[rvz]

That's if you trust and believe that the LLMs themselves are 'correctly' scoring.

I wouldn't immediately even agree with an assessment made by these LLMs if I were Gary Marcus as that could immediately contradict any of the claims he even made and falling into the trustworthy trap. I'd remain skeptical as ever...

...because this is the worst of the red flags that ultimately supports Gary's argument that the LLM results may be untrustworthy:

All verdicts are LLM-scored, not human-verified.

People should check for themselves and draw their own conclusions.

> The crash hasn't come.

yet.