Deep Ethics Integration in Technology Research and Development: Bridging the Principles-to-Practice Gap in AI Ethics

Abstract In this paper I investigate how integrating ethics deeply into technology research and development can help developers overcome a major obstacle in the ethical alignment of innovation processes, namely, the “principles-to-practice gap”. By reference to a specific ethics intervention made by Berlin Ethics Lab featuring an AI system under development, I show that the gap is not about a failure to operationalize principles for research and development practice, but rather that it originates in divergent styles of thinking: ethical reflection and design thinking. Based on these insights, I elucidate how ethical reflection can be integrated on a foundational level into research and development processes. An essential starting point is the creation of a space for reflection to explore potential designs, a space dedicated to contextualizing and re-conceptualizing technology. This enables an ethical reframing of the design problem, systematic ethical analysis, and an assessment of potential implications. I conclude by presenting the ways in which this approach of deep ethics integration challenges other approaches to integrating ethics at an early stage (such as embedded ethics, STIR, VCIO, ethics by design, and value sensitive design) and demonstrating what additional benefits it has to offer. Similar content being viewed by others 1 Introduction What is discriminating? The Bluetooth interface! This insight proved to be a turning point in one of the Berlin Ethics Lab's interventions. A company had asked for a workshop to conduct an ethics test on its most recent product under development. The aim was to design an app for mobile phones based on an AI-driven identification of walking patterns and supported by data exchange using a Bluetooth interface. Being highly specific to individuals, walking patterns make it possible to personalize many tasks, including the automated unlocking of doors, as envisaged in this case. During the workshop ethical issues were brought to light, as expected. The developers had expected privacy issues to come up, so to address this, precautionary action had been taken to ensure data would be stored locally on the mobile phone. The ethicists had expected to encounter problems around implicit bias, an issue that became apparent when the tool for reflection used in the workshop enabled participants to explicate their assumptions about the future user. When these individual ideas were shared, a surprisingly coherent image of a male, white, young and energetic walker emerged, effectively neglecting the experiences of people with other walking patterns due to injuries, say, or different footwear (consider high heels in contrast to sports shoes), or with no walking patterns at all (as with wheelchair users). However, ethical issues also cropped up unexpectedly, for both the developers and the ethicists, at a point which came as a surprise for everyone. While contextualizing the future artefact methodologically and exploring the conditions underlying its potential existence as well as the implications it entails, the Bluetooth interface became a focal point of concern. From the developers’ perspective, implementing the most recent Bluetooth standard (which guarantees greater security and better addresses health concerns) was a reasonable way to go when developing a new product that was to be designed in the best way possible. However, this logical step within the chosen design framework would have had a discriminatory effect – namely, that of excluding users who are either unable or unwilling to buy the latest mobile phone on the market. How is it that a tiny technical detail such as the choice of a Bluetooth version – which at first glance seems like the epitome of technical neutrality – can raise major ethical concerns? Why is it that the workshop participants (tech experts) were not able individually to notice the discrimination but could only do so when collaborating with the workshop facilitators (ethics experts), assisted by a reflection tool? In this paper I seek to show why the act of integrating these perspectives in a shared space for reflection was able to close the so-called principles-to-practice gap. It is often in the seemingly unimportant technical details where abstract values become concrete material operations and affect the handling of an artefact. It was the developers’ resolute intention not to discriminate. Yet the reflective move that appeared logical given a specific design logic – namely, ensuring that the components of the new app reflect the latest standards – had to be interrupted. Considering the design in its intended context of use enabled the link between the Bluetooth interface and the issue of social participation to become apparent; only now could the ethical consequences of design decisions be systematically interrogated and alternatives developed. This “re-framing” was made possible by the company’s developers and ethicists from the Berlin Ethics Lab coming together in a three-hour workshop. The workshop enabled the former to step outside their everyday routines and to analyse the newly developed product from an ethical point of view in a sheltered setting, supported by reflection tools that provided guidance in collaboratively exploring issues and drawing conclusions. The principles-to-practice gap is currently one of the biggest problems in AI ethics. Countless guidelines have been approved to ensure that the development of artificial intelligence is based on ethical principles (e.g. Corrêa et al., 2023; Fjeld et al., 2020; Hagendorff, 2020; Jobin et al., 2019), including perhaps one of the best known, the European Ethics guidelines for trustworthy AI with its corresponding assessment list (European Commission AI HLEG, 2019, 2020). What such guidelines have in common is that they identify a set of values (also referred to as principles) intended to provide orientation for developers of AI. Unsurprisingly, given that these discursive processes are influenced by but conducted independently of one another, there are overlaps as well as divergences. The major flaw in these guidelines is that they stop precisely where the actual challenge begins, namely, in finding answers to how abstract values can be translated into practice for a given technical application. In AI ethics this challenge is currently referred to by the catchphrase “principles-to-practice gap” (e.g. Bleher & Braun, 2023; Corrêa et al., 2024; Floridi, 2019; Morley et al., 2020, 2021, 2023; Taddeo et al., 2024; Zhou & Chen, 2023); the issue at stake is the operationalization and specification of AI ethics. The challenge remains even when – as happened in August 2024 with the European AI Act, the world’s first comprehensive regulatory instrument for AI – guidelines are formulated as laws: the step that still needs to be taken is that of moving from the abstract to the concrete in specific instances, regardless of whether guidelines or laws are to be formulated. However, the extent of the so-called principles-to-practice gap is larger still. AI guidelines may give the impression that there is an established canon of values; there is none. The question as to which values are most important for a given product or a given area of application must be determined for each individual case because the impacts of artefacts are highly context-specific (e.g. Gogoll & Zuber, 2026). In value sensitive design, determining the relevant values for a specific product to be developed (also known as “value elicitation” or “value discovery”) serves as a methodological starting point (Borning and Muller 2012; Friedman et al. 2013; Friedman and Hendry 2019; Friedman and Kahn 2002; van de Poel 2020b, p. 302). Where this case-based advancement is not considered, we see inadequate ethical analyses and assessments which fail to recognize that AI algorithms are not isolated entities but rather part of far-reaching technical systems. It is not the AI algorithm on its own but rather its embeddedness in the broader technical ensemble along with all the latter’s components and the resulting interactions that needs to be considered in the contexts in question. Thus, we are dealing with an explosion of complexity such that the purported solutions peddled by generalized approaches stop short of delivering on their promises, based as they are on an established canon of values and on predefined operationalizations relating to fixed criteria and indicators (e.g. Hallensleben et al., 2020; VDE Verband der Elektrotechnik & Elektronik Informationstechnik e.V., 2022). Before using the example outlined at the start of this paper to elaborate on why deep ethics integrationFootnote 1 can provide a response to this challenge, I want to offer two preliminary comments. The first has to do with terminology. Depending on one’s affinity to a given school of theory – the ethics of biomedical principles (whose influence is predominant in AI ethics) or value sensitive design (an influential current in the ethics of technology) – the main subjects under discussion are principles or values (though fairly often both terms are lumped together) (e.g. Corrêa et al., 2024; Hallensleben et al., 2020). The ideas presented here about integrating ethics throughout a design process remain valid regardless of whether we wish to speak of principles or of values; the reader is welcome to substitute the concept of principles with that of values as they wish.Footnote 2 The second comment relates to the theoretical framing of this paper. I do not share the prevalent interpretation that the principles-to-practice gap is about the gap between theory and practice (Bleher & Braun, 2023). The juxtaposition of theory and practice is an outmoded dichotomy, and has been such – not just from the perspective of philosophy of science – since the practice turn. If, alongside designing and developing, ethical reflection also is to be understood as an epistemic practice (Ammon, 2017c, 2019), then it is not just a question of transferring theory into practice in order to bridge the gap. Rather, as I argue here, the challenge lies in integrating two epistemic practices in such a way that it becomes possible to deal with ethical issues from a design perspective. As I will show, it is precisely in this theoretical re-framing of the so-called principles-to-practice gap that the key to overcoming the problem lies; additionally, it becomes clear why approaches based on abstract operationalization – though they may provide valuable building blocks – nonetheless fail to build the actual bridge. In what follows, my focus is on technical research and development projects and is guided by the question of how ethical issues can be integrated as early as possible into technoscientific thinking and reasoning in order to trigger changes in design and to overcome the so-called Collingridge-Dilemma (Collingridge, 1980; Grunwald, 2019, pp. 189–194). I will clarify, first, what characterizes the different disciplinary styles of thinking that are typical of the respective epistemic practices (Sect. 2). What I seek to show is that, to be successfully integrated, these practices with their respective epistemic strategies need to be combined into reflective movesFootnote 3. It then becomes possible to investigate the design according to ethical considerations and to embed these directly into processes of generating knowledge about the future product. As I seek to show, this process of deep ethics integration rests on multiple preconditions in several respects. For one thing, it involves reconceptualizing the concept of technology by starting with contextualization in order to understand artefactsFootnote 4 from the point of view of a processual and praxeological approach (Sect. 3). Extending the space of reflection in this way enables the different epistemic practices to be brought together at those points where it is necessary for ethical reflection to occur. In Sect. 4, I look at the features that are typical of this integrative praxis and point out which of their elements are significant. Finally, I explore the challenges posed by deep ethics integration in comparison to those of other approaches (such as embedded ethics, STIR, VCIO, ethics by design and Value Sensitive Design) in order to show what it has to offer that is new (Sect. 5). Looking ahead in Sect. 6, I explain finally how this form of ethical empowering may provide an additional piece of the puzzle when it comes to designing and producing technology in a responsible way. 2 The Art of Reflection: Reframing the Principles-to-Practice Gap Acknowledging that technology (henceforth: Technik)Footnote 5 is conceptualized in different ways is an initial, crucial step towards integrating ethical considerations into processes of technical research and development. The workshop mentioned at the start of this paper that brought together