What Are We Going to Do About Artificial Intelligence

On Tuesday, 9 Feb I attended an Open Space event on this topic, hosted by Lloyd Davis and Helen Keegan at WeWork on the Southbank.

After some small group introductions to get started, we broke out into self-selecting streams based on discussion topics suggested by participants. I was motivated to propose one on, ‘When does AI get political and what do we do about it?’. So now I get to write it up!

We had some uncertainty about what constituted true AI as opposed to complex rules-based probability decision systems. At what point does a ‘ghost in the machine’ arise?

The participants were generally all agreed that AI was already politicised, but that the establishment in general was not keen for it to be seen as political; and that AI industries and business interests would probably actively work to stop it becoming part of the political scene.

We discussed how a recent study showed Google ads targeting to be socially biased — based on a name. If you sounded like an inner city dude then you got served more ads related to crime, for example. Google may say it’s just the algorithm and we can tweak it, but someone must be programming it to start with, and it’s definitely political!

At what point does politics and AI get to be a real test case? Maybe when the automated driver-less car has to make a decision between the life of its passenger or the life of a person outside the car? How is the law going to deal with that one?

How can we get involved? Is it a global level issue, and we need some trusted third parties to act as intermediaries (like NGO’s), or do we act at a micro level? We didn't get too far in this discussion. It’s clear society still lacks any kind of oversight framework in this space.

There are some interesting discussions happening in this area over at the Cybersalon.

Next Tuesday we’ll be having more Open Space conversation about ‘What do we do about the Internet of Things?’.