I think loops serve a valuable purpose when the work needs to be repeatable and there’s a certain measure of precision involved. I do think, for more abstract problems, free thinking and open question prompting can really explore possibilities not otherwise thought of.
Context rot is always going to be a challenge and it evolves over time. It’s interesting to see the dumb harness idea, i do believe in taking large concepts and workshopping them at a smaller level to prototype and iterate quickly.
I do something similar. I don’t necessarily call it AGENT or STATE and every project has its own files. I have architecture documents that accompany change log descriptions that load technical knowledge that the agent can readily use.
I find it also necessary to have a principles document outlining the particular problems that the software is supposed to solve and guard rails to not cross. I call it promise driven development.
I have a running theory that although AI will continue on (just as the internet did). That between the rapidly accelerating technology and methodologies as well as the thriving open source community. That no company will be able to extract meaningful value that they can return to share holders, and that the margins will remain low on the back of high energy costs.
I don’t think this situation is going away and is likely going to accelerate. As systems get more complex, it becomes increasingly difficult to keep track of the changes and impacts to a large code base (especially if you are working on a monolith). Increasingly it becomes a cat and mouse game of tracking race conditions and edge cases that might be the subtile degradation of multiple components.
Automation methodologies like IaC have helped and some advances in AI support tools have helped to parse data faster. New problems have emerged as well and I would say case complexity has also increased.
I would not say it’s capitalism but rather industrialization. Children are assets when they can help with farm work but become an expense when work moves to a factory/office setting.
I wonder what kind of bone yard is going to come out of all the mania spending. I am picturing a sea of GPU's being liquidated from startups no one's ever heard of.
I think anything to reduce the overall power and consumption of the AI is the long term win. I do see nature as being a catalyst for developing new technologies. I don't think WallStreet has the patience for this development and research. I wonder even if there can be value extraction from this, or if this will eventually be supplanted by the next hot idea that sounds good at the next shareholder meeting.
I don't mind having a phone call, it's the multiple layers of screening interviews with ill prepared questions that feels like a giant waste of time. I'd rather answer all those questions in an email and have the phone call as a formality. I do recognize the need to confirm that the person writing is indeed the person you are speaking to but I should at least be able to get a straight answer on salary and comp before I pick up the phone.
Stories like these really help highlight how bad the bubble some people surround themselves with. Imagine being so out of touch you describe another human being as "lower value human capital".
I have been seeing so many bad commencement speeches that it's good to see someone actually deliver a nuanced and grounded approach that speaks to the issues of our time. Woz has definitely improved a lot in his public speaking over the years.
Everyone was hating on the Google CEO but I really almost had a crash out of how out of touch Scott Borchetta sounded on stage too. Glad there's one good Apple out there.
Context rot is always going to be a challenge and it evolves over time. It’s interesting to see the dumb harness idea, i do believe in taking large concepts and workshopping them at a smaller level to prototype and iterate quickly.