Background and relevance to quantum AI

“Ethics by committee/commission” is a popular approach among practitioners who are interested in ethical aspects of their decision-making. This approach consists of assembling a team of usually both inhouse and external experts who are supposed to come up with a set of ethical guidelines that organizations and businesspersons ought to follow (Why You Need an AI Ethics Committee 2023). As it turns out from time to time, however, not every committee’s work or not all ethical guidelines are taken seriously. They might have little to no impact on informing actual decision-making of the practitioners that stepped up earlier and made a case for establishing the committee in the first place. In other words, there exists a veritable risk of “ethics washing” if (too) many (industry) people are on the board that have an interest in coming up with vague ethical goals and in avoiding ethics-based regulation, while appearing to care about ethics to strengthen the company’s license to operate.Footnote ²

Presentation of the argument

P1: People A, B, C, … claim that quantum computing and AI ethics are important in business.

P2: People A, B, C, … are experts in quantum AI and business.

Thus: Quantum AI ethics are important to businesspersons.

This is a fallacious argument. If A, B, C claimed that it is important to businesspersons to know the number of certified sommeliers in their organizations, it would also not necessarily follow that this is of actual importance. Yet, arguments from authority and, particularly, ethics by committee are quite common and do have steering power for the practice of quantum AI technology.

AI and particularly quantum technology (Vermaas, Reference Vermaas2017) are complex and abstract. Quantum mechanics, with its counterintuitive principles such as superposition and entanglement, is even being used to inspire and further general philosophical discourse in ethics as well as the humanities and social sciences (Voelkner and Zanotti, Reference Voelkner and Zanotti2022). Unsurprisingly, therefore, many myths and half-truths exist around quantum computing, propagated by many “authorities”. For example:

• Quantum computers will (soon) replace all classical ones;

• Quantum computers will make all calculations faster;

• Quantum computers require a lot of energy;

• Quantum computers are best for (classical) big data problems.

Hence, the constitution, competence and working processes of a quantum AI ethics committee must be even more carefully considered.

Example

Leveraging internal research as well as published work that illuminates different facets of ethics committees and approaches to the topic (Schrag, Reference Schrag2011), the center of competence for quantum and AI “QuantumBasel” (part of the innovation campus “uptownBasel” in Switzerland) has from the start been exploring the role ethics guidelines and advisors can and should play in an ecosystem which develops cutting-edge technology. In order to foster proactive engagement with ethics and minimize “ethics washing,” the approach has encompassed:

• Exploration of guiding principles and their applicability to projects conducted with partner companies;

• Advisors on quantum AI and ethics who have served in other ethics committees and roles and know about the risks around “ethics washing”;

• Empowered employees and advisors who proactively raise topics that should be addressed.

Recommendations

• Articulate clearly how a quantum AI committee and advisors should make decisions – and on what grounds.

• Carry out “postmortem” workshops to proactively address “ethics washing” and related risks.

• Empower the committee and advisors.

Background and relevance to quantum AI

Business leaders often care about maximizing profits and not about maximizing the number and impact of ethical actions per se. At least this has been the traditional approach, reflected in the key target of maximizing shareholder value, although recent developments have brought greater attention to aspects such as sustainability and general social responsibility (Denning, Reference Denning2019).

Still, perhaps there is room for ethics to assist with reaching profit goals. Indeed, ethics are (at least) of instrumental value for companies to develop products which meet regulatory requirements. Engagement with ethics thus turns out to be a competitive advantage, becoming increasingly relevant in a hyperconnected world where brands and reputations are built or impacted within weeks, if not days.

Ethical standards and strong moral beliefs of society members are sooner or later expressed and reflected in regulation. The following risks and insights should be highlighted to clarify the relationship between ethics and regulation further:

• Beware of overregulation.

  • ○ Regulation that results in new bureaucracy and potentially unnecessary overhead should not be created hastily. Since such bureaucracy requires financing and triggers follow-up costs in the economy, one needs to carefully consider if sufficient value is created. However, before this question is answered, another one should be addressed. Rationality requires one to first tackle the question of whether to regulate and according to which principles. Both aspects point to the realm of ethics. Only then the “how” becomes relevant, which is of primary concern to legislation.

• Responsibility guidelines are needed first.

  • ○ It is not clear which governmental and institutional bodies should be in the lead for which ethical questions. Thus, this must be contemplated before concrete principles and laws can be developed.

• Legislation is restricted to a certain jurisdiction. Ethics, by contrast, can be argued to have universal features.

