AI's hidden costs: Privacy, copyright, and identity
Video and transcript
As part of our inspiring women in AI series of interviews, Dr. Gry Hasselbalch, author of Human Power: Seven Traits for the Politics of the AI Machine Age, expresses concern about mainstream AI systems' ethical issues and potential threats to human identity.
She also highlights more positive AI developments, particularly those from minority communities and projects focused on health and sustainability.
Transcript and audio description
Continuous shot of Gry face-on in an office environment.
I have to be honest and say that I'm more worried than I'm excited at the moment.
Of course, I do see some, I think that's mostly projects that come from minority communities or from ecosystems that are not part of that global AI race.
Because the biggest AI systems, the ones that we hear most about at the moment, have some huge flaws. And I think it's not only in terms of the here and now – like how are they trained and were they trained on copyrighted material or do they invade our privacy? Are they censored?
I find that really problematic, but it is, as I said before, the kind of the values that they've been assigned with. So the more systemic impacts on both on our democracies, but also, our human identity. They're somehow designed with this idea that they can basically solve all human problems and replace a very flawed human identity.
There's not much confidence in what humans can do and who we are as a species. And I find that very problematic. So I would say I'm more worried than excited at the moment regarding the dominant AI field.
Then of course there is what we don't hear much about... There are breakthroughs and there are, in other ecosystems than the most dominant ones at the moment, developments in terms of AI for health, AI to support the Sustainable Development Goals. Things like that.
And those are the areas where I would say I would put most of my faith in, but that's not what we hear about in the public discourse.