developers and ethicists involved an encounter between two styles of thinking that could hardly be more different given the rigours of disciplinary training. In using the term “style of thinking” I want to emphasize that component of epistemic practices which refers to a specific way of comprehending the world and is associated with certain characteristic reflective moves and epistemic attitudes.Footnote 6 Clearly the epistemic praxis of designing (Ammon, 2017c, 2019) differs markedly from that of ethical reflection if we look at their respective tools of reflection or the ways their respective outcomes manifest; this is less obvious with their associated styles of thinking. The subtle interplay between the material conditions underlying the milieu of reflection and the style of thinking used sets the stage for addressing certain problems, which then forms the object of thinking. The specificity of the style of thinking sets an epistemic focus and gives rise to a characteristic set of questions to be worked on. In the following, as I detail the specific features of the styles of thinking called upon in developing an artefact or reflecting on an ethical issue, the limitations of such an approach become obvious. There is no one single style of thinking. To answer the questions raised, I must necessarily resort to overstatements and simplifications to make the disciplinary styles of thinking recognizable in terms of their peculiarities, their inherent rigour, and their conceptual logic. In my understanding, the individual manifestations of styles of thinking include subjective nuances based on experiences, preferences, and personality traits. Nonetheless, a contrastive approach based on abstraction makes it possible to illustrate the differences in the way each one comprehends its object of thinking and the specificities of how each generates knowledge – this is all that matters to me here. My intention is not to establish a valid, watertight description of the two styles of thinking; instead, what I seek to facilitate is a basic understanding of the gap that exists between the two modes of comprehension, and thus to render visible the integrative task required of deep ethics integration. Naturally, my own perspective – that of a bridge builder at home in both worlds and thus familiar with both styles of thinking – feeds into this discussion. The style of thinking used in designing something is tailored to the challenge of creating a technical object (cf. Table 1). Anyone who adopts the role of designer will often develop a close emotional connection to their own creative output. In many cases this special closeness is borne along by the enthusiasm felt for the design idea which is advanced and defended during the development process. Here, the strengths of the development are put up front – not least to convince other people. In this respect, the style of thinking is heavily influenced by a desire to figure out how something can work: the job of enabling becomes the primary focus. In the development process, exploratory work is geared epistemically toward an outcome that is good enough – or “satisficing”, a neologism coined by design theorist and economist Herbert Simon by merging the terms “satisfy” and “suffice”, (Simon, 1996, p. 27 f). This often means, in quite pragmatic terms, making a technical system work. Efficiency – that is, achieving an outcome rapidly within a limited budget – matters. The important thing is to put the ideas into practice and test them, because many things will only come to light during implementation. This major dependence on context along with dynamic ever-changing interactions between the technical system, society and the environment means that subsequent improvements must be made in an iterative process to reveal errors and potential weaknesses. Yet whether an emerging artefact is good enough, and which criteria are relevant to decide this, depends not only on the technology itself but also on the context of application and the cultural setting. The factors identified as relevant need to be examined repeatedly in a critical light and negotiated, especially in the case of emerging technologies but also in the case of existing technologies. Yet these broader circumstances are seldom considered during the development process. One of the biggest challenges in making the development task manageable is to reduce its spiralling complexity. To do this, it is necessary to identify the key leverage points of adjustment in the process so that these then form the focus in the exploratory design moves and the testing phases. This occurs based on disciplinary priorities that implicitly determine which aspects are to be considered and which simplifications introduced. One result of this is to limit the focus to issues addressed as “technical,” including mode of operation, construction principles, production procedures and choice of materials. This tendency is reinforced by the reflective move that involves in to the level of technical detail – a move of considerable significance, because it is here that crucial choices are made that effect the entire product. Seemingly minor details can often have a major impact and may be the make or break of a given solution; focusing too closely on the solution, however, entails disregarding contextual factors. Values have an important role to play in deciding between different variants and alternative developments. They provide orientation in the design process, enabling developers to decide in which direction the design should go, what limits and priorities should be set, and how to weight the various criteria. “Technical” and economic values are foremost here, albeit consideration of sustainability in its narrow ecological sense is increasingly gaining in importance. Safety and health have already long been included in quality assurance due to painful lessons learned from the adverse impacts of technologies. When it comes to considering ethical issues, the seemingly obvious approach would be simply to add more ethical values to the ones already being considered during the development phase. This would be a false move, however. While it is important to extend the range of values for consideration, it is not nearly enough. Deeply embedded in styles of thinking are fundamental differences in the way different actors conceive of technical objects; these differences provide a sense of why integrating ethics into technical development processes is such a great challenge. Before I turn, by way of a contrast, to a detailed discussion of the specificities of the conceptual moves involved in ethical reflection, a preliminary comment is in order. The term ethical reflection here does not refer to the broad range of ethics in its entirety, but rather to the specific technoethical perspective on the world. In the ethics of technology, the general ethical question of how we human beings can co-exist beneficially is regarded through the lens of how this beneficial co-existence might look in the context of human-Technik relations and their mutual effects. The distinctive feature of technoethics is the adoption of a critical attitude toward technology more generally and the concrete emerging artefact. The aim is to highlight potential ethical shortcomings; to do that, it is necessary to generate a certain distance between the ethics analyst and the object under consideration (see Table 1). The historical distancing frequently used in the humanities cannot be applied with newly emerging technologies. It may certainly be possible to draw parallels between new and older technologies and to derive certain insights by forming analogies. However, such procedures have their limits when dynamic societies shaped by changing values encounter technical applications that have little in common with existing familiar ones. To maintain a critical distance, it is important to distinguish the promises proclaimed by narratives of progress and hype from the actual technical development. Systematic scepticism is required to identify ethical shortcomings. It takes a critical gaze to question what appears obvious and not to be overhasty when accepting answers. Assumptions must be brought to light and analysed, implicit decisions rendered explicit, impacts explored and assessed. The probing must be relentless to identify ultimate grounds upon which to develop a line of argumentation based on truth or rightnessFootnote 7 in reasoning. To this end, the ethical study must be comprehensive so as not to overlook any aspects. This demands an extremely thorough approach, which is incompatible with external time pressure. By its very nature ethical reflection is not about the interests or needs of any specific person; the answers it generates aim to achieve supra-individual validity. As such, many reflective moves in ethical analysis shift from the concrete to the abstract, enabling answers to be found which can then be applied back to specific cases. Questions such as what constitutes the good life in the context of human-Technik relations are central to this, as are the good practices that emerge when interactions are guided by certain values. Crucial to making these issues manageable is the act of broadening the field of view far beyond the technical object itself: its effects on relations between people, on the environment, and on relations between humans and the environment are all considered. Thus, “zooming out” is a characteristic move that seeks to render larger contexts apparent. As much complexity as possible must be allowed in this process to enable an understanding of the artefact’s impacts. Despite the simplifications involved in this comparison of the two styles of thinking, it becomes abundantly clear where the actual challenge lies when it comes to integrating ethical considerations into the development process. It is not a matter of a gap between theory and practice – or, more pointedly, a gap between principles and practice. Ethical reflection and design constitute very different kinds of epistemic practices. The above analysis has shown why it is not sufficient merely to add ethical values to so-called technical values and to operationalize them in a top-down approach. The gap is much more fundamental. The problem begins with the very fact that there is no shared object of thinking and that, due to the differences between the styles of thinking, a range of problems arises that have barely any areas of overlap. The contrast reveals a collision between different ways of conceiving Technik – even as, in their very contrariness, they complement each other. It is this area of tension that deep ethics integration can use to beneficial effect. To embark on a joint process of generating knowledge, the different styles of thinking – each of which is based on an inherent logic and makes sense in terms of its own epistemic approach – need to be brought into a shared space of reflection to feed into convergent reflective moves. Thus, dealing with ethical questions, assessing possible consequences, and highlighting societally desirable development alternatives becomes the challenge to integrate ethical exploration into design thinking. The necessary first step, therefore, is to find on a shared epistemic focus, thereby bringing conceptual issues to the fore. 3 Foundations: Conceptualizing Artefacts To understand the differences in the object of thinking, let us return to our initial example and thus to the question of why a seemingly inconspicuous Bluetooth interface in an AI application could be suspected of disadvantaging certain user groups, thereby prompting ethical concerns. On the website of a major electronics dealer, I find the relevant product on offer. The product description tells me that BL652-SA-01-T/R is based on Bluetooth standard v5.0 – a standard which, at the time of the workshop, had not yet been integrated across the board into mobile phones.Footnote 8 The product photos show a circuit board with blue varnish and white print, on which rests a large shiny silver metal casing alongside several smaller electronic components (Fig. 1). The list of features on the dealer’s website tells me that BL652-SA-01-T/R has an integrated antenna and several serial interfaces, that it operates with a data transfer rate of 1 MBits/s, a maximum transmission power of 4 dBm and a receiver sensitivity of -96 dBm. I discover details about the product’s data memory, its power consumption, its operating temperature as well as the component’s dimensions: at 14 mm long, 10 mm wide and 2,1 mm high and weighing under 1 gramme, it can readily be inserted into a mobile phone. The accompanying data sheet contains further detailed information about hardware specifications, a functional description, and mechanical details. I can find no cause for ethical concern here, though what I do find is the information that the product has been tested for compliance with relevant standards for the EU market in accordance with the requirements