Evgeny Morozov: ”What kind of Artificial Intelligence do we want?”

Artificial intelligence is the next big technological step for humankind. But what happens to our society when machines have more and more influence on our lives? And how do we regulate this process? Evgeny Morozov, AI expert and author of "The Net Delusion: The Dark Side of Internet Freedom", answers our questions in this interview.
13 May, 2022 by
Evgeny Morozov: ”What kind of Artificial Intelligence do we want?”
GDI Gottlieb Duttweiler Institute

GDI: The Internet: The holy grail of democracy, freedom of speech, and autonomy. A hope that has not come true. Will AI now be the game-changer?

Evgeny Morozov: None of the emancipatory promises made in the name of digital – and many other – technologies would become true, unless we work to make them true. There are so many different ways to design technologies – but also to shape the context in which they are used, by whom, and with what purpose. Obviously, there’s much harm that can happen due to AI. But there are also many gains – once again, as with most technologies.

The real challenge is to understand what kind of AI we want – and how we need to change society to make it more hospitable to that kind of AI. Of course, we won’t have a one-size-fits-all solution: we’ll all have to make compromises and it’s precisely in the making of that compromise – between citizens and governments, companies and consumers, schools and students – that the most important debates about AI will unfold.

In science fiction, the higher intelligence of machines is mostly misused for exploitation and oppression. Are there examples of democratic, constructive AI applications?

It depends on what we expect from AI and how we imagine it to be. I like to think of it more as “intelligence amplification” – that is, relying on technology to make us, the human decision-makers, smarter. It’s not a matter of automating us or of replacing us entirely. In this sense, yes, there are plenty of constructive AI applications – any technology that allows us to make better decisions by spotting patterns and trends that a human mind can’t spot. Even the abacus would be part of this great arsenal of technologies of intelligence amplification. How to make all this stuff democratic is another issue, but here we have to distinguish between various meanings of “democracy.” Do we want to make access to such technologies wider? That’s one meaning of democracy. Or do people want to participate in the design and refinement of such technologies? That’s quite a different meaning. The former is probably easier to achieve than the latter but we shouldn’t entirely abandon this second, more utopian option.

Do we need more or fewer regulations for AI developers?

Well, we need a broader framework, a strategy, and a goal of what we want to do with all these tools for amplifying intelligence. Are we amplifying the intelligence of big companies only? Do we want to spread all this data wealth a bit more equally among the population? Regulations should follow a strategy but without a strategy they would not be very useful. We are missing this bigger, more strategic framework for thinking through these issues.

Who benefits if machines learn faster and faster?

In most cases, of course, it’s those who own the machines and who control their inputs and outputs. Again, I don’t think that there’s much controversy about wanting to have more ability to delegate simple and non-controversial decisions to machines. But what kind of decisions and in what levels of society, and with what consequences for the weak and the unprotected: this is the open question we need to resolve.

Knowledge and speed are power: Will AI change the balance of power in society?

I hope it does! Otherwise, it means that our society has become so insensitive and resistant to change that even vast changes in its technological infrastructure leave no impact. I do think that given the many unresolved problems, from inequality to climate change, technology, concentrated in the hands of the few, might make many of our current dilemmas much worse. But this is no reason to pretend that we can go back to a pre-technological universe.

We still have to articulate – and then implement – a vision for making all these technologies work for a very different political and economic project, the one that would address problems like climate change and inequality – and would hopefully do so without sacrificing the tools of technology and science.

Evgeny Morozov will be speaking at the conference Discovery on Steroids: How AI will Speed up Innovation on 5 July 2022 at the Gottlieb Duttweiler Institute. Sign up now!

Share this post