Interview with Daron Acemoglu: the future of AI, data and sustainability

I do not see the technological and cultural transformation of society and the economy without an overhaul of our property rights; in particular, I must own my medical records or my digital agency, and be rewarded for it if artificial intelligence or other technology stacks will use that (my) data to learn and serve society at large with better products and services. Do you see this happening? Can Meta, Alphabet or Microsoft, for example, accept the fact that they will have to pay more stakeholders than today? Is this an illusion?
Absolutely. Data is a fundamental input for AI and at the moment there is neither a market for data nor ownership rights over data nor incentives for people to create high quality data. This situation is unfair (people's data is stolen) and unproductive (AI models are trained on low quality data, contributing to hallucinations and other problems with the available models). The solution must involve data markets and data property rights. But it is complicated, because individual property rights do not work. Billions of people produce valuable data, but transactions with billions of people would be prohibitively expensive. Also, many people produce highly substitutable data (everyone can recognise a cat, so platforms can pit one group of users against another to drive down the price of data). Therefore, we need a new infrastructure for data markets with collective ownership of data, for example, in the same way that the Writers Guild of America plays the role of an intermediary that protects the rights of creative artists and induces a type of collective ownership;

Do you think the financial and banking system will sooner or later accept cryptocurrencies, or even just bitcoin? What is your opinion on the future of the financial system, which should be open to all and where financial fees should become a commodity, i.e. zero?
I hope not. Right now, cryptocurrencies have value for three simple reasons: they allow illicit activities; there is a certain premium because if one becomes widely accepted, the company that runs it will become fabulously wealthy; and there is a bubble component to many cryptocurrencies, with gullible investors sometimes being encouraged to pay excessive valuations for certain currencies. None of these valuation sources are socially beneficial. In my opinion, cryptocurrencies might be useful in some very narrow circumstances, but they should be very heavily regulated. 

Do you think sustainability has had bad marketing so far? Shouldn't we rather talk about a total technological and marketing transformation of the world, in which the world is run more efficiently? Should sustainability advocates see technology as the only way to make this world less resource-intensive and less distorted, and stop talking about ESG as a separate item on our to-do list?
The ESG is a mixture of many different things, very poorly specified. Right now, it only gives managers the opportunity to do what they want. Investors who care about the social good must be an important component of the market economy of the 21st century. For example, I believe that investor pressure on fossil fuel companies could be very effective. But this requires a very clear set of guidelines with measurable parameters. In the case of carbon emissions, we can do that. So, my preference would be to move away from ESG and focus on carbon emissions and perhaps other important issues, such as labour practices (e.g. whether companies create highly unsafe environments for workers or exert coercive pressure on their employees).

The EU's AI Act seems to be the only legislation with any substance at the global level. Are we facing a world in which the rights of individuals will be sacrificed in the name of progress, if we look at the case of the US and China? Who sets the guidelines on what AI should do?
Yes, absolutely, that is the real danger and China is unlikely to play a leadership role in this area. The Chinese government is the biggest collector of data. The Chinese Communist Party is unlikely to be interested in protecting individual rights and the evolution of the technology industry in China over the past two decades has created a culture where intrusive data collection and surveillance has become normal. US regulators are still very much beholden to the whims and desires of the technology industry. So far, the leadership to reduce abusive practices in AI and the tech industry more generally has to come from Europe. This is not sustainable unless Europe itself becomes a major player in the field of AI, and it does not look like it will be easy. The hope is that the political pendulum in the United States swings in favour of more robust regulation and less influence of technology companies on policy and that, in the meantime, Europe starts to become an innovator, not just a regulator, of AI.

Are you optimistic or pessimistic about the future of the world?
Neither. It is possible to use AI in a pro-worker way. It is possible to use AI to support democracy and better communication. It is possible to use AI while respecting individual rights, privacy and individual autonomy. But no, we are not going in that direction. So if I believed that there is not going to be a major reorientation of technology and a major change in institutions in the US and elsewhere, I would be very pessimistic. But I cling to the hope that it is possible to course-correct, check out the tech giants, and start investing in artificial intelligence that is good for people (and in the meantime abandon the crazy dreams about general artificial intelligence and disabuse ourselves of continuing to trust the tech giants).