of CE marking. Nothing in the description of technical parameters suggests that BL652-SA-01-T/R might be discriminatory. The theory of technology neutrality appears convincingly watertight in the face of this specific component.Footnote 9 But is the issue really that straightforward? I conduct more detailed research. Bluetooth, I discover from an article in Wikipedia, is an industrial standard established in the 1990s to enable wireless data transfer at short distances.Footnote 10 It is part of the field of radio technology in which electromagnetic waves in the radio frequency range are modulated in such a way that they transmit signals.Footnote 11 The Bluetooth standard makes it possible to connect different electronic devices with one another and to exchange data. This gives rise to small local networks for data exchange through to pairing and bonding, in which two devices (in our example a mobile phone and an electronic door lock) connect with each other, enabling access to rooms to be controlled remotely. What is striking about the product presentation is it implies a shared understanding of what counts as belonging to a technical description. “Security” and “health” issues in connection with radio technology are clearly part of the “non-technical” information offered on the website. This is where the technology’s shortcomings are in evidence. The wireless transfer of data between two devices which first must find each other makes Bluetooth especially vulnerable to issues of security: scintillating expressions such as “eavesdropping”, “man-in-the-middle” and “denial-of-service attacks” refer to opportunities for intercepting and manipulating information, issues the more recent specifications attempt to address. In addition, though, the radio technology requires special care when it comes to health aspects. The Bluetooth standard is situated between of 2.402 GHz and 2.480 GHz on the electromagnetic spectrum and is thus non-ionizing. Since it is not possible to rule out harmful impacts from this part of the spectrum, it has been classified by the International Agency for Research on CancerFootnote 12 as potentially carcinogenic.Footnote 13 Here too more recent standards promise improvements based on lower power consumption. My research yields no information regarding issues of discrimination, however. Indeed, how can it? What had already become obvious in the description of the component is represented precisely the same in the description of the technology.Footnote 14 Neither in the presentation of the technology’s abstract operating principles nor in the portrayal of the component’s material conditions does the issue of social interaction make an appearance. An implicit “self-immunization” thus occurs: technology is everything that can be considered “technical” information, along with everything directly connected to it. By virtue of this conceptualization, all other issues lie beyond the realm of the technical and are “external” to technology. Common to both conceptualizations is a conspicuous generalization which effectively de-contextualizes Technik and ensures that the discriminatory potential of the Bluetooth interface does not even enter the picture. Neither considering Bluetooth as an abstract technology nor viewing it as a component (where the technology, although now materialized, becomes an isolated element removed from its operational context by the way it is conceptualized) allows us to draw any inferences about why the Bluetooth interface should even be capable of discrimination. As its very conceptualization suggests, a component constitutes a basic element that can be inserted into countless technical contexts. Naturally, then, the conceptualization foregrounds precisely those characteristics that extend beyond any specific mode of use. Having said this, certain ethical values certainly can and do enter the picture. As we have seen, the field of radio technology raises questions regarding health impacts as well as privacy issues which are directly linked to the technology’s mode of functioning. However, the abstract perspective is geared toward general functional characteristics, meaning that specific instances of actual use do not come into consideration. Conceptualizing the Bluetooth interface as a concrete component does not exclude consideration of ethical values either. Through the materialization and embodiment of the technology, aspects associated with manufacture and materiality come into focus, prompting questions about the raw materials used and where they came from, the conditions in which the component was manufactured and produced, as well as its durability and whether it can be repaired or recycled. This touches on the value of sustainability, which comes into view with the instantiation of the technology. These examples show how each conceptualization brings with it a certain range of considerations, opening a space for reflection which, with its own epistemic focus, governs the object of thinking and therefore the inclusion or exclusion of certain issues. They also show that the concepts of “technology” and “component” do not allow for the investigation of issues relating to discrimination. Discrimination in our case only comes to light when we adopt a broader focus by exploring the actual artefact – not, however, from an “artefact-only” perspective (often encountered in design reasoning), but as a technical object fully embedded in a complex socio-technical-environmental system (Rohde et al., 2026). The revised conceptualization of Technik I want to draw on is rooted in a praxeological, situated and processual approach. In this way, the artefact’s impact on behaviour and on relationships between people, society and the environment comes into view, thus laying the groundwork for bringing exploratory design and ethical analysis into a shared space for reflection.Footnote 15 Without going into too much detail, let me briefly introduce the praxeology of artefacts. As an ensemble, artefacts only come to be constituted as such through the practices associated with them. Accordingly, it is not the status of being a thing that is the crucial point of departure (as language would suggest) but rather the praxis in which actions and thing are woven into a whole. It is the opening and closing of the door, rather than the key, that must be foregrounded. It is in actual praxis that different elements are integrated into an ensemble. In our example, these include an app, a mobile phone, the phone’s Bluetooth interface, a door, and a door lock with a Bluetooth interface and home automation technology; these are encountered by a regular user as well as indirect stakeholders, and therefore need to be considered in terms of how they work as a whole. To put it more precisely, we are talking not about a single practice but about a group of practices: for example, the actions of opening and closing in the narrower sense also include practices associated with preparedness and maintenance. Neither should we forget the unintentional practices that can occur and which can lead to creative forms of appropriation as well as to practices we might categorize as “misuse” or undesirable use. Artefactual practices are always situated. This necessitates acknowledging and looking closely at the complex real-world conditions in which they occur; further, each act of situating influences the distinctive form these practices take. The aspects of situatedness that are especially important in the context of ethical analysis have to do with the setting of the surrounding environment, including the cultural and temporal context as well as political and legislative conditions. Another important dimension in this regard is change – not only change in the artefact itself but also change in the world in which it is embedded. In other words, the practices are not just anchored in dynamic contexts; these contexts are themselves dynamic. Furthermore, artefacts need to be understood from the point of view of their processuality. This is not only because of the inherent processuality of the practices themselves but also because of the artefact’s mode of existence. We can divide the mode of existence of an artefact roughly into the four primary phases of design, manufacture, use and dissolution. A technoethical analysis that starts at an early stage must conceptually anticipate all four phases from the start, which inevitably entails the multiplying of the relevant artefactual practices. It begins in the design phase: although the future artefact is not even embodied materially, there are nonetheless manifestations of the design in the form of descriptions, drawings, models, and prototypes that help in developing and further refining the future artefact (Ammon, 2017a, 2017b).Footnote 16 The manufacturing process translates the design into its material dimensions; this involves a fine-meshed process of feedback between design and manufacture, characterized by iterative adjustments. In such cases, improvements may be done even when the product is well into the use phase to enable flaws in planning and manufacture to be corrected. The ageing artefact is subject to phases of maintenance and repair. In addition, changes in use may increasingly occur, especially with artefacts that have a long service life. Dissolution denotes the phase in which an artefact ceases to be perceived as a well-defined object and its components are given over to other practices (from spare parts storage to waste incineration). There is a close link between the specificities of artefactual practices and the way artefacts are designed. The way an artefact is designed suggests certain actions in relation to it, whereas others are made more difficult; this generates behaviour patterns.Footnote 17 Artefacts engender normative effects when the way they are designed suggests or hinders specific actions and behaviours. The normative power generated by these affordances of the design elements is influenced by the scale on which they occur – that is, depending on whether an individual or many people change their behaviour frequently or less frequently. Large numbers of behavioural changes may lead to technologies and the artefacts associated with them having a transformative impact and triggering large-scale change. Thus, the formativity of artefacts comes into view when we look not just at human behaviour but also at environmental conditions in a broader sense.Footnote 18 They require infrastructures and many other prior conditions in order to function, even as they establish a specific milieu with their emissions and other consequences of their operation, through which they “educate” their environment. Conceptualizing Technik as an artefact enables us to broaden our view and thus to render potential discrimination perceptible in the exploratory design phase. A new set of questions (to which we are oblivious when talking about “component”, “technology”, or “artefact-only”) then arises: What happens if I want the newest Bluetooth specification to be used in the ensemble for practices of opening and closing a door automatically? If the assumption is that the newest generation of mobile phones is required to open doors, would this entail disadvantaging low-income groups? Do the impacts of discrimination become greater, perhaps, in interaction with the AI-driven app when, in the case of people with impaired mobility, less reliable results are available due to such cases having not been adequately represented in the training material? In order to clarify whether a form of discrimination may exist, the situatedness of the artefact needs to be taken into account. Does the locking mechanism control access to a public space or to private rooms? Is it to be installed on a company’s premises where only a few employees are given access to a specially secured area, or does it affect the entire workforce, who are provided with company phones? Are there certain cultural particularities or established legal regulations that need to be considered? These exploratory questions give an initial impression of how varied the answers may turn out to be in relation to instances of potential discrimination when more contextual factors are taken into account. Depending on what kind of practices occur in which settings, the conclusions drawn may lead to different results. Within the adjusted conceptual framework, it at last becomes possible to take a differentiated look at concrete circumstances and examples of particular situations of use, and to examine the effects of the technical specifications with regard to, among other things, issues of discrimination. 