In our new CONVOCO! Podcast Corinne M. Flick speaks with Carl B. Frey, Professor of AI and Work and Director of the Future of Work Program at Oxford University, about:
Will AI surpass human intelligence?
Here’s what he said:
We will increasingly see tools able to replicate in the virtual domain what one may call human social intelligence. But I strongly suspect that this is only going to put a premium on in-person communication.
There’s a risk that the internet will deteriorate as a potential source of training for AI, simply because it’s going to be flooded by AI-generated content.
A general theme I worry about with the computer revolution is that as it allows us to do things more productively, we seem to opt for quantity over quality, for doing more things rather than fewer things with greater care.
There’s been a lot of hype around ChatGPT surpassing human performance in various tests. But we’re constantly seeing that these technologies don’t perform as well when new circumstances emerge.
Generative AI seems to help low skilled workers and novices more. That has given some people reason for optimism that AI might reduce wage inequality going forward.
Somewhat ironically, the more we emphasize the dangers of AI and try to control it, the more we’re pushing the monopolization of that technology.
It’s important to remember that generative AI is not all of AI. There are many use cases in transportation, logistics, or warehouses, where it’s early days of the revolution.
We should be better at devoting our computational resources to harder problems. Right now, it doesn’t matter what you ask generative AI to do. It devotes the same computational resources to basically every question