Google AI Mode is pretty much doom scrolling. Its frontier in its own way.
the generated doc is largely coherent. but you feel the blurriness as you read through it. you also sense that what you need is in there. but you wont use it because it will only make things worse. alas, you have no option but to generate it again. Output Blur
its time your coding harness guardrails start behaving more like a CFO. they have the tech locked down.
generating slop is fine. spin the wheel, shape, paint, and bake it too.
There must be some dimension along which progress since GPT‑3.5 has been roughly linear. I wonder which dimension that is.
can someone build me a social media style feed out of what my agents did or are doing right now. auto generated memes will make it even better!
if the 5.6 benchmarks are to be believed, i do not see how anthropic maintains corporate buy in for much longer without significantly cutting token costs
are saas folks who have been left on the sidelines until now fighting back with swarms?
the shift from “understand how to do it and then do it” to “finish doing it and then understand how it was done” is truly crazy
Governing AI is a choice about where freedom gets constrained. Unrestricted access to powerful models means more surveillance to manage misuse. A private, free society means more limits on access to powerful models.
a prompt is not a requirement, a plan is not permission, a test is not acceptance, code is not intent
looks like model version numbers are definitely in the S portion of the curve leading indicator?
what year will we stop writing “this is agentic coding please do everything in one go” at the end of every prompt
wish agents just gave multiple answers instead of arguing about everything in a single answer and i can just branch the chat at the one i like
information architecture should naturally facilitate thinking depth selection and give agents the ability to stop correctly at the right spot
Less pedantry about expression does not translate into greater precision about meaning.