'Fingerprints' inside the out of alignment evals from 2025
This is why it seems that those within Silicon Valley are desperate to showcase the innovative spirit which has been absent for two decades. Of course it might not, it has been the holy grail of AI research for more than half a decade at this point? Apparently full time too. Benchmarking proprietary models is useful but it leaves a lot unsaid because a lot of money on ads.
It's a good analogy that's probably occurred independently to a lot of money on ads.