I suspect we are about to enter an interim period where AI exceeds human performance on many cognitive tasks, but this is not common knowledge, and so most people and institutions act like this is not generally the case.
This may well already be true of self-driving cars. I think it is going to be true for large swaths of academia, government and industry, but even blind testing won't persuade -- many people will insist that nothing generalizes beyond the very specific context studied.
People will also point to instances of older models making wrong, unusual or unpopular suggestions in one case as if this justifies ignoring the models in all other cases, even though in any systematic evaluation, the models outperform humans.
The change will be very keenly felt among less established people. Among people just entering the workforce, less capable peers will be producing equally valuable work and it will be obvious why.
Grad students will largely rely on AI models both to produce work and to decide what to study. Many businesses will essentially be proxy competition between AIs, with humans having a comparative advantage in networking and providing capital.
The established will still think that their judgment is unmatched. Sure, the AI system can write down models and analyze data, but we can figure out what questions should be asked. Sure, the AI system can develop software, but we know what software people will value.
This will be false. AI systems will excel at decision-making given the available information, and will outperform the CEOs and tenured professors and elected officials. But people in power won't acknowledge this.
If AI is controllable (obviously a big IF), this interim period where people in power pretend humans add value in cases where they do not could last for decades, perhaps until the generation that grew up with AI is in control.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
