246 - Nonsensical Questions
Several of the most frequent nonsensical questions I’ve encountered in the tech and AI space are “If that is true, then why haven’t I heard of you yet?” and “Why aren’t you funded?”. Of course, these are the same question. The fundamental question that they both re-skin is:
“Why aren’t my personal default assumptions about the world painting a coherent narrative about you?”
The answer is inevitably that cognitive biases, while useful tools, are also wrong. When they fail, they fail systematically, like LLMs, which is about the only human thing that LLMs are like. When humans stick to assumptions that systematically fail then they become less effectively intelligent than a literal random number generator, or as Prof. Tetlock colorfully put it, a “dart-throwing-chimpanzee”.
These biases are also often circular, self-fulfilling, self-reinforcing, and ego-centric post-hoc rationalizations. “If I haven’t heard of you then you must not be that (good/important/etc.)”, “If other people aren’t investing in you (yet) then I won’t”, or as was comically the case recently “If other people haven’t benchmarked your systems already then we won’t.”
Frequently the sub-chimpanzee-performance human paints the most cognitively lazy narrative that offers them a sense of coherence, “You must be lying (because if you aren’t then I’d have to use my brain and revise my worldview, and who has time to do that?)”. Fortunately for those who they target, these people usually offer no non-trivial value, whether they are successfully educated thereafter or not, so the effort required to correct them can be put to better use.
Cognitive biases deliver a simplified world, where attending Harvard is a sign of prestige or competence, not of the far more likely statistical base rate of the financial incompetence required to pay full tuition at such an institution in this day and age. Such biases may, at one point, have held an actual correlation (not a causal relationship), but even that they can easily age out of without being let go of by society’s collective cultural memory.
If you haven’t heard of us it is because people in the media share the same cognitive biases, and dart-throwing chimpanzees can outcompete them too on outliers where those biases systematically fail. Investors share them, plus a few of their own. Some cognitive biases are culturally reinforced, and some are cultivated in specific domains.
Cognitive biases make the world go round, frauds like Altman weaponize them, and many never see beyond them. Humans can’t perform at a “human level” without them, but they are useful until they aren’t. Recognizing when they aren’t is a skill that more people need to develop. They’d ask much smarter questions if they did.
