gpawley.bsky.social
@gpawley.bsky.social
League Spartan > Inter > *
November 21, 2025 at 7:47 AM
Looks nifty. And I can totally relate to writing tools as procrastination 😂
November 5, 2025 at 1:15 PM
Just the easy bit then 😅
August 1, 2025 at 1:28 PM
Interestingly there have been successful usages of LLMs in genetic algorithm like ways, i.e. FunSearch / AlphaEvolve.
August 1, 2025 at 1:27 PM
I generally agree that business drivers are not helping, that a lot of answers we find are in some sense cultural rather than rational, and that LLMs still seem limited in fundamental ways. I think I read "AI" more abstractly into your original post, rather than "LLMs".
August 1, 2025 at 1:27 PM
That's looking really good Martin
August 1, 2025 at 7:35 AM
I agree that progress has been the result of this diffuse distributed process as often, or more frequently, than it has been about the lone genius. That said, if it were possible to build a really smart Thing, it is probably possible to cheaply build a million really smart Things.
August 1, 2025 at 7:31 AM
someone at Unity thinks Mac users have poor eyesight
July 31, 2025 at 7:13 AM
Pretty clear! Looks good.
July 31, 2025 at 7:11 AM
I think I would like a diagram/visual of a city with each city feature described in a text box with a line to show the feature on the diagram.
July 31, 2025 at 7:08 AM
I feel like I have a good sense of what the game will be like - the first page is excellent, and the right place to start. I struggled with the second page. It's vital information, but there's a lot I need to remember and I can't integrate it until I see a more fleshed out / visual example.
July 30, 2025 at 6:21 AM
To clarify - do you mean you're seeing low rates of internal AI adoption at many companies?
July 3, 2025 at 7:50 AM
It's somehow brilliant that in a book that was so seminal in terms of predicting the future, the first line is a reference that was more or less out of date at the time the book was printed.
July 2, 2025 at 8:31 AM
It also validates the approach of most providers to focus on creating general agents. We'll basically have general agents participating in organizations as co-workers lite, before we have (if ever) fully automated task pipelines.
July 2, 2025 at 8:19 AM
I would suggest that this makes it easier to adopt AI, because it can seep in through the same chaotic messy process. And in fact, that is what the research is showing in terms of how adoption is happening, right?
July 2, 2025 at 8:18 AM