“`html
Are We Training AI Too Much Like Humans? A Sociologist’s Perspective
Table of Contents
A University of Chicago sociologist and data scientist, argues that training AI to think like humans is hindering scientific progress.
According to James Evans, a sociologist and data scientist at the University of Chicago, current artificial intelligence models may be limiting scientific exploration because they are being trained to think too much like humans. In a recent episode of the Big Brains podcast, Evans discussed how these models could be narrowing the scope of revelation.
Evans is advocating for the development of AIs that think “differently from us”-what he terms “cognitive aliens.” The podcast explores the potential of these “alien minds” to unlock breakthroughs and the steps required to create them.
Could artificial intelligence with a non-human cognitive framework lead to unexpected advancements? The podcast delves into this question and the challenges of building such systems.
For a deeper dive, you can read the transcript of this episode.
The Risks of Human-Like AI
The push for AI that mirrors human thought processes may inadvertently restrict the potential for AI to uncover novel solutions and insights. By encouraging AI to think differently, researchers hope to overcome these limitations and unlock new possibilities in various fields.
“We’re training AI to think too much like humans-and it’s holding science back.”
Exploring “Cognitive Aliens”
The concept of “cognitive aliens” involves creating AI systems with fundamentally different cognitive architectures than those found in humans.This approach aims to foster creativity and innovation by allowing AI to explore unconventional solutions that might be overlooked by human-centric AI models.
