The result of a new research study suggests that advanced AI models are better at acting dumber than they really are.
With a children-oriented theory of mind with similar characteristics, the researchers sought to determine if Large Language Models (LLMs) like OpenAI’s GPT-4 can pretend to be less capable than expected.
Based on 1,000 trials and cognitive tests, these ‘simulated child personas’ did seem to be advancing much the same as children do, verifying that LLMs can indeed pretend to be less intelligent.
Ultimately, these findings could aid the development of artificial superintelligence (ASI), the next step after human-level artificial general intelligence (AGI).
MORE