4 The Praxis of Deep Ethics Integration: Bridging the Gap Between Styles of Thinking As we saw in the previous section, the re-conceptualisation of Technik as an artefact lays the ground for deep ethics integration. The shared epistemic focus on artefactual practices opens up a space for reflection in which different styles of thinking encounter one another so that ethical questions can be asked and corresponding technical answers developed. Successful integration of this kind generates knowledge about the potential impact of the artefact on human beings in general, on groups, society, and the environment; it provides guidance for a societally and ethically desirable development and shows leverage points for adjustment to steer the design accordingly. But how can the gap between the thinking styles of design reasoning and ethical reflection be closed, given that they differ in terms of guiding values, creative relationship and emotional connection, inherent reasoning and epistemic values (cf. Table 1)? This question leads us to the issue of strategies for integrating ethical reflection into research and development processes, and thus to the practice of such integration, the most significant challenges of which I would like to outline briefly. 4.1 Opening up Spaces for Reflection The integration of ethical considerations into technological research and development processes requires a critical stance, the ability to distance oneself emotionally and to allow for scepticism as a methodological attitude. Initially, this means giving up the maker perspective, and, quite pragmatically, finding a time and a safe space to explore potential weaknesses and blind spots of the envisioned design. In our example this was a three-hour facilitated workshop in which different tools for reflection were used. The workshop made it possible for the developers to step outside the daily development routine for a limited time in order to jointly explore issues concerning the ethics of technology and impact assessment. This enabled a more distanced perspective to be gained on the planned AI application and to examine it more closely and critically in structured reflective moves. 4.1.1 Cultures of Collaboration To achieve this critical distance in exploratory ethical studies, it is crucial to be able to address weaknesses openly. This requires giving every voice a hearing, regardless of hierarchies, roles, gender, or disciplinary and cultural backgrounds. Robust generation of knowledge can happen when different perspectives come together in the process to broaden the knowledge and values base. In this respect, from an epistemic point of view, good communication takes on a key role which must be actively established in integrated ethics interventions. This is even more important when one considers that temporary ethical interventions always operate within the limits set by the existing organizational culture, the composition of the participants, and their styles of behaviourFootnote 19. Values are central when it comes to establishing this culture of communication in processes of ethics integration: values in how we deal with one another, values for the discourse, and values for the processes of interdisciplinary knowledge generation. Thus, collaborative knowledge generation is always also about good practice; the “how” of generating knowledge has a powerful influence on the outcome – which brings us to the ethics of integration. Allowing different styles of thinking to enter a shared space for reflection and thus into shared knowledge generation requires a specific way of designing the process and a special culture of collaboration. Accepting divergent epistemic perspectives, being highly flexible in the ability to adopt new points of view, trusting others’ expertise, feeling safe enough to speak freely of potential problems, and being able to point to difficulties and errors, appreciation and mutual respect are all important parts of this. Given this, the epistemology of integration is inherently linked to the ethics of integration: they both go hand in hand in the art of reflecting. 4.1.2 Reflexivity Processes of integrated ethics also require an awareness of the limits of one’s own knowledge as well as an awareness of the strengths and weaknesses of the style of thinking one has learned in the course of acquiring specialist expertise. If a disciplinary lens is not recognized as being such, and is instead regarded as a neutral view of the world, it is hardly possible to remain open to other points of view or to adopt different perspectives. In this regard, ethics integration should also prompt an awareness of one’s own internalized style of thinking. This reflexivity needs also to include the situatedness of one’s style of thinking in personal, biographically influenced insights and cultural backgrounds, all of which are linked to emotions and moods as well as personal values and attitudes. The development process involves making value-oriented decisions and drawing on the imagination, which is inevitably shaped by individual experiences. This being the case, it is crucial that, as prefigurations, these shaping influences are not obscured but are uncovered and worked with in a constructive way. Ethicists and product developers are always also people with lifeworld experiences and values, and with ideas about what constitutes a good life (not always the same as one’s own personal interests or preferences), all of which co-shape the development process.Footnote 20 4.2 Reasoning and Valuation Re-framing the design problem (Schön, 1983, p. 85) from an ethical standpoint allows us to bring the different logics of reasoning and reflective moves into dynamic interaction. Once a re-conceptualization of Technik has established a shared epistemic focus, we can see how ethical concerns become a part of the exploratory design process, as design moves explore the design space step by step (Schön, 1983, p. 79) and generate knowledge about the future artefact (Ammon, 2017c, 2019). Systematic probing of the design configurations makes it possible to settle on reliable, well-grounded and ethically tested versions of the design. Ethically undesirable consequences as well as insights into ways of preventing these by trying out potential variants can be scrutinized. What impacts would the choice of an older Bluetooth standard have on use practices? Are there alternatives to a Bluetooth connection? Can undesired effects be offset by a change in the algorithmic architecture? In this way technical specifications and details can be explored in terms of their impacts, and the set-up of the ensemble re-examined. We encounter an iterative interplay between “zooming in” at the level of adjusting specific technical parameters in the narrower sense and “zooming-out” at the level of the artefact’s impacts in its contextualization. The exploratory design moves enable us to study various options for designing the artefact in terms of its potential impacts. Effecting repeated, targeted changes to the parameters and subjecting their potential impacts to critical scrutiny in relation to the contextualized artefact makes it possible to identify dependencies and interactions that might give rise to ethical problems. It thus becomes possible to identify leverage points of adjustment in the design that are linked to undesirable consequences. Ethical considerations and exploratory design moves enter a shared space for reflection at this point, where knowledge about potential consequences can be generated and undesirable effects avoided by the choice of alternative technical options.Footnote 21 4.2.1 Mind Your Language! With the problem re-framed, language becomes a focal point when talking about the technology and the future artefact. In the process of re-framing, words, metaphors, and images become conceptual vehicles that guide our imagination. Naming the software for walking pattern-based authentication on a mobile phone “artificial intelligence,” “app” or “key” exerts an influence on our imagination. The linguistic framing shapes our expectations of the artefact and directs our attention to putative strengths and weaknesses, which in turn influences the emerging design. Every description, every metaphor, every image is suggestive of certain ideas even as it blocks others out. This is all the more important in the conceptually driven design phase, because here there is no direct experience of the future artefact. In this respect it is crucial to find appropriate ways of speaking and imagining the artefact to get as clear a view as possible of the interactions and practices entailed by it. Because this is possible to only a limited extent, it is important to generate awareness of these prefigurations in processes of integrated ethics, and to observe the effects of terminological framing with a critical gaze.Footnote 22 4.2.2 Ethical Triangulation and Good Praxis The results of this process of exploring and testing can be evaluated for the purpose of revising design decisions as well as coming up with more general recommendations and design guidelines. Ethical triangulation can help to provide an initial assessment in the valuation process where “technical” values such as efficiency, effectiveness and economic feasibility need to be brought into alignment with a broad set of ethical values and ideas of the good life. Ethical triangulation involves comparing existing practices, ideal practices and the hypothetical practices associated with the envisaged new development. Comparing existing practices with good opening and closing praxis makes it possible to locate the new development on this spectrum in order to establish how the practices would either improve or get worse. This kind of feedback provides orientation and makes it possible to carve out alternative development trajectories at an early stage if undesirable impacts become apparent. This triangulation means that ethical exploration and testing and the process of developing recommendations are tied directly into the design moves. The critical attitude adopted in doing so serves as a heuristic which, systematically underpinned by the acts of “zooming out” and contextualizing, enables the ethical investigation to be conducted.Footnote 23 Not only must those involved check whether the future product meets ethical demands; they must also consider whether the knowledge gained regarding possible impacts is good enough to formulate recommendations. Due to the processuality of artefacts, this analysis needs to be conducted for all phases of the artefact’s lifecycle. 4.2.3 The Role of Ethics Tools Ethics tools play a crucial role in bringing together different styles of thinking.Footnote 24 As epistemic tools, they become a stimulus for ethical exploration; they prefigure reflective moves. They may lay the foundation for re-conceptualizing Technik and initiate discussion about specific ethical issues in addition to guiding the process of testing and evaluating the object in question. A tool for reflection such as the “implication fan,” which was deployed in our above-mentioned intervention and enables the systematic exploration of preconditions and consequences of a future artefact, constitutes a repository of knowledge of ethically relevant issues and epistemic strategies that come into effect when handling the tool. These explorations, facilitated by the the tool and steered by its implicit framing, enable the connection between ethical issues and technical leverage points to be established, thus rendering visible the normative assumptions present in design decisions. Making these issues explicit by means of visualization makes it possible, from a design perspective, to question decisions and develop technical alternatives.Footnote 25 Since the way a tool is both designed and deployed has a direct influence on reflective moves during its use, a critique of the tool must reveal conditions for reflection and conceptual affordances in order to render the effects on ethical analysis and ensuing judgements transparent and potential limitations of knowledge generation recognizable. 