  • ○ As the world is still far away from globally applicable legislation in most fields (if this is desirable at all, given that at least countries with a long tradition of liberalism may be skeptical here), the question arises if ethics-related regulation can be formulated generally enough to be widely adopted without ending up being watered down too much.

Regulation on technology can be severe if society deems it necessary (consider human cloning (Langlois, Reference Langlois2017) or biological weapons (Biological Weapons Convention 2023)). Companies voluntarily pulling out of facial recognition research (Why It Matters That IBM Has Abandoned Its Facial Recognition Technology 2020) and the calls for a global AI moratorium (Miotti and Wasil, Reference Miotti and Wasil2023) indicate that at least certain aspects of quantum AI are likely to face regulation.

In order to strengthen trust in the responsible use of such emerging technologies, it is very important to sensitize the public to the potential risks of quantum AI applications. Clear principles must be developed to ensure that AI is used responsibly, ethical standards are adhered to and data protection is guaranteed. These include transparency, fairness, security and accountability. Ethical and legal aspects must be taken into account in order to prevent abuse and discrimination. Quantum AI models reflect the data with which they are trained and the people who program them. It is crucial that existing biases and discrimination in society are not reproduced.

Furthermore, if quantum AI really does live up to its lofty expectations, this will bring the question into the foreground concerning who has access to such powerful technology. AI, including large language models and foundation models, already cost millions, and soon perhaps billions, of US dollars to train (The cost of training AI could soon become too much to bear 2024). Access to quantum computers is (still) relatively limited, and there are significant disparities between different regions (State of Compute Access: How to Bridge the New Digital Divide 2023). Organizations or countries who are not leaders in this space may argue that they should not be excluded from the positive benefits of breakthroughs, for instance longer life expectancies due to the design of novel drugs.

Presentation of the argument

P1: Moral beliefs in society shape tomorrow’s regulation.

P2: Ethics is the branch of philosophy which deals with questions of human morality.

P3: Items of regulatory importance are of practical importance to businesspersons.

Thus: Ethics is important to businesspersons.

Premises 1 and 2 are straightforward. Premise 3 can be rationalized once one recalls that regulation decides on what defines a market-compliant product which businesses aim to sell. In this way, engaging with ethics increases revenue or at least prevents revenue loss (The Business Risks of Poor AI Ethics and Governance 2022).

Example

A variety of guidelines has been issued across the world around ethical AI (Jobin et al., Reference Jobin, Ienca and Vayena2019). While the precise relationship between ethics and AI legislation continues to stir debate (Anderson, Reference Anderson2022; Catanzariti, Reference Catanzariti2023), governments are moving from AI guidelines toward binding regulation now. For example, the EU AI Act represents one of the world’s first efforts to regulate AI (EU agrees “historic” deal with world’s first laws to regulate AI 2023). It is designed to help AI uptake while banning risky AI technology and applications (Schuett, Reference Schuett2023). Such legislation will significantly influence how businesses engineer and apply AI models.

One contentious point in the act concerns the development and application of foundation models; (ethical) arguments around regulating foundation models include their use in the development of general-purpose AI. However, restriction of foundation models would arguably limit the competitiveness of EU companies compared with the rest of the world (Will Disagreement Over Foundation Models Put the EU AI Act at Risk? 2023). Hence, the ethics surrounding (quantum) AI regulation have a direct impact on businesspersons.

Recommendations

• Be clear about which primary targets your business has – shareholder value? Ethics? Social goals? Long-term survival?

• Be flexible and adapt as new laws loom based on shifts in technological progress and societal values.

• Be mindful of international differences in regulation.

Background and relevance to quantum AI

The picture that quantum and AI industries and practices present us with is not black and white, but rather complex, both from a descriptive and from a normative perspective. The focus of reform efforts toward responsible and trustworthy AI has been on legal regulation: changes in the rules and regulations that govern AI systems. While doubtlessly important, however, one wonders whether such changes in legal regulation are sufficient or whether they need to be accompanied by changes in the ethos and the culture of AI and quantum-related markets.

In systems as complex as today’s economic systems, governance by rules requires getting the incentives right, which is no easy task. Rules are, by definition, rather rigid and may not be ideal as the only tool for regulating a wide variety of cases. If circumstances change quickly, for instance due to the rapid and often not-easily-understood advances and developments of quantum AI, rules may become quickly outdated and those setting them may have trouble catching up. Rules need to be applied to concrete cases, which can require some degree of judgment and a joint understanding of the practices they are supposed to regulate.

