Fascinating thing i find studying evolutionary psychology and evolution of AI is how often ideas cross pollinate. A human behavior can become an useful lens for algorithmic concept.
For e.g., contrarians are human XOR gates.
Whatever the crowd thinks, they think the opposite.
@TechieUltimatum The interesting part will be how it handles uncertainty. More personal health data could make its answers feel more authoritative, even when they shouldn’t.
In a lot of cases, token price doesn’t tell you much. What matters is the cost of getting the task completed. Here’s a useful comparison of coding-agent harnesses based on cost per completed task:
https://t.co/O7yHYpNrJ3
@sama Project brings Khan Academy to languages it doesn’t yet support, helping ~30M students learn in their own language. Starting with Finnish, Nepali, and Hausa.
Built with latest drops: Sites, 5.6 Sol. Tested with Codex Browser + Computer Use
https://t.co/4yJwuphgPF
@paulg I’m paraphrasing Daniel Kahneman here
“What would have to change for that to be right?”
It reframes disagreements into understanding the assumptions. Helps me a lot.
Congratulations to BAIR Faculty @matei_zaharia who has been awarded the 2025 ACM Prize in Computing "for his visionary development of distributed data systems and computing infrastructure, which has enabled large-scale machine learning, analytics, and AI at global scale."
https://t.co/suAIEms6vC
When iPhone launched, YouTube icon was a TV set. Notes a legal pad. Newsstand a bookshelf.
People weren't ready for digital. So Apple dressed the future in familiar clothes (skeuomorphism)
Next billion users for AI, the late adopters are waiting on this.
The loudest voices around new technology are always the extremes, hype or panic. The truth is usually quiet, boring, and somewhere in the middle.
So new tech adoption is mostly emotional, not rational.
Applies to PCs, smartphones, cloud computing and now AI. @rorysutherland
My reading hack: Find the author on YouTube first, then read the whole book in their voice.
Tried this on @Ada_Palmer’s Inventing the Renaissance.
#books#reading#hacks
Candidates share screen,debug live and explain thinking process of their approach. We can tell a lot by their thinking process and comfort of tool use. Early results of this approach is encouraging.
Do you forsee any downside to this approach or situations this won’t work?
I tried a different approach for technical interviews. Instead asking candidates to code to solve problems, i ask them to debug buggy code. Level of bug introduced in the code varies depending on seniority of the position.
1/2
I turned @paulg essays into 3 Claude skills for founders & VCs.
They help you: Think clearly, validate ideas & spot great founders
Links below.
Here’s how I built them:
Attempt #1: Failed
Just prompted Claude with links to PG’s essays.
Result are mediocre and generic.