4.3 Process Modulation Technical research and development processes are powerfully shaped by an iterative approach in which the future artefact becomes more and more concrete. A successful integration of ethical considerations must be adapted to this process of concretization and monitor it. In this way, integrated ethics reveals itself to be inherently processual: if the process is designed accordingly, ethics becomes a co-designer of the emerging artefact. Goodness, not truth, as a processual epistemic value provides orientation: goodness means deciding when the results are good enough (or satisficing, as Herbert Simon named it in his design theory) to proceed. In our example, the ethical input ought not to be limited to a brief intervention during the early stage of product development; rather, repeated ethical explorations should be conducted throughout the entire development process. The ethicist’s role in this process turns out to be twofold. To enable an exploration of the company’s development approach, as requested, it was necessary to bring in the specialist expertise of people trained in AI ethics and, more broadly, in technology ethics and technology assessment methods. At the same time, the ethicists were responsible for modulating the process of ethics integration. While the scheduling of the workshop and the selection of participants were largely determined by the company’s stated needs, the manner of intervention had to be planned in such a way that the ethical reflections could be embedded usefully into the current stage of the development process. Especially in the case of longer-term support for integrating ethical issues, process modulation can be expected to constitute a key task of the ethicists involved, who – besides planning the overall ethical co-design process – are also responsible for devising the specific interventions. The workshop structure with its specific sequence of stages of reflection can then provide a systematic heuristic for guiding reflective moves. According to the stages of the process, it takes a suitable degree of immersion and tailored forms of intervention to raise the ethical questions appropriate to a given stage of development. Since the artefact is most amenable to shaping in the research and early development phases, ethical issues in these phases need to be pursued very carefully, despite the vagueness and non-knowledge involved in these phases. However, contrary to what widespread development models suggest (such as the V-model or technology readiness levels, TRL), ethical reflection should not end when the artefact is launched on the market (Fiedler et al., 2025). Ethical involvement must go beyond conceptual anticipation in the development phase and critically monitor also the manufacture, use and dissolution of the artefact, so that relevant knowledge can be fed systematically into iterations of the artefact’s development. What applies to all developments is even more crucial when it comes to applications involving new technologies – at a point when there is still little experience of how they are actually used. In this respect, artefacts and knowledge about their impacts should only ever be considered provisional, and there should be a willingness to make new adjustments if negative consequences for society and the environment become apparent.Footnote 26 5 Discussion: Deep Ethics Integration in Research and Exploratory Design – What Makes the Difference? As argued above, a key element of integration consists in embedding the praxis of ethical reflection into design thinking. Ethical exploration needs to take hold in the thinking of precisely those who are involved in advancing research and development, because many undesirable consequences of technology have their roots in specialist technical issues that can only be addressed with the relevant specialist expertise. This requirement goes far beyond what Responsible Research and Innovation (RRI) addresses and practises as ethics, in both its institutional and academic versions (Burget et al., 2017). In advancing RRI the European Commission has focused on research ethics (Iphofen, 2020; Koepsell, 2017), with emphases on research integrity, protection of the object of research, and ethical acceptability (Berghaeuser et al., 2024; European Commission, 2015). The role of ethics commissions in particular is double-edged. Outsourcing ethical questions to an expert body prior to implementation might have a boomerang effect if efforts aimed at integrating ethics stop there. A division of labour might tempt tech people to regard ethics as not belonging to their “actual” job of researching and developing; at the same time, the different styles of thinking do not enter a shared space for reflection.Footnote 27 A one-off review at the start of a technological research or development programme suggests that, once it has been authorized, all ethical problems will have been resolved. Such an ethical approach to research is useful when it comes to critically illuminating the research design so that good research practice is guaranteed – that is, ensuring that the project does not overstep any lines from an ethical perspective (to the extent this can even be judged prior to the start of a project). The same goes for ethics checklists, which often are used as a means of operationalizing the activities of ethics commissions. However, these approaches are not enough. Ethical reflection must become an integral part of processes of technical innovation and must be adapted to each stage of development, something neither ethics commissions nor ethics checklists can accomplish. The praxis of ethical reflection necessary for this integration also goes beyond what has become established as the academic understanding of ethics in the context of RRI. Based on the values of anticipation, reflexivity, inclusion and responsiveness (the dimensions of responsible innovation named by (Stilgoe et al., 2013), the resulting broad-based understanding of ethics unquestionably forms the fertile substrate for ethical integration. However, to conduct exploratory moves that probe the design space based on ethical considerations and enable direct feedback at the leverage points of adjustment, a systematic approach is required that is simultaneously anchored at a deep, namely, conceptual level. As I have already shown, successfully integrating different styles of thinking must be done based on a revised understanding of technology, one that renders artefacts accessible from the standpoint of their contextualized practices. It is the iterative process of exploring, testing, evaluating, and revising which, by deliberate use of methodological distancing, makes it possible to probe the future artefact for weaknesses and to generate robust design knowledge about the impacts on relationships, groups, society, and the environment. Finally, then, I would like to explain in more detail what deep integration of ethics has to offer to approaches that likewise seek to strengthen ethical reflection in research and development processes of technical innovation, and where the commonalities and differences lie. The focus will be on methods and measures that directly address research and development practices (also known as midstream modulation (Fisher et al., 2006) or intra methods (Reijers et al., 2018) and intervene during the early stages (Doorn et al., 2013) at project level. Would it be possible to identify the potential discrimination of the AI app from our example? Do these approaches enable us systematically to probe the design for ethically problematic aspects and to identify weak points such as the Bluetooth specification selected? Participating as an ethicist and thus contributing an additional viewpoint in research and development processes, as the structurally oriented approach of embedded ethics (McLennan et al., 2020; Van Der Burg & Swierstra, 2013; Willem et al., 2024) proposes, can indeed, when it works, lead to convincing outcomes (e.g. Tubig et al., 2024). However, often it is not enough and yet is still too much. It is fine to trigger disconcertedness by introducing the other perspective during encounters in everyday research practice and picking up on ethical issues as and when they arise – but it leaves much to mere coincidence. Ethical analysis of technology and an assessment of the consequences of new developments requires a systematic approach that is not given by simply embedding an ethicist in a work group. The latter strategy additionally requires setting aside considerable resources in terms of time, something which may be feasible in certain research projects but is difficult to scale up due to the number of people required. It is important and proper to embed an ethicist in a project team in cases where, from an ethical point of view, new critical research and development is going on, where there is little experience of the artefact in practice, and where many questions are thrown up that entail entering uncharted territory. In addition to this embedding, however, additional interventions are required so that ethical issues can be raised in a targeted and systematic way, one that fits in well with the different research and development phases. The crucial point here is not to fall into the “division-of-labour trap” where ethical reflection is outsourced to the embedded ethicists. As we have seen, ethical issues need to be dealt with in research and development thinking directly, building on the technical expertise located there. In this regard, the question arises as to whether more might be achieved by less: more ethical reflection through a structured procedure in targeted, temporally limited interventions, that take as their starting point the reflective skills of the researchers, as the method of socio-technical integration research (STIR) proposes (Fisher, 2007; Fisher & Schuurbiers, 2013) and has already been implemented on many occasions (e.g. Flipse et al., 2013; Schuurbiers, 2011; Smolka, 2020; Smolka et al., 2022). STIR stimulates researchers’ capacity for self-reflection by encouraging them to critically question their own activities and to develop alternative ways of doing things. Applied to design praxis, critically questioning decisions is an important requirement for ethical reflection. Ethical reflection must build upon the expertise of the developers, as only they possess the knowledge of technical details required to establish a connection between concrete development trajectories and potential effects. It is this technical knowledge that is needed to identify the leverage points of adjustment and to devise alternative development options. However, the extent to which, prompted by the STIR method, reflections regarding one’s own activities touch on questions of this kind remains uncertain because there is no systematic exploration of ethical issues and aspects of technology assessment embedded in the reflection tool – which constitutes a major shortcoming of this approach. The principles-based approaches promise to redress this shortcoming. A limited number of core principles – (Floridi & Cowls, 2019) argue that only five are sufficient: beneficence, non-maleficence, autonomy, justice, and explicability – offer guidance for novel developments in the field of AI. As (Taddeo et al., 2021, 2024) have shown for the defence domain, subsequent methodological steps make it possible to modulate principles for specific areas of application. While this is the right approach to address domain-specific ethical concerns, more is needed when it comes to addressing ethical issues in the development of concrete artefacts. The most thorough approach to date, namely, VCIO (values, criteria, indicators and observables) (Hallensleben et al., 2020; VDE Verband der Elektrotechnik & Elektronik Informationstechnik e.V., 2022), may serve as an illustration. Its aim being to facilitate an ethics label, the VCIO method breaks down the six values [sic! ] of transparency, accountability, privacy, justice, reliability, and environmental justice – listed as general principles of AI ethics – to the design level via three stages. For example, the abstract value of transparency is broken down into criteria such as “disclosure of properties of algorithm/model used,” which in turn relate to certain indicators (Hallensleben et al., 2020, p. 20). By articulating the latter as questions (for example, “Is it possible to inspect the model so far [sic! ] that potential weaknesses can be discovered?”), they can be linked to concrete observables that usually elicit a yes/no answer (Hallensleben et al., 2020, p. 20). However, this operationalization of values into observables is precisely not capable of building a bridge to exploratory design, as demanded by a deep integration of ethics. It is an approach that remains at the level of what is observable and thus fails to enable any systematic exploration of design thinking, so that the concrete realization of the artefact remains uncertain. This is due not only to the lack of any area of application but also to the blind spot constituted by the mutual interplay of the technologies involved. AI systems are embedded in larger technical systems and generally form complex ensembles; thus, we should not be looking at AI systems in isolation if we want to test their ethical impacts. In the case of the walking pattern-based authentication built into the door opening system, the potential for discrimination lies not only in the AI algorithm but also in interactions with the Bluetooth interface built into the mobile phone – not to mention still other ethical problems associated with this app. There is no doubt that the operationalization used in the VCIO method may prove helpful in cases where general AI problems can be addressed by generalized solutions. However, due to the lack of context around the application and the resultant failure to investigate other ethical aspects (which always need to be examined in each individual case), this approach lulls potential users into a false sense of security and promises simple solutions that are rarely given in emerging technologies. This applies also to the ethics-by-design approach (Brey & Dainow, 2023), which goes a crucial step further. It proposes operationalizing ethical values throughout the different stages of the process of technical development (limited here to AI systems). From specification through to evaluation, the integration of ethical issues is tailored to the respective problems that crop up during the development process. However, it is hard to see how a pre-determined list of values and the restriction to a specific technology can enable ethical issues to be identified beyond these pre-established parameters, and how different technological constellations (which, as in our case, often present in a mixed form) can be addressed: the Bluetooth problem would not have become apparent here. This lack of flexibility in terms of identifying relevant values for each individual development project – common to many approaches seeking to integrate ethics into early design stagesFootnote 28 – can be countered by Value Sensitive Design (VSD) (Friedman et al. 2017; Friedman and Hendry 2019; van de Poel 2020b). Tools such as envisioning cards (Friedman et al., 2024; Friedman & Hendry, 2012) make it possible to take each individual design process as a point of departure. Would it have been possible to use this approach to identify early on the ethical problems of the gait-based authentication app that arise at the interface between AI and Bluetooth technologies? Perhaps. Attention has been drawn to the need to extend the VSD approach specifically for AI applications, and suggestions for doing so have been put forward (Umbrello and van de Poel 2021; van de Poel 2020a). Answers to the question of how to respond better to the lack of differentiation between individual stakeholders‘ preferences and ethical values (Manders-Huits, 2011) inherent to VSD have also been offered (Umbrello, 2020). However, given that the focus of the envisioning cards is on the values of (direct and indirect) stakeholders in relation to the dimensions of time and pervasiveness, any systematic connection between the choice of Bluetooth specification and potential user discrimination can only be established via highly convoluted design moves– if at all. It is the lack of any systematic exploratory approach for studying the effects of detailed technical decisions on the manifestation of specific practices that makes the above-stated “perhaps” unsatisfactory and points to a missing piece of the puzzle, a desideratum. Systematically exploring the impacts of the technology and locating values theoretically in an analysis of practices based on an ethical triangulation between existing, emerging, and ideal practices may prove helpful in this regard. What would also be needed is a process-based model of ethics integration for feeding the results of these studies into the development iteratively, so that each stage of the development can be adapted accordingly. Grunwald recognized very early what was necessary (Grunwald, 2011): technology assessment, the ethics of technology, and an STS-based exploration of societal impacts must be merged together. The early stages of technological research and development require a kind of real-time technology assessment (Guston & Sarewitz, 2002) as part of exploratory ethical investigations that produce initial, “quick and dirty” assessments. Integrating ethics at a deep level is a key to achieving this, because initial investigations can be conducted directly with the exploratory designs – in a way that is at once technologically integrated and analyses the artefact emerging from the interplay of two or more technologies.Footnote 29 This brief account has hopefully illustrated, at least, the importance of deep ethics integration for a rigorous ethical alignment of technological developments that are desirable for society, and where to find its respective systematic entry point for designing suitable tools and interventions. A further, related desideratum, though, is concealed behind this “perhaps.” There is at present no systematic methodical analysis that might provide more precise information about where the strengths and weaknesses of the different approaches lie. What are they suited to demonstrating? What design moves do they encourage? For which stages of development and in which fields of technology or areas of application are they especially useful? What expectations do they place on their users? All these and many more would be pertinent questions to ask. With the theoretical foundation of deep ethics integration in styles of thinking and conceptualizations, we now also have a basis for systematically assessing the different methods and their respective effects in search of further improvements. 6 Conclusion: Empowering Ethical Co-Design Using Deep Ethics Integration As we have seen, deep ethics integration is necessary for integrating ethical reflection directly into exploratory design processes. It is exactly here that it must occur, to ensure that ethical considerations are firmly anchored at the very earliest stages of technological research and development. This does not mean, conversely, that the integration of ethics can stop at this point. It must occur in close conjunction with the development process and does not end, either, with the market launch of the technological product. Deep ethics integration is an important building block and part of an overarching strategy for strongly orienting technical developments towards what is desirable from an ethical standpoint. As with all that is included within the catchword “innovation ecosystems” (Herzog & Blank, 2024; Stahl, 2021, p. 84, 2023), it requires an array of measures – from laws and regulations through to changes in structures of disciplinary training in the technical domains – of which the ethics of technology and technology assessment need to become a taken-for-granted part. Deep ethics integration enables us to address ethical problems in direct relation to exploratory design processes. It thus contributes crucially to ensuring that the developers themselves are actively involved in shaping the ethical orientation of new artefacts. Using the right methods and tools, ethical reflection and technological development are no longer in opposition to each other as contrasting styles of thinking, but rather are brought into a shared space of reflection. The purported gap – not between principles and praxis but between styles of thinking – can be closed by ethical reflection becoming embedded in the reflective moves of exploratory design; this makes it possible to generate sound knowledge about the ethical impacts of the artefact to be developed. In this way the ethical analysis can become part of the iterative development process. The dynamic interplay between “zooming in” and “zooming out,” checking technical details for their societal impacts, and adopting alternating perspectives of maker and critic make it possible to ensure that exploratory designs are geared toward solutions good enough (from an ethical perspective) to proceed with the development. This progress occurs in iterative steps until the knowledge generated regarding ethical impacts is good enough for the current stage of development. With a deep integration of ethics, the role of ethics itself changes as well – from that of a gatekeeper, pronouncing on a (possibly already far advanced) technology, to that of a co-designer of new artefacts. Integrating ethical issues into the reflective moves of technical design effectively eliminates the old division of roles. Research and development teams are placed in a position to make initial assessments about ethical impacts directly in the exploratory design stages and can develop an ability to reflect and talk about such issues. Ethical exploration and testing and the evaluation of potential impacts must progressively be extended and conducted in greater detail in each phase of development as the future application gradually becomes more concrete. Even if the integration of ethics does not end here – it must start with deep ethics integration. For ethics to become a co-designer of technology, it is crucial to enable researchers and developers in the technical disciplines to consider ethical issues as well and to feed the analysis of potential impacts of the technology into their exploratory design processes and development work. As we have seen, it is in the exploratory interplay between technical details and the concrete artefactual practices that the gap described above can be closed. Decisions reached regarding material, form, construction solution, training data and, as in our example, the version of Bluetooth are linked to impacts on user behaviour, society, and the environment. Technological developments can be successfully guided by ethical values when even apparently insignificant technical decisions are repeatedly placed in relation to their consequences within a contextualized investigation. Sounding out different technical design options and excluding societally undesirable consequences requires detailed technical expertise that is simultaneously capable of adopting a reflective ethical attitude. When developers are empowered to address ethical issues in their technical design without themselves having to become ethicists, they can take responsibility for its ethical orientation. This transfer of responsibility constitutes a significant step towards the shared design of technological futures based on ethical values, and calls for a new self-understanding of the technosciences. Data Availability Not applicable. Notes Deep ethics integration refers to an intervention that targets the style of thinking (e.g. basic concepts, reflective moves) and thus laying the foundation for any the content-related and case-specific questions. Ethics integration as an overarching concept addresses the challenge of making ethical considerations effective in technical research and development processes, thus allowing the developer to embrace an ethical perspective throughout the whole life cycle. Ethics integration can be promoted in different ways which should be combined, e.g. by public discourse (developing visions for desired technological futures, identifying key values for good human-technology relations, communicating technology as something that can be shaped), regulations (guidelines, laws, standards), education (technology ethics, technology assessment as a natural part of the curricula in the techno-sciences, an adjusted disciplinary self-image in the technosciences), as well as research and development practice. Ethics integration in research and development practices comprises different kinds of measures such as (a) structural measures (diverse composition of development teams, embedded ethicists in the team, ethics consultation during the development process), b) procedural measures (early integration of different stakeholders, maintaining decision-making leeway, iterative approach), c) ethical quality assurance measures (regular ethics tests, audits, TÜV). My intention here is not to obscure the fact that there is a marked theoretical difference between principles and values, reflected in their different roles in research and development processes. To elaborate on this issue, however, would go beyond the scope of this paper and must be saved for a later study. The concept of reflective moves borrows from Donald Schön’s concept of “web of moves” as a key element of “reflection-in-action” (Schön, 1983, p. 128 ff.). The term ‘artefact’ as I see it can be conceptualized quite broadly and covers much of the spectrum of what is made, from the material thing through to process design. What I am not referring to is its widespread meaning in the engineering sciences as a distortion or measurement error. Due to a lack of differentiation in the English language comparable to that between “Technologie” (abstract operating principles) and “Technik” (overarching term) in German, I will use the term “technology” to refer to abstract operating principles, “artefact” when I turn to the concrete technical object and its embedded practices, and the German word Technik as an overarching term that refers to technical matters in a broad sense, such as techniques, technology, artefacts, components etc. The adjectives “technological” and “technical” are employed respectively in the following. My use of the term ‘style of thinking’ constitutes a loose reference to philosopher of science Ludwik Fleck and his concept of thought styles (Fleck, 1980), which highlights disciplinary differences. How it relates to concepts from historical epistemology such as Michel Foucault’s ‘episteme’, Thomas Kuhn’s ‘paradigm’, Imre Lakatos’s research programme, or Ian Hackings ‘styles of reasoning’ (Sciortino, 2017) would need to be discussed elsewhere. Rightness refers here to an epistemological concept introduced by Catherine Elgin and Nelson Goodman (Elgin, 1996; Goodman, 1988, 1954, 1978); cf. (Ammon, 2009, 2016). Grounded in a processual epistemology, rightness