Furthermore, the control of whether rules are obeyed is time-consuming and costly. These various factors imply that it is highly desirable that complex systems be regulated not exclusively by rules but also by a joint ethos of those who participate in the systems – a professional ethos that embodies an orientation toward the standards implicit in the practices they are active in.Footnote ³

Quantum AI exacerbates these issues and can easily result in black box models and approaches. For example, models developed with deep neural networks are typically difficult for humans to comprehend, and their outputs are not easily explainable. However, it has been argued that a “right to explanation” is present when algorithmic decision-making is involved (Kim and Routledge, Reference Kim and Routledge2022). Quantum computing, by its very nature, is probabilistic; this may create replicability challenges, further complicated by the noise affecting current generations of quantum systems. Thus, the complexity and pace associated with the technology mean that individual ethos must always accompany general regulation.

Presentation of the argument

P1: Quantum AI technology is complex and embedded in complex social systems.

P2: Complex systems cannot be steered and governed by regulation and rigid rules only.

P3: If regulation and rigid rules are insufficient for steering and governing complex systems, then there is a strong need for individual ethos and ethics.

Thus: Quantum AI systems require ethos and ethics from developers and users so that they are employed in a way which is desired by society.

Example

It has been estimated that insiders cause 20% of security breaches (DBIR: Data Breach Investigations Report 2008-2022). However, security budgets are primarily allocated to external threats (Cybercrime Is An Inside Job 2020); individual care and ethos tends to be neglected. This imbalance is reflective of the challenge of containing a budding technology such as quantum AI. Even the best (external) regulation will not be able to prevent individuals from crossing red lines if they lack the ethos that prevents them from doing so.

Recommendations

• Inculcate employees, be they managers, scientists, or developers, with awareness around the potential impact of quantum AI technology and their individual responsibilities.

• Anticipate developments and plan for appropriate responses, for example through scenario analysis and postmortem exercises.

• Achieve clarity around individual incentives and motivation – and address these proactively, particularly should they disincentivize individual ethos.

Background and relevance to quantum AI

Sustainability, as a normative concept and social goal for people (including future generations (MacAskill, Reference MacAskill2022)) to co-exist on Earth over a long time along typically the three dimensions of the environment, economy and society, has been a megatrend for some years if not decades.Footnote ⁴ Prominently, the United Nations agreed the Sustainable Development Goals (SDGs) in 2015, which set a global agenda for sustainable development, with a deadline of 2030.

For our purposes, the attention toward sustainability from the business practice is remarkable in this context. To give just three examples, this large and growing attention is documented by 1) a variety of clean-tech, social impact and food-tech startups nowadays that respond to the SDGs, 2) novel business and policy concepts such as carbon credits and 3) the sheer number of sustainability reports that have been voluntarily published by many companies throughout different sectors. This, in turn, has spun the wheel of a flourishing ecosystem through developments that include, for instance, metrics such as Bloomberg’s environmental and social governance (ESG) ratings and the Dow Jones Sustainability Indices (DSJI), which use private and public information from companies, and serve as proxies for actual sustainability performance. Another example is the formation and growth of organizations such as the Ellen MacArthur Foundation and the World Business Council for Sustainable Development.

To make a case for ethics in quantum AI markets, one can take inspiration from how sustainability has transitioned from an academic topic to a policymaker issue to a theme of interest for most companies, large or small. The idea is that one compares where quantum AI stands today with where sustainability was years/decades ago. One can then use this comparison for predictions of quantum AI development and adoption.

Presentation of the argument

P1: Sustainability has transitioned from a policymaker topic to a theme of interest for most companies.

P2: Sustainability and quantum AI ethics are sufficiently similar, for instance in respect to their cross-societal impact and close connections to the future of human wellbeing.

Thus: Quantum AI ethics will transition from a policymaker topic to a theme of interest for most companies.

Example

Sustainability has been argued to be indispensable to corporate strategy (Why sustainability is crucial for corporate strategy 2022) and to have a clear business case (Whelan and Fink, Reference Whelan and Fink2016). Moreover, early movers have already reaped rewards, an example being organic pioneers who enjoyed greater price premiums before the organic produce market significantly increased (Early movers in sustainability reap rewards 2020). Companies are increasingly considering sustainability and AI together now in order to achieve competitive advantages and further social objectives, cases in point being Microsoft and GE Research (The Opportunities at the Intersection of AI, Sustainability, and Project Management 2023). Quantum computing, in addition, is also being explored to help achieve climate goals (Cooper et al., Reference Cooper, Ernst, Kiewell and Pinner2022).

