fastThinking

7 ms·

1. 1.▲
   by fastThinking 7mo ago

   I think you’re missing the point a bit. Not every tool needs to be Figma, and honestly, that’s a good thing. I’ve been using Figma for a while, and true, it’s powerful. At the same time it becomes increasingly complex, difficult, bloated ov
2. 2.▲
   by fastThinking 8mo ago

   The writing is beautiful, also accountable, but doesn’t the moral agency here belong to the company rather than the model?
3. 3.▲
   by fastThinking 8mo ago

   If well-formed sentences now read as LLM output, that’s unfortunate. If you disagree with the point(s) I’m happy to discuss that instead.
4. 4.▲
   by fastThinking 8mo ago

   No LLM here, just a habit from academic writing. The reason the tobacco analogy works for me is that both cases optimize around reinforcement, not outcomes.
5. 5.▲
   by fastThinking 8mo ago

   Good one
6. 6.▲
   by fastThinking 8mo ago

   This reads less like nutrition science and more like addiction engineering. The tobacco analogy isn’t rhetorical, it’s structural.
7. 7.▲
   by fastThinking 8mo ago

   Being ahead of Google is less about raw model quality and more about shipping usable products fast. Anthropic’s advantage seems organizational as much as technical. If Sonnet 5 really halves inference cost while improving reasoning, that’s
8. 8.▲
   by fastThinking 8mo ago

   As a parent without nearby family support, this feels uncomfortably real. What the model exposes is a system where illness isn’t a failure but an incentive, quietly shifting the cost of optimization onto families.
9. 9.▲
   by fastThinking 8mo ago

   So Copilot is for customers, Claude is for getting actual work done?
10. 10.▲
    by fastThinking 8mo ago

    So Copilot is for customers, Claude is for getting actual work done?
11. 11.▲
    by fastThinking 8mo ago

    This feels like a glimpse into a future where ‘organ failure’ is a solvable engineering problem.