Four Prevalent Themes in the HN Discussion
-
Medicine pragmatically accepts treatments without full mechanistic understanding
Many argued that effective drugs are routinely used despite unknown mechanisms (e.g., anesthesia, Tylenol), and efficacy—not mechanistic proof—is the standard for approval."We don't even have the mechanisms behind general anesthesia nailed down; do you want to forgo it when the docs cut you open to remove your tumors?" — squidmaster
"Indeed, aren't medical trials based on its effects, rather than how it works?" — gste
"Logic would dictate that medicine should be pragmatic and accept working cures that aren't understood." — qup -
Concerns about AI exploiting weaknesses in clinical trial or validation processes
Tao's core argument centered on whether an AI could "game" trial systems to produce false positives, requiring scrutiny of alignment beyond surface-level efficacy."Could the AI solution be somehow misaligned by exploiting a weakness in the trial process or its math models?" — jsenn (calling this "the crux of his argument")
"If we saw AIs reward-hacking clinical trials to get drugs passed, that would be an extraordinary outcome..." — dekhn
"Do you think the current drug approval process is bullet proof enough that a completely novel AI-generated drug candidate... can be considered safe if it makes it through the process?" — jsenn -
Tao is perceived as out of touch with how ordinary people actually make medical decisions
Critics contended that most people lack deep scientific understanding yet rely on trusted experts and empirical results—not personal comprehension—when accepting treatments."This guy might have the highest IQ on the planet, but it's clear he hasn't spent much time around average people." — qup
"Half of people are below average! They understand nothing at the level that Tao means. Literally nothing." — qup
"People were not against a COVID vaccine... They were against a rushed vaccine. They were against a vaccine without human trials." — qup (countering Tao's implied skepticism) -
AI is disrupting mathematical culture, identity, and the value of traditional proof-based work
Many expressed grief over AI shifting mathematics from an understanding-driven discipline to an outcome-oriented one, undermining the intrinsic joy and communal aspects of proof-solving."Mathematicians are primarily understanding-oriented. Leveraging AI to tackle new frontiers without true understanding converts mathematicians to engineers." — hansvm
"Several reported OpenAI drop included answers put a wrap on problems they had been working for years... like losing an old friend or a lover." — samuelknight
"The field needs to get it through their heads that their old problems are no longer ambitious..." — samuelknight (on the need for new frontiers)