is framed by Elgin and Goodman as a dynamic process of finding an equilibrium which is marked by fitting (a claim into a context) and working (as the effect of the resultant whole which leads to an epistemic ascend if successful). https://en.wikipedia.org/wiki/Bluetooth#Bluetooth_5.4 [accessed: 21.11.2025]. The theory of technology neutrality assumes that it is not the artefact itself but rather the people who use it that decides whether positive or negative consequences will ensue. The artefact itself is neutral regarding the actions carried out with it. A knife can be used for eating or for killing – according to the neutrality theory it is not the shape or design of the knife but the intentions of the person using it that decide whether they will commit a murder with it or still their appetite. In line with the theory of neutrality, then, whether or not BL652-SA is discriminatory has nothing to do with the component’s design and everything to do with the way it is used. https://de.wikipedia.org/wiki/Bluetooth [accessed: 21.11.2025]; https://en.wikipedia.org/wiki/Bluetooth [accessed: 21.11.2025]. https://de.wikipedia.org/wiki/Funktechnik [accessed: 21.11.2025]. https://en.wikipedia.org/wiki/International_Agency_for_Research_on_Cancer [accessed: 7.3.2025], https://www.iarc.who.int/ [accessed: 7.3.2025]. https://en.wikipedia.org/wiki/Bluetooth#Health_concerns [accessed: 21.11.2025]; for more information on the IARC’s classification of Radiofrequency (RF) as ‘possibly carcinogenic’ see also https://www.iarc.who.int/video/the-annual-science-and-wireless-series-hosts-iarc-scientist-in-november-2011-the-event-provided-an-opportunity-to-consider-the-implications-of-the-international-agency-for-research-into-cancer-iarc/ [accessed: 7.3.2025]. The component can be interpreted as a material instantiation of the technology. This re-conceptualization of the concept of technology is influenced by ideas put forward by philosopher of technology Gilbert Simondon (Simondon, 2012), some of which are echoed in the above account; the same applies to actor-network theory. A more detailed account of these approaches, and one that compares them to divergent accounts of artefacts, must be elaborated elsewhere. Of key importance here is the concern to demonstrate the need for a processual, contextualized account of Technik in contrast to components, technology, and artefacts-only perspectives. A closer consideration (not undertaken here, given the issue at hand) would affirm that this is especially true of completely new product developments. Otherwise, variants of the artefact do exist in physical form when, for example, only certain of their aspects are to be re-worked or updated. During later stages of product development work is done on prototypes and mock-ups, which serves to clarify certain specific issues about the material object. Normativity thus becomes a specific instance of formativity. Openness to getting involved in discussions, curiosity, and a constructive attitude are helpful for entering into a shared process of generating knowledge. Anti-social behaviour that operates with disparagement and intimidation, is unable to deal with others on an equal footing, that takes up a disproportionate amount of speaking time, demands sustained focus and has to be the centre of attention, can make entry into a shared process of knowledge generation difficult to impossible. Attempts to counter such behaviour can be made by actively establishing a culture of collaboration that includes rules of communication. Without wishing to discuss this in detail, it should be mentioned at this point that the second type of immunization preventing the integration of ethics is removed in this way. The first type was the immunization of technology, which has been attributed a supposed neutrality; as we have already seen above, this is incorrect. The second type of immunization is that of the developer who is reduced to being the bearer of disciplinary knowledge which they contribute to the development process. The reduction of the developer to a repository of disciplinary knowledge is factually incorrect as well as being a misleading ideal. In our case – a concrete technological development – those leverage points of adjustment are of particular interest which can actually be influenced in the development process. The choice of priorities depends on the composition of actors involved and the goal being pursued. If, for example, the idea is to develop organizational ethical guidelines or a legal framework, other adjustable elements need to be looked at more closely. Depending on the product, it may also be crucial to explore these aspects in the context of aesthetic requirements. The crucial role of concepts in the development of new technologies has also been pointed out in the context of conceptual engineering (Veluwenkamp, 2025; Veluwenkamp et al., 2022). In the development process, they can be framed as specific concepts which offer entry points for strategic conceptual changes, as has been argued by Ibo van de Poel (van de Poel, 2025). By integrating ethical considerations into design thinking, these explorations can be conducted without any speculative assumptions about possible futures which, in most cases, would entail an overwhelming degree of complexity and uncertainty. For an overview of existing tools, see https://reinforcing.eu/oss-platform/services/tools [accessed 11.3.2025]; to browse some of the tools used for systematic exploration of design and development processes by the BerlinEthics Lab, see https://www.tu.berlin/en/philtech/berlin-ethics-lab/ethics-toolbox[accessed 18.3.2026]. Prohibition needs to be one of the options for revision. “Provisionally approved”, as a figure of thought is inspired by Karl Popper’s notion of theories being corroborated (Popper, 1959, p. 10). Popper insisted that scientific progress lies in the tentative acceptance of a theory until and unless it is falsified; this requires an ongoing awareness of indicators of failure. Although beyond the scope of this article, it should be mentioned that a comprehensive ethics integration is grounded in a shift in the self-understanding of the tech disciplines along with an integration of technology ethics and assessment in engineering (e.g. Ammon et al., 2022, 2023) and computer sciences curricula (e.g. Grosz et al., 2019). This, however, remains wishful thinking in many parts of the world, even more so when considering the current backlash against ethics. (Sætra & Danaher, 2022) have drawn attention to the danger of ethics proliferation. What is needed is not just an integrated ethics approach but also an integrated technology approach. References Ammon, S. (2009). Wissen verstehen: Perspektiven einer prozessualen Theorie der Erkenntnis (1. Aufl., p. 206S). Velbrück Wissenschaft. Ammon, S. (2016). Explaining understanding, understanding knowledge. Explaining Understanding. Routledge. Ammon, S. (2017a). Epilogue: the rise of imagery in the age of modeling. In S. Ammon & R. Capdevila-Werning (Eds.), The active image: architecture and engineering in the age of modeling (pp. 287–312). Springer. https://doi.org/10.1007/978-3-319-56466-1_12author Ammon, S. (2017b). Image-based epistemic strategies in modeling: Designing architecture after the digital turn. In S. Ammon & R. Capdevila-Werning (Eds.), The active image: architecture and engineering in the age of modeling (pp. 177–205). Springer. https://doi.org/10.1007/978-3-319-56466-1_8 Ammon, S. (2017c). Why designing is not experimenting: design methods, epistemic praxis and strategies of knowledge acquisition in architecture. Philosophy & Technology, 30(4), 495–520. https://doi.org/10.1007/s13347-017-0256-4 Ammon, S. (2019). Drawing inferences: thinking with 6B (and Sketching Paper). Philosophy & Technology, 32(4), 591–612. https://doi.org/10.1007/s13347-018-0323-5 Ammon, S. (2020). Design methods and validation. The Routledge Handbook of the Philosophy of Engineering. Routledge. Ammon, P. P., Derda, S., Hildebrandt, M., Kühne, T., S., & Klein, N. (2023). Eng+ | Ethics for sustainable development in engineering programmes: A praxis report of TU Berlin’s Think Tank Technology Reflection. Practice Papers. https://doi.org/10.21427/F2YD-N075 Ammon, S., Kljagin, A., Rettschlag, J., & Vortel, M. (2022). The Berlin ethics certificate: Conceptualizing interdisciplinarity as a core building block of ethics in engineering education. Towards a new future in engineering education, New scenarios that european alliances of tech universities open up, 913–924. https://doi.org/10.5821/conference-9788412322262.1422 Berghaeuser, H., Prass, M., & Lindner, R. (2024). Researchers’ practice and perception of research ethics and the role of institutional support: Insights from a pan-European researcher survey. Journal of Academic Ethics. https://doi.org/10.1007/s10805-024-09590-z Bleher, H., & Braun, M. (2023). Reflections on putting AI ethics into practice: How three AI ethics approaches conceptualize theory and practice. Science and Engineering Ethics, 29(3), 21. https://doi.org/10.1007/s11948-023-00443-3 Boenink, M., & Kudina, O. (2020). Values in responsible research and innovation: From entities to practices. Journal of Responsible Innovation, 7(3), 450–470. https://doi.org/10.1080/23299460.2020.1806451 Borning, A., & Muller, M. (2012). Next steps for value sensitive design. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, CHI ’12, 1125–1134. https://doi.org/10.1145/2207676.2208560 Brey, P., & Dainow, B. (2023). Ethics by design for artificial intelligence. AI and Ethics. https://doi.org/10.1007/s43681-023-00330-4 Burget, M., Bardone, E., & Pedaste, M. (2017). Definitions and conceptual dimensions of responsible research and innovation: A literature review. Science and Engineering Ethics, 23(1), 1–19. https://doi.org/10.1007/s11948-016-9782-1 Collingridge, D. (1980). The social control of technology (p. 200 S.: graph.,Darst.). Pinter. Corrêa, N. K., Galvão, C., Santos, J. W., Pino, C. D., Pinto, E. P., Barbosa, C., Massmann, D., Mambrini, R., Galvão, L., Terem, E., & de Oliveira, N. (2023). Worldwide AI ethics: A review of 200 guidelines and recommendations for AI governance. Patterns, 4(10). https://doi.org/10.1016/j.patter.2023.100857 Corrêa, N. K., Santos, J. W., Galvão, C., Pasetti, M., Schiavon, D., Naqvi, F., Hossain, R., & Oliveira, N. D. (2024). Crossing the principle–practice gap in AI ethics with ethical problem-solving. AI and Ethics. https://doi.org/10.1007/s43681-024-00469-8 Davis, J. L. (2020). How artifacts afford: The power and politics of everyday things. The MIT Press. Doorn, N., Schuurbiers, D., Van De Poel, I., & Gorman, M. E. (Eds.). (2013). Early engagement and new technologies: Opening up the laboratory (Vol. 16). Springer Netherlands. https://doi.org/10.1007/978-94-007-7844-3 Elgin, C. Z. (1996). Considered judgment. Princeton University Press. European Commission AI HLEG (2019). Ethics guidelines for trustworthy AI | Shaping Europe’s digital future. https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai European Commission. (2015). Indicators for promoting and monitoring responsible research and innovation: Report from the Expert Group on policy indicators for responsible research and innovation. Publications Office of the European Union. https://data.europa.eu/doi/https://doi.org/10.2777/9742 European Commission, A. I. H. L. E. G. (2020). Assessment List for Trustworthy Artificial Intelligence (ALTAI) for self-assessment | Shaping Europe’s digital future. https://digital-strategy.ec.europa.eu/en/library/assessment-list-trustworthy-artificial-intelligence-altai-self-assessment Fiedler, L., Remmers, P., & Ammon, S. (2025). Ethics consultation for integrating ethics in technology readiness levels (TRL). Journal of Responsible Innovation, 12(1), Article2445321. https://doi.org/10.1080/23299460.2024.2445321 Fisher, E. (2007). Ethnographic Invention: Probing the Capacity of Laboratory Decisions. NanoEthics, 1(2), 155–165. https://doi.org/10.1007/s11569-007-0016-5 Fisher, E., & Schuurbiers, D. (2013). Socio-technical integration research: Collaborative inquiry at the midstream of research and development. In N. Doorn, D. Schuurbiers, I. van de Poel, & M. E. Gorman (Eds.), Early engagement and new technologies: Opening up the laboratory (pp. 97–110). Springer Netherlands. https://doi.org/10.1007/978-94-007-7844-3_5 Fisher, E., Mahajan, R. L., & Mitcham, C. (2006). Midstream modulation of technology: Governance from within. Bulletin of Science Technology & Society, 26(6), 485–496. https://doi.org/10.1177/0270467606295402 Fjeld, J., Achten, N., Hilligoss, H., Nagy, A., & Srikumar, M. (2020). Principled Artificial Intelligence: Mapping consensus in ethical and rights-based approaches to principles for AI. SSRN Scholarly Paper 3518482. https://doi.org/10.2139/ssrn.3518482 Fleck, L. (1980). Entstehung und Entwicklung einer wissenschaftlichen Tatsache: Einführung in die Lehre vom Denkstil und Denkkollektiv (1. Aufl). Suhrkamp. Flipse, S. M., van der Sanden, M. C. A., & Osseweijer, P. (2013). Midstream modulation in biotechnology industry: Redefining what is ‘part of the job’ of researchers in industry. Science and Engineering Ethics, 19(3), 1141–1164. https://doi.org/10.1007/s11948-012-9411-6 Floridi, L. (2019). Translating principles into practices of digital ethics: Five risks of being unethical. Philosophy & Technology, 32(2), 185–193. https://doi.org/10.1007/s13347-019-00354-x Floridi, L., & Cowls, J. (2019). A unified framework of five principles for AI in society. Harvard Data Science Review, 1(1). https://doi.org/10.1162/99608f92.8cd550d1 Friedman, B., & Hendry, D. (2012). The envisioning cards: A toolkit for catalyzing humanistic and technical imaginations. