They paved paradise (and put up a parking lot)
It was the combination of being in NYC and a references to asphalt that made me think about Joni Mitchell's Big Yellow Taxi during the conference closing key note by sociologist and Oxford Institute Senior Research Fellow Gina Neff. Professor Neff used the asphalt analogy to describe how our world is increasingly being paved by AI technology. While we may be talking about how to make better AI asphalt, we really haven't had an inclusive conversation about where or when to use it, who decides where it goes and what aspects of our world we may not want to pave.
She called for the community to expand our ethical capacity and bridge the gap between the technologists and social scientists. She also called for AI researchers (and industry) to go into communities first and find out what they want rather than just building and deploying the technology. It was the right message to close the jam-packed two day conference. As a researcher in the humanities, and a huge proponent of stakeholder engagement, it spoke to me on a very personal note about the need to broaden the conversation beyond the technologists and the role I can play in this work.
Should we turn everything into a math problem?
The day started with a series of great papers from savvy researchers presenting their version of how to make things better. Yet, the common thread seemed to be an approach of making better asphalt. I wondered about the efficacy of attempting to turn everything into a math problem and framing the solutions to ethical challenges strictly within this paradigm. Even if it can be done (debatable) should it be the way forward? It seems to me that it leads us down a rather narrow path.
Ethics is about appropriateness, not accuracy
Professor Peter Dabrock is a German theologian and Chair of the German Ethics Council. He regularly consults with big tech companies on ethical issues on a volunteer basis, so as not to be beholden to the companies (super ethical!). His talk centered around a new framework that's being launched by the German Ethics Council with respect to data sovereignty. The basic idea is to give users control of their data. They can determine how it should be used, they can change their data use settings at any time and for different circumstances. In short, they have full autonomy. Data agents and a data trustee works in the background to seamlessly administer the system in real-time. This is, according to Professor Dabrock, technically possible.
I have to wonder if it's politically possible. The current business models are built on people not having data sovereignty. It doesn't have to be that way but I suspect, at least in North America, it will take some big challengers and a lot of political will to change the system. Europe is far ahead of North America when it comes to ethical data policies and Germany may be the world leader in this area. Here is an article that provides more context about Germany's plans.
The problem with intelligence
Professor Stephen Cave, philosopher and Executive Director for the Leverhulme Centre for the Future of Intelligence, walked us through a genealogy of the word intelligence. Intelligence is a value-laden term, an evaluative term, that's been used to oppress and dominate over various people groups. It was used as a rational way to justify colonialism. IQ tests were the Social Credit Score of it's day, used to measure a person's worth.
He raised concerns about AI's reverence for the ideal of intelligence. One of the many interesting points he made was about how its primarily white, male, elites that seem to be concerned about AI having super intelligence (thus, having the right to assert itself because its more intelligent) and the resulting oppression. Others, he noted, might be less concerned about super intelligent AI oppressing humans because they are already being oppressed by other humans!
Professor Cave's talk came shortly after a paper discussing the two camps of AI ethics researchers - those calling to fix present issues with discriminatory AI and those focused on more futuristic, existential risk. The paper did not present who specifically was in each camp, but anecdotally, there seems to be more women and people of colour (Crawford, Whittaker, Sweeney) focused on present day issues and more white males (Bostrom, Kurzweil, Tegmark) focused on the long-term existential issues.
There were so many other incredible papers, researchers and ideas - too much to capture in a short blog post. Plus, it's Saturday night and I'm in NYC...
Big thanks to the conference co-chairs, students and other organizers for a great event!
Fun fact: It was Norbert Wiener that came up with many of the early ideas around AI and how it would impact society in his book The Human Use of Human Beings. He called it cybernetics. However, John McCarthy, who had a personal disdain for Wiener and also wanted to lay claim to the field, coined the term Artificial Intelligence and it stuck.