Back then, people who were bad at interviews were often undervalued. I think interview culture itself will soon be devalued and eventually become obsolete. People who build real technology and solve real problems will be unaffected, while the glorified Valley slop built around interview performance will become obsolete.

My optimism comes from AI's ability to solve genuinely hard mathematical problems this year. September and October could go down as two of the most remarkable months in the history of AI If, and only if, those proofs are correct and independently verified by real experts. Even if the proofs have been formally verified in Lean, expert scrutiny still matters.