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, CHI ’12, 1145–1148. https://doi.org/10.1145/2207676.2208562 Friedman, B., & Hendry, D. G. (2019). Value Sensitive Design: Shaping Technology with Moral Imagination. The MIT Press. https://doi.org/10.7551/mitpress/7585.001.0001 Friedman, B., & Kahn, P. H. (2002). Human values, ethics, and design. The human-computer interaction handbook: Fundamentals, evolving technologies and emerging applications (pp. 1177–1201). L. Erlbaum Associates Inc. Friedman, B., Kahn, P. H., Borning, A., & Huldtgren, A. (2013). Value sensitive design and information systems. In N. Doorn, D. Schuurbiers, I. van de Poel, & M. E. Gorman (Eds.), Early engagement and new technologies: Opening up the laboratory (pp. 55–95). Springer Netherlands. https://doi.org/10.1007/978-94-007-7844-3_4 Friedman, B., Hendry, D. G., & Borning, A. (2017). A survey of value sensitive design methods. Foundations and Trends® in Human–Computer Interaction, 11(2), 63–125. https://doi.org/10.1561/1100000015 Friedman, B., Nathan, L., Kane, S., Lin, J., Yoo, D., Logler, N., Ballard, S., & Hendry, D. G. (2024). Envisioning Cards: A Value Sensitive Design Toolkit (2nd Edition). University of Washington. https://envisioningcards.vsdesign.org Gogoll, J., & Zuber, N. (2026). Introduction to ethical software development. Springer Nature Switzerland. https://doi.org/10.1007/978-3-032-06195-9 Goodman, N. (1906–1998). (1954). Fact, fiction & forecast (p. 126 S.). London: Athlone Press. Goodman, N. (1906–1998). (1978). Ways of worldmaking / (p. Illustrationen.). Indianapolis, Indiana:Hackett publishing company. Goodman, N. (with, & Elgin, C. Z. (1988). Reconceptions in philosophy and other arts and sciences (1. publ.). Routledge. Grosz, B. J., Grant, D. G., Vredenburgh, K., Behrends, J., Hu, L., Simmons, A., & Waldo, J. (2019). Embedded EthiCS: Integrating ethics across CS education. Communications Of The Acm, 62(8), 54–61. https://doi.org/10.1145/3330794 Grunwald, A. (2011). Responsible innovation: Bringing together technology assessment, applied ethics, and STS research. Enterprise and work innovation studies, 7, 9–31. Grunwald, A. (2019). Technology assessment in practice and theory. Routledge. Guston, D. H., & Sarewitz, D. (2002). Real-time technology assessment. Technology in Society, American Perspectives on Science and Technology Policy, 24(1), 93–109. https://doi.org/10.1016/S0160-791X(01)00047-1 Hagendorff, T. (2020). The ethics of AI ethics: An evaluation of guidelines. Minds and Machines, 30(1), 99–120. https://doi.org/10.1007/s11023-020-09517-8 Hallensleben, S., Fetic, L., Fleischer, T., Grünke, Hagendorff, T., Hauer, M., Hauschke, A., Heesen, J., Herrmann, M., Hillerbrand, R., Hubig, C., Kaminski, A., Krafft, T., Loh, W., Otto, P., Puntschuh, M., & Hustedt, C. (2020). From Principles to Practice: An interdisciplinary framework to operationalise AI ethics. https://www.bertelsmann-stiftung.de/fileadmin/files/BSt/Publikationen/GrauePublikationen/WKIO_2020_final.pdf Herzog, C., & Blank, S. (2024). A systemic perspective on bridging the principles-to-practice gap in creating ethical artificial intelligence solutions – a critique of dominant narratives and proposal for a collaborative way forward. Journal of Responsible Innovation, 11(1), 2431350. https://doi.org/10.1080/23299460.2024.2431350 Iphofen, R. (Ed.). (2020). Handbook of research ethics and scientific integrity. Springer. Jobin, A., Ienca, M., & Vayena, E. (2019). The global landscape of AI ethics guidelines. Nature Machine Intelligence, 1(9), 389–399. https://doi.org/10.1038/s42256-019-0088-2 Koepsell, D. (2017). Scientific Integrity and Research Ethics. Springer International Publishing. https://doi.org/10.1007/978-3-319-51277-8 Manders-Huits, N. (2011). What values in design? The challenge of incorporating moral values into design. Science and Engineering Ethics, 17(2), 271–287. https://doi.org/10.1007/s11948-010-9198-2 McLennan, S., Fiske, A., Celi, L. A., Müller, R., Harder, J., Ritt, K., Haddadin, S., & Buyx, A. (2020). An embedded ethics approach for AI development. Nature Machine Intelligence, 2(9), 488–490. https://doi.org/10.1038/s42256-020-0214-1 Morley, J., Floridi, L., Kinsey, L., & Elhalal, A. (2020). From what to how: An initial review of publicly available AI ethics tools, methods and research to translate principles into practices. Science and Engineering Ethics, 26(4), 2141–2168. https://doi.org/10.1007/s11948-019-00165-5 Morley, J., Elhalal, A., Garcia, F., Kinsey, L., Mökander, J., & Floridi, L. (2021). Ethics as a service: A pragmatic operationalisation of AI ethics. Minds and Machines, 31(2), 239–256. https://doi.org/10.1007/s11023-021-09563-w Morley, J., Kinsey, L., Elhalal, A., Garcia, F., Ziosi, M., & Floridi, L. (2023). Operationalising AI ethics: Barriers, enablers and next steps. AI & SOCIETY, 38(1), 411–423. https://doi.org/10.1007/s00146-021-01308-8 Norman, D. A. (2013). The design of everyday things. Basic Books. Revised and expanded edition. Owen, R., Macnaghten, P., & Stilgoe, J. (2012). Responsible research and innovation: From science in society to science for society, with society. Science and Public Policy, 39(6), 751–760. https://doi.org/10.1093/scipol/scs093 Owen, R., Bessant, J. R., & Heintz, M. (2013). Responsible innovation: Managing the responsible emergence of science and innovation in society. Wiley. Popper, K. (1959). The logic of scientific discovery. Harper & Row. Reijers, W., Wright, D., Brey, P., Weber, K., Rodrigues, R., O’Sullivan, D., & Gordijn, B. (2018). Methods for practising ethics in research and innovation: A literature review, critical analysis and recommendations. Science and Engineering Ethics, 24(5), 1437–1481. https://doi.org/10.1007/s11948-017-9961-8 Rohde, F., Nasruddin, Z., Kotova, E., & Ammon, S. (2026). Navigating justice in AI lifecycles: Ethical perspectives on infrastructural prerequisites and their environmental impacts. AI and Ethics, 6(1), 156. https://doi.org/10.1007/s43681-026-01008-3 Sætra, H. S., & Danaher, J. (2022). To each technology its own ethics: The problem of ethical proliferation. Philosophy & Technology, 35(4), 93. https://doi.org/10.1007/s13347-022-00591-7 Schön, D. A. (1983). The reflective practitioner: How professionals think in action. Basic Books. Schuurbiers, D. (2011). What happens in the Lab: Applying Midstream Modulation to Enhance Critical Reflection in the Laboratory. Science and Engineering Ethics, 17(4), 769–788. https://doi.org/10.1007/s11948-011-9317-8 Sciortino, L. (2017). On Ian Hacking’s notion of style of reasoning. Erkenntnis, 82(2), 243–264. https://doi.org/10.1007/s10670-016-9815-9 Simon, H. A. (Ed.). (1916–2001). (1996). The sciences of the artificial / (3. ed., p. XIV, 231 S.: Ill.). Cambridge, Mass. :MIT Press. Simondon, G. (2012). Die Existenzweise technischer Objekte (M. Cuntz, Ed.; 1st edn). Diaphanes. http://www.gbv.de/dms/weimar/toc/682277738_toc.pdf Smolka, M. (2020). Generative critique in interdisciplinary collaborations: From critique in and of the neurosciences to socio-technical integration research as a practice of critique. R(R)I NanoEthics, 14(1), 1–19. https://doi.org/10.1007/s11569-019-00362-3 Smolka, M., Fisher, E., Pickering, C., & Peate, L. (2022). Traveling through the past and into the future of Socio-Technical Integration Research (STIR): Midpoint report on the 2022 STIR Seminar Series. EASST Review, 41(2). https://easst.net/easst-review/41-2/traveling-through-the-past-and-into-the-future-of-socio-technical-integration-research-stir-midpoint-report-on-the-2022-stir-seminar-series/ Stahl, B. C. (2021). Artificial Intelligence for a better future: An ecosystem perspective on the ethics of AI and emerging digital technologies. Springer Nature. https://doi.org/10.1007/978-3-030-69978-9 Stahl, B. C. (2023). Embedding responsibility in intelligent systems: From AI ethics to responsible AI ecosystems. Scientific Reports, 13(1), 7586. https://doi.org/10.1038/s41598-023-34622-w Stilgoe, J., Owen, R., & Macnaghten, P. (2013). Developing a framework for responsible innovation. Research Policy, 42(9), 1568–1580. https://doi.org/10.1016/j.respol.2013.05.008 Taddeo, M., McNeish, D., Blanchard, A., & Edgar, E. (2021). Ethical Principles for Artificial Intelligence in National Defence. Philosophy & Technology, 34(4), 1707–1729. https://doi.org/10.1007/s13347-021-00482-3 Taddeo, M., Blanchard, A., & Thomas, C. (2024). From AI ethics principles to practices: A teleological methodology to apply ai ethics principles in the defence domain. Philosophy & Technology, 37(1), 42. https://doi.org/10.1007/s13347-024-00710-6 Tubig, P. A., Brown, T. E., McCusker, D., Pham, M., Klein, E., & Goering, S. (2024). From an ethics ‘side dish’ to ‘your modus operandi’: Neurotechnology researcher perspectives on the impacts of a decade of embedded ethics collaboration. Journal of Responsible Innovation, 11(1), 2414493. https://doi.org/10.1080/23299460.2024.2414493 Umbrello, S. (2020). Combinatory and complementary practices of values and virtues in design: A reply to Reijers and Gordijn. Filosofia, 65, Article65. https://doi.org/10.13135/2704-8195/5236 Umbrello, S., & van de Poel, I. (2021). Mapping value sensitive design onto AI for social good principles. Ai and Ethics, 1(3), 283–296. https://doi.org/10.1007/s43681-021-00038-3 van de Poel, I. (2020a). Embedding values in Artificial Intelligence (AI) systems. Minds and Machines, 30(3), 385–409. https://doi.org/10.1007/s11023-020-09537-4 van de Poel, I. (2020b). Values and Design. The Routledge Handbook of the Philosophy of Engineering. Routledge. van de Poel, I. (2025). The distinction between generic and specific concepts and why it matters for conceptual engineering. Inquiry : A Journal Of Medical Care Organization, Provision And Financing, 0(0), 1–25. https://doi.org/10.1080/0020174X.2025.2556750 Van Der Burg, S. & Swierstra, T. (2013). In T. (Ed.), Ethics on the Laboratory Floor. Palgrave Macmillan UK. https://doi.org/10.1057/9781137002938 VDE Verband der Elektrotechnik & Elektronik Informationstechnik e.V. (Eds.) (2022). VCIO based description of systems for AI trustworthiness characterisation. https://www.vde.com/resource/blob/2242194/a24b13db01773747e6b7bba4ce20ea60/vcio-based-description-of-systems-for-ai-trustworthiness-characterisationvde-spec-90012-v1-0--en--data.pdf?trk=public_post_comment-text Veluwenkamp, H. (2025). Impactful conceptual engineering: Designing technological artefacts ethically. Ethical Theory and Moral Practice, 28(1), 155–170. https://doi.org/10.1007/s10677-024-10459-8 Veluwenkamp, H., Capasso, M., Maas, J., & Marin, L. (2022). Technology as driver for morally motivated conceptual engineering. Philosophy & Technology, 35(3), 71. https://doi.org/10.1007/s13347-022-00565-9 Von Schomberg, R. (2011). Towards Responsible Research and Innovation in the Information and Communication Technologies and Security Technologies Fields (SSRN Scholarly Paper No. 2436399). Social Science Research Network. https://papers.ssrn.com/abstract=2436399 Willem, T., Fritzsche, M. C., Zimmermann, B. M., Sierawska, A., Breuer, S., Braun, M., Ruess, A. K., Bak, M., Schönweitz, F. B., Meier, L. J., Fiske, A., Tigard, D., Müller, R., McLennan, S., & Buyx, A. (2024). Embedded ethics in practice: A toolbox for integrating the analysis of ethical and social issues into healthcare AI research. Science and Engineering Ethics, 31(1), 3. https://doi.org/10.1007/s11948-024-00523-y Zhou, J., & Chen, F. (2023). AI ethics: From principles to practice. AI & SOCIETY, 38(6), 2693–2703. https://doi.org/10.1007/s00146-022-01602-z Acknowledgments Not applicable. Funding Open Access funding enabled and organized by Projekt DEAL. Author information Authors and Affiliations Contributions Not applicable. Corresponding author Ethics declarations Competing Interests Not applicable. Additional information Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. About this article Cite this article Ammon, S. Deep Ethics Integration in Technology Research and Development: Bridging the Principles-to-Practice Gap in AI Ethics – and Elsewhere. Philos. Technol. 39, 165 (2026). https://doi.org/10.1007/s13347-026-01134-0 Received: Accepted: Published: Version of record: DOI: https://doi.org/10.1007/s13347-026-01134-0