š F,P! Week In Review, Briefly #6
Also: Key Up Wing and Down Wing items from the week that was
In case you missed it ...
šš¦ The Giant, Bioengineered Crab Bias theory of AI optimism (Monday)
⨠AI as a ānormal technologyā would still be amazing (Tuesday)
šļø Time to build ... new US cities (Thursday)
š¤ AI risks and rewards: My chat (+transcript) with AI researcher Miles Brundage (Friday)
⤓⤵ Up Wing/Down Wing: Farming with drones; Huntingtonās treatment; GLP-1 pills; tariff troubles; China-US space race; H-1b visa fees.
šš¦ The Giant, Bioengineered Crab Bias theory of AI optimism (September 22, 2025)
Just a Pinch. Maybe my AI optimism is irresponsible ⦠or maybe a little more risk tolerance is just what we need!
Introspection. Iām admittedly optimistic about the future and its technology. Progress usually brings net benefits, but never free of cost. Unfortunately, our culture tends to over-index on risks ā which brings me to what I jokingly call my āGiant, Bioengineered Crab Biasā named in honor of a phenomenal Onion clip. In it, cheerful anchors hype 75-foot acid-spitting crabs, engineered to aid construction projects, breezily dismissing any B-movie trope-style dangers. Could that willingness to wave away apocalyptic warnings describe me when it comes to AI?
In a Snap. Letās briefly take a look at my thinking:
AI might kill us. Some experts, like Anthropicās Dario Amodei, put catastrophic odds at 25 percent. To be fair, we canāt realistically put the odds of disaster at zero, so governments should fund safety research and think seriously about national security.
AI might help us. Goldman Sachs estimates AI could boost US growth by half a percentage point above baseline this decade ā and thatās without AGI. Potential upsides include scientific discovery, medical breakthroughs, and productivity gains.
Permissionless innovation should be the default. Barringn any extraordinary evidence to justify pre-market regulations. Letās take a āwait and seeā approach.
Market incentives count. Firms naturally avoid backlash, giving them reason to manage existential risks proactively.
Defensive AI is underappreciated. Governments should fund tools to counter malicious AI use and prepare for extreme scenarios. We can have good foresight without making rash decisions.
Up Wing Up Shot: Letās err on the side of hopefulness. Watchful waiting beats panic. Markets incentivize quality products and services. Build contingency plans, not bans. If that qualifies as Giant, Bioengineered Crab Bias, Iāll stand by it for the time being.
⨠AI as a ānormal technologyā would still be amazing (September 23, 2025)
Keep reading with a 7-day free trial
Subscribe to Faster, Please! to keep reading this post and get 7 days of free access to the full post archives.