Google researchers may have just turned the race to AGI upside down with a single paper
Business Insider
|
Contributed by: Bill Russell
Summary
Google researchers reveal transformers, tech driving large language models (LLMs) like ChatGPT, struggle with tasks outside their training data. The findings, a setback for achieving artificial general intelligence (AGI), highlight transformers' limitations in generalization and simple extrapolation tasks. Despite the blow to AGI ambitions, advanced AI forms may have better generalization.
