Vatican AI Ethics Pledge Will Struggle To Be More Than PR Exercise – Forbes

The Vatican is seeking to encourage more tech companies to consider the ethical implications of ... [+] technology when designing and using AI systems. (Photo by Tiziana FABI / AFP) (Photo by TIZIANA FABI/AFP via Getty Images)

The Vatican cares about AI. Last week, it signed an ethical resolution on the use of artificial intelligence. Co-signed by IBM and Microsoft, this resolution lays down a number of principles for the development and deployment of AI-driven technology. It also commits the co-signatories to collaborate with the Roman Catholic Church in order to "promote 'algor-ethics', namely the ethical use of AI."

Superficially, the Vatican's resolution is timely and very well-intentioned. However, its unlikely to succeed in making AI more ethical, for a number of significant reasons.

Dubbed the Rome Call for AI Ethics, the resolution voluntarily commits signatories to uphold six principles when designing AI:

Given that artificial intelligence already has a bad rap for discriminating against women and ethnic minorities, the need to address its ethical implications is growing stronger by the day. As such, it's not surprising to hear the declaration's co-signatories herald its signing as a milestone in the development of artificial intelligence.

"Microsoft is proud to be a signatory of the Rome Call for AI Ethics, which is an important step in promoting a thoughtful, respectful, and inclusive conversation on the intersection of digital technology and humanity," said Microsoft President Brad Smith.

Likewise, IBM's VP John Kelly praised the initiative for focusing on the question of who will benefit from the proliferation of AI. "The Rome Call for AI Ethics reminds us that we have to choose carefully whom AI will benefit and we must make significant concurrent investments in people and skills. Society will have more trust in AI when people see it being built on a foundation of ethics, and that the companies behind AI are directly addressing questions of trust and responsibility."

There's no doubt that the AI and wider tech industry has serious problems involving the ethics of its activities. However, it's highly unlikely that the Vatican's AI initiative will make much of a difference in ensuring an ethical deployment of AI that benefits everyone, rather than just the corporations and governments that exploit AI for economic and political purposes.

First of all, despite talk of collaboration between the Church, academia, and tech companies, the Call for AI Ethics resolution outlines no practical, day-to-day strategy for working towards its wider aims. There's no practical timetable, no scheduled meetings, workshops, conferences, or projects, so it's hard to envisage how the laudable call for more ethical AI will actually be put into practice and implemented.

The Call for AI Ethics is intended more as an abstract incitement to AI companies to work towards ethical AI, rather than a concrete blueprint for how they might actually do this on the ground. This is suggested by Archbishop Vincenzo Paglia, the President of the Pontifical Academy for Life, who signed the Call on behalf of the Vatican.

"The Calls intention is to create a movement that will widen and involve other players: public institutions, NGOs, industries and groups to set a course for developing and using technologies derived from AI," he tells me. "From this point of view, we can say that the first signing of this call is not a culmination, but a starting point for a commitment that appears even more urgent and important than ever before."

Secondly, the six principles themselves are vaguely worded and open to considerable subjective interpretation. Moreover, anyone who's had any recent experience of each of the principles on their own will know that corporations and people conceive of them quite differently.

For example, "privacy" for a company like, say, Facebook is arguably not real privacy. Yes, Facebook can generally perform a reliable job of ensuring that other members of the public don't somehow get to view your Facebook posts and photos. Nonetheless, seeing as how pretty much everything you do on and off Facebook is monitored by Facebook itself, this isn't complete privacy. It's privacy from other people, not from companies.

Analogously, tech companies may in the future be great at ensuring that no cybercriminal hacks into the data their AI algorithms have mined from you. Still, the explosion in the use of AI to mine data will inevitably result in a concomitant explosion of personal data mined by tech corporations and sold off to other corporations. Again, privacy from people, not from companies.

Very similar points could be made about the other principles. In the case of transparency, "explainable AI" generally only works at certain levels of complexity, so that not every aspect of an AI system could be fully transparent and explainable. More fundamentally, tech companies may be able to explain the parameters they've set for their AI models, but not the wider business, commercial, social and even political ramifications these models could have once deployed.

On top of this, some of the principles are basically tautological, to the point of being almost meaningless. The third principle, that of "responsiblity," declares that "those who design and deploy the use of AI must proceed with responsibility." Put simply, to be ethical you have to be responsible. Very helpful indeed.

Then there's a deep misapprehension which undermines the substance of two of the other principles, "Impartiality" and "Inclusion." According to the Call for AI Ethics, impartiality dictates that AI developers should "not create or act according to bias." Well, perhaps developers can avoid being deliberately and maliciously biased, but bias is inevitable when designing any kind of AI. That's because developers have to select a certain data set when training their AI models, and they have to select certain factors or parameters that any algorithm will use to process said data. This entails a certain degree of bias. Always. Because an AI can't incorporate all possible data and all possible parameters.

In sum, the Vatican's AI principles are too insubstantial and fluffy. But more fatally, they also make the mistake of approaching the whole issue of AI ethics from back-to-front. That is, the problem that really needs to be addressed here is not AI ethics but, rather, the ethics of every company and organisation that seeks to develop and deploy AI, as well as the ethics of the economic and political system in which these companies and organisations operate. Because it's no good obsessing over the transparency and reliability of an AI system if it's going to be used by a company whose business model rests on exploiting workers, or by a military whose main job is killing people.

The Vatican recognises this aspect of the issue, even if the Call for AI Ethics doesn't explicitly address it. Archbishop Vincenzo Paglia tells me, "There is a political dimension to the production and use of artificial intelligence, which has to do with more than the expanding of its individual and purely functional benefits. In other words, it is not enough simply to trust in the moral sense of researchers and developers of devices and algorithms. There is a need to create intermediate social bodies that can incorporate and express the ethical sensibilities of users and educators."

Indeed, if organisations aren't really committed to being ethical in general, then no number of ethical AI initiatives is going to stop them from using AI in unethical ways. And in this respect it's interesting to note the lack of signatories to the Vatican's AI principles. So far, it would seem, the vast majority of the globe's corporations want to use AI for unethical purposes.

That said, Archbishop Paglia confirms that the Vatican is working towards attracting other corporations. "Certainly the work continues," he says. "There are contacts with other companies to create a wide convergence on the contents of the Call. For this we already have an appointment scheduled in exactly one year, for a verification of the work done."

But without a bigger body of signatories, without more detail on the six principles, and without addressing the underlying issues of social, economic and political ethics, the Vatican's Call for AI Ethics isn't likely to achieve much. At the moment, it seems like a glorified PR stunt, one way the Roman Catholic Church can appear relevant, and one way big tech powerhouses like IBM and Microsoft can appear ethical. But let's hope history proves such scepticism wrong.

Link:

Vatican AI Ethics Pledge Will Struggle To Be More Than PR Exercise - Forbes