iamtrask Says AI Uses Vague Language to Avoid Mistakes

iamtrask · x · 2026-07-07

Andrew Trask (iamtrask) shares observations on AI handling professional tasks: to avoid hallucinations, AI adjusts responses to a "perfectly vague" level to avoid errors, producing much empty content. He believes this cannot be solved by alignment but requires human taste, trust, and scarce high-quality data.

Related event: iamtrask: Human-AI Interface is the Frontier of Value Creation(2 posts)→

Original post →

More from Models

Models channel →