An exchange, following my appearance at Applied Machine Learning Days. Blogged by permission.
Dear Victor,Dear Professor Bryson,My name is Victor Kristof and I'm a PhD student in Machine Learning at EPFL. I saw you yesterday on stage at the Applied Machine Learning Days.I wanted to thank you for the very interesting talk and the extremely sharp inputs during the panel discussion.Particularly, I highly resonated with you saying that the two problems of our time are sustainability and inequality. I also resonated with you mentioning the fact that too few researchers and engineers are working on these big problems.I had a question that I unfortunately could not ask due to the limited time: According to you, what incentives should we—as a society and as a community—provide to AI/ML researchers and engineers to tackle these issues, since apparently saving the human race is not enough (or at least, the moral motivation seems not to be sufficient)?Pardon the cold reach and sorry to take your time, but I believe you’ll have great insights.Sincerely,Victor Kristof.
I think the answer depends on who you mean by "we"? You say "as a society", but societies don't really make executive decisions collectively, at least not intentionally. As with every political decision, every individual has to decide how to divide their time in attempting to have efficacy. Generally (but not always in specific) where you have the highest probability of impact you have the lowest chance of success, e.g. motivating your household vs your country, so there's an infinite number of equally valid ways to divide your attention. However, even here every individual will be different.
I believe in general, unless you are working in a computer science department or tech company, it's not about motivating ML people in general but everyone to have attention to these issues, and then if you find a potential source of funding or other resources (after school clubs, whatever) you can suggest ML as one way to have impact.
I know one specific thing in the USA which is that the environmental protection agency (EPA) is specifically asking for ML people to help them visualise the pollution and sustainability challenges of the present age. This is because they held a similar competition in the 1970s with photographs, and now can show just astonishing improvement that their regulation brought about. But now there is nothing obviously visible left, so they need help from those good at visualising data.
I'm sure there are other similar initiatives that can be brought up by universities or cities. My city of Bath has a digital community called "Bath Hacked" that uses the city's data to try to help its citizens, including showing things like pollution graphs and visualisations.
Of course larger concerns like the population pressure on biodiversity aren't easily handled on the city level, but a city or school may still start an initiative that somehow communicates the information more effectively.
One reason that people don't act on things like global warming is that they believe science will just "fix" them, and of course we might! Working on ways to say increase plant growth or presence in urban environments might also be a positive action. But ultimately given the physical size of humans there can only be so many of us on the surface area of the planet, however small our houses are. So there are all kinds of social issues that emerge from rethinking the economy etc. when population growth wanes and reverses, which is something parts of Europe already lead in.
So there are things from practical biochemical / engineering / power responses, to education, to tools for governance, to simply making perception more effective of the impacts of individual lives. And or course lobbying individuals to choose these things to do with their free time or their public or private funding or even their start ups.
I have to say I was very impressed by all the energy at EPFL. It shouldn't be discouraging that no one person can do everything, because there are so many people who want important things to do!