Recent AI model progress feels mostly like bullshit

Re:Recent AI model progress feels mostly like bullshit by adamiscool8 on Sunday April 06, 2025 @09:12PM (#43606465)

I don't think this reflects a flaw in LLM's at all. It is true that Paul Newman is not "known for" having alcohol problems. Paul Newman is known for acting, and salad dressing. Paul Newman has been "known to" have had alcohol problems, which formed as a question correctly one-shots accurate results for me across a few common LLMs. Someone like Betty Ford could be "known for" having alcohol problems. Paul Newman's alcohol problems are relatively specialized knowledge.

[ Parent | on HN ]

Re:Recent AI model progress feels mostly like bullshit by xdavidliu on Sunday April 06, 2025 @09:17PM (#43606505)

good point, though a great LLM answer would DWIM here and explain both interpretations as you have done here, as opposed to (what presumably happened) focus on one and ignore the other.

[ Parent | on HN ]

Re:Recent AI model progress feels mostly like bullshit by moregrist on Sunday April 06, 2025 @09:22PM (#43606548)

And yet the New Yorker article was the first thing that my Google search for “Paul Newman alcoholism” returned.

Apparently it isn’t so specialized that a pretty obvious old fashioned web search on Google wouldn’t immediately return an authoritative source.

[ Parent | on HN ]