Preliminary remark: This article was first published on medium.com on November 3, 2021.
On September 8–9 2021, ForHumanity, a public charity formed to examine and analyse the downside risks associated with the advance of AI, organized their first international conference. Practitioners, experts, regulators, and sector representatives discussed how comprehensive, pragmatic, and implementable audit rules and standards can be used in every corner of the economy to achieve this vision.
As a fellow of ForHumanity it was my pleasure to take part in the “Practitioner’s Corner” on Day 1, together with Merve Hickok (AIethicist.org), Krishna Gade (Fiddler), Navrina Singh (Credo), and William Llewelyn (Ethicsgrade). The session was moderated by ForHumanity’s founder and executive director, Ryan Carrier. This piece presents a summary of my key contributions to the lively and engaging discussion.
Question: What is your business and where does the demand for AI ethics come from?
Dorothea Baur: My background is in business ethics. I had an academic career focusing on corporate social and environmental responsibility but six years ago, I founded my own company because I wanted to have a bigger impact in the real world. My mission is to integrate the whole AI ethics debate into the overall ESG, CSR, or business ethics debate, because just like companies need to be yaccountable to their stakeholders for their impact on the environment, on society, etc, they also need to be accountable for what they’re doing in terms of using and developing technology.
However, the demand for AI ethics in Switzerland is not very strong yet. There is a saying: The Swiss always ‘wake up early but get up late’. This also applies to AI ethics: we’re still trying to get up when it comes to AI ethics. When I talk to people, it’s mostly about raising awareness, about telling them why is it important for their business, and why this is the case even though most of the AI ethics scandals we read about come from the US, the UK, or from China. I explain to them why it is important to jump on the bandwagon now and get ready.
Question: We want to explore where these conversations happen. Who reaches out to you usually? Is it CEOs? Chairmen? Or design and development teams?
Dorothea Baur: Usually, the first call is to hire me as a speaker or a lecturer on continuing education level, where I teach in front of people from the industry. But as I said, it’s very much awareness-raising.
We find it fascinating to read about Silicon Valley, but my challenge is to transfer its relevance to Switzerland. You always need to take into account the social and political and economic context.
We don’t have the same divisions in our society as the US has, and a lot of AI ethics scandals or problems revolve around e.g. issues of discrimination that don’t come in the same shape in Switzerland. It’s a different society, we have a different system of social welfare, we don’t have the kind of omnipresent credit rating system that the US has, etc. So, for me, the challenge is to explain to people in Switzerland why similar stuff could happen here, even though it won’t be the same along the same lines of division.
Anyway, I’m being called in as a speaker and for awareness-raising and then I can tell them what tools there are out there, and how you could integrate all those AI ethics issues into your overall sustainability or responsibility strategy.
But I’m mostly called in on C-level. I haven’t ever been called by design teams. After all, AI ethics can be thought of to be on a path of maturity, where Switzerland lags a bit behind. There is this typical path of maturity from the sustainability debate, where the first stage is defensive, where companies deny responsibility or refuse to acknowledge the problem. The second stage is “okay we acknowledge that there is some compliance element to it”. The third stage is when companies acknowledge the strategic relevance of an issue, and that’s when for example AI auditing could come into play, or maybe already a compliance stage. The fourth level is when companies start to claim “ethics is in our DNA”. At least this is what they claim about sustainability now, and I want to come to the point where this claim is being expressed about responsible AI under the condition that it is true. But in Switzerland we are in the ignorance stage yet, so we are lagging behind. But I’m working on it.
Question: What do you think is going to motivate the adoption of AI audits in businesses over the next three to five years? Is it cost-benefit? Is it driving revenue through trust? Is it reputational or regulatory? Is it about good ethical practices?
But I’m mostly called in on C-level. I haven’t ever been called by design teams. After all, AI ethics can be thought of to be on a path of maturity, where Switzerland lags a bit behind. There is this typical path of maturity from the sustainability debate, where the first stage is defensive, where companies deny responsibility or refuse to acknowledge the problem. The second stage is “okay we acknowledge that there is some compliance element to it”. The third stage is when companies acknowledge the strategic relevance of an issue, and that’s when for example AI auditing could come into play, or maybe already a compliance stage. The fourth level is when companies start to claim “ethics is in our DNA”. At least this is what they claim about sustainability now, and I want to come to the point where this claim is being expressed about responsible AI under the condition that it is true. But in Switzerland we are in the ignorance stage yet, so we are lagging behind. But I’m working on it.