Moreover, sustainability and digital technologies are becoming more and more closely intertwined; key challenges no longer just relate to digital innovation but increasingly also to governance, bringing together “the green and the blue” (Floridi Reference Floridi2020). This intimate connection between the green and the blue is intricate and multilayered. For instance, while AI systems can make processes more efficient (which saves resources and cuts emissions), their development and training also increase carbon emissions; thus, a system of systems view is required (Gaur et al., Reference Gaur, Afaq, Kaur Arora and Khan2023). This includes quantum computing, which is increasingly being explored to improve the energy efficiency of computing, including AI applications (Chen, Reference Chen2023).

Recommendations

Learn from the emergence of sustainability and other transitions and avoid:

• Ignoring the topic until it is “too late”;

• Not proactively developing a perspective on how quantum AI can be used for good;

• Not reaping early-mover rewards, such as positive branding and novel business models.

Background and relevance to quantum AI

Good predictions, such as on the future of quantum AI, crop up in tandem with good explanations (Toulmin, Reference Toulmin1961). Often something more than mere prediction is desired. One needs to have information about the underlying mechanisms in order to make accurate and robust predictions about what will happen when the system such as a certain market breaks down or modifies itself in various ways in response to disrupting new technologies that are penetrating it. To embark on this arduous journey, humans in the business practice and elsewhere are often counterfactual learners, that is, they imagine worlds that do not exist and infer reasons for observed phenomena. “As-if” exploratory modes and imagination should not be frowned upon but rather treated as an integral substrate of progress (Hoffmann, Reference Hoffmann2022).

The sixth argument pays tribute to this insight by imagining a possible world where one, as a businessperson or company, decides not to consider ethical aspects of quantum AI. What would be the consequences of this decision? It is argued here that the consequences could easily be severe and unacceptable to good and reflected business leaders since a number of disastrous risks could materialize, such as:

• Losing reputation and damaging the brand;

• Steering into regulatory difficulties in the near future;

• Failing to attract young talents who wish to improve the world;

• Having one’s license to operate withdrawn by society.

Consider the fourth point: younger generations no longer choose employment based on a small set of factors such as salary and manager. They expect to work on cutting-edge problems with a strong positive impact. Quantum AI presents such opportunities par excellence because it is in the transition from lab to application (meaning many interesting research questions must still be addressed, and there are many white spaces to explore) and because it has tremendous transformative potential; quantum AI even raises fundamental questions about the nature of reality – for example, it has been argued that the mere existence of quantum computers provides evidence for a multiverse (Deutsch, Reference Deutsch1997; Tegmark, Reference Tegmark2003).

Given quantum AI’s complexity coupled with its revolutionary potential, “getting it wrong” could easily lead to operating a business that is not in line with society’s zeitgeist and thus considerable reputation and regulation problems. For instance, quantum computing has significant potential in military applications (Krelina, Reference Krelina2021) and, as with any technology, could also be used for nefarious goals. A prominent example here is in cybersecurity and data privacy where many current cryptographic standards are being threatened by future generations of quantum computers (“harvest now, decrypt later” attacks further highlight the importance of considering such issues already today).

Presentation of the argument

P1: Ethical aspects of quantum AI are irrelevant to businesspersons.

P2: If ethical aspects of quantum AI are irrelevant to businesspersons, then costs follow for businesspersons.

P3: Businesspersons aim to avoid costs.

Thus: Ethical aspects of quantum AI are not irrelevant to businesspersons. In other words, the risk is not to be in quantum; the risk is not to be in quantum.

Example

The Facebook-Cambridge Analytica data scandal, where personal data from millions of Facebook users was used via machine learning for political advertising that arguably had a significant influence on political outcomes, led to the greatest crisis for Facebook in its history and to Cambridge Analytica ceasing its operations (The Cambridge Analytica scandal changed the world – but it didn’t change Facebook 2019). It highlights the above risks and potential consequences for businesses who do not explore quantum AI ethics with sufficient care.

Recommendations

• Consider if the risks to your business of not exploring quantum AI compared with the risks of exploring quantum AI.

• Estimate the impact of certain scenarios (for instance, failure to attract quantum talent or failure to own intellectual property rights for a key quantum AI application) and use these to guide prioritization of quantum AI initiatives.

• Continually reassess – quantum AI technology is developing so quickly that plausible future scenarios today may already be replaced by new ones tomorrow.