An experiment run in Washington DC schools holds valuable lessons for the implementation of new technology
An important fundamental question is whether humans can ultimately control AI, which remains to be seen. Microsoft’s Tay Twitterbot demonstrated that AI can adopt the most deplorable forms of human expression, quickly. In less than 24 hours, that experiment was shut down. Similarly, a Facebook demonstrated that AI is capable of developing its own language, which may be nonsensical or even undecipherable by humans. So risks and rewards both need to be considered.
Many [business leaders] trust the technical experts but I would argue that they’re ultimately responsible if one of these models has unexpected results or causes harm to people’s lives in some way.
The fear of AI taking jobs manifests itself in the private-sector side, for sure. There’s an added element of that in the public sector, where people feel like if jobs are going to be at risk, we just shouldn’t do it.