Question: What do you think is going to motivate the adoption of AI audits in businesses over the next three to five years? Is it cost-benefit? Is it driving revenue through trust? Is it reputational or regulatory? Is it about good ethical practices?
Dorothea Baur: I think it’s going to be shareholder activism. We’ve seen that already a bit with Amazon where some shareholders protested against Rekognition, their facial recognition software. I hope that the same will happen with for example Palantir, who just relocated to Switzerland, which for me is proof that ethics awareness here is not that high. You can operate without being found out
Anyway, it’s going to be shareholder activism hopefully. That’s why merging AI ethics with ESG is so important. For shareholders, the key for deciding whether to engage on an issue or not is, whether the issue is material. Not material as in financial reporting but material as we know it in the sustainability debate, where you always ask which ESG issues have an impact on a company’s stakeholders and simultaneously have an impact on the company.
As soon as companies acknowledge the materiality of responsible AI, proper AI governance, privacy et cetera, this will give a huge boost to AI auditing, etc.
Context: In response to remarks about the importance of shareholder pressure, Merve Hickok emphasized the importance of values and culture, arguing that a company with a “move fast and break things” culture will not necessarily going to invest much into ethics and responsible design and will not incentivize their employees to act ethically. Navrina Singh then agreed with the importance of shareholder value arguing that even if culture is also important, unfortunately, “nobody’s thinking about that right now because there’s so much ambiguity in what is responsible AI, why do I need to do it, how am I going to make that happen right now”. ESG is now “top of mind for a lot of organizations and it is becoming a differentiation for companies that do it, but that took a while and it took a movement around what does shareholder value mean”.
Dorothea Baur: “I could subscribe to every single word Navrina said and in hindsight, I’m a bit ‘shocked’ to be the one who mentioned like shareholders first, when you asked who will drive the whole thing. I very much agree with Merve that culture plays a really important role. If you ever want to internalize ethics credibly, you need to have a fertile ground for ethics to fall on. I agree that a receptive corporate culture can lead companies to engage in AI ethics on a level of e.g. stakeholder engagement or with the help of an advisory board etc. But when it comes to a big commitment like AI auditing, this means you are ready to put yourself out there on being compliant or not it. For this to happen, it will take more than just an ethical corporate culture.
It will take a stronger business case and an agreement or pressure from shareholders because shareholders like this risk management element to AI audit. I’m afraid I need to admit that even Milton Friedman would agree with AI audits. He wouldn’t call AI audits a theft from shareholders. If he understood AI and AI audits he’d probably have to agree with it. This is certainly not decisive for my personal support of AI audits, but I’d have to say for once Milton and me would be on the same page…
It is kind of frustrating for an ethicist to admit that you need to demonstrate the business case and the return on investment of ethics. When I was in academia I would have refused this kind of discussion. But I have learned my lesson.
Question: Any last words from you?
Dorothea Baur: I want to slightly correct my scepticism about the readiness for independent AI audits, because actually for me as an ethics consultant AI audits will potentially solve two problems at once: first of all, they make it clear that AI ethics or business ethics, in general, is not just about philanthropy. Audits link ethics to a risk management approach. So they take away the suspicion that ethics is ‘soft’, and just ‘nice to have’ et cetera. The other thing that AI audits can help to address is that standard reaction to any company which is addressing AI ethics, namely the allegation that they are only doing ‘ethics washing’. So, independent audits have the potential to kill two birds with one stone. They make it clear that ethics is not just philanthropy and neither ethics washing. Why would you commit to this kind of rigorous audit if you wanted to do ethics watching?
For me, thorough independent audits offer a chance to get out of this vicious circle of ‘ethics washing’ allegations from one side, and ‘it’s only philanthropy’ allegations from the other side.
*********************************
Interested to know more? Get in touch!