Heat exchange is used instead of refrigerating the coolant. Makes sense. How do they manage the indoor climate for the humans working there though? Eventually everything will be at 45C in the building, will it not?
Requesting human attention demands demonstrating human effort[1]. Slop at its worst: obscuring actual insight. The main one being that there are no exponential growth curves, no hockey-sticks, they're all logistic s-curves at most. Sure, there is a lot happening right now, but although technologies such as trains reshaped society, society did not come to only consist of railways.
The FFT is essentially just a matrix multiplication, or two. No need for fancy conversions. Just a huge amount of training data and a very large array.
The power needs of a house of course vary with many parameters. Our 100 square meter house in Nordic climate draws 1.2 kW on average over a year including charging the car. What are common numbers elsewhere?
This was of course dependent on yolo mode, but automatic approval has also been pulling stunts like this. A recent example is data that was purposely kept away from Codex in a folder far far away. When it found a single reference it just went for the data when having an issue. Lesson learned, keep essential data and Codex separated on different machines. Codex remote ssh actually helps here.
I'm starting to long for the age after AI. When the generative euphoria has settled and all outputs are formally verified based on exquisite architectures and standards.
I see your point. Many of my prompts for reasoning ends with: No code. Planning mode is sort of the workaround for this specific situation. Sometimes it is useful for the AI agent just to think. It looks like I need a screwdriver in addition to the aforementioned hammer, a pozidriv screwdriver to be precise.
To be fair, there is likely not much training data on the difficult conversations you need to handle in a senior position, pushback being one of them. The trouble for the agents is that it is post hoc, to explain themselves, rationalising rather than ”help me understand” beforehand.
I think that it is a fair perspective to allow role play, and it's useful too, when explicit. Does not really make sense for AI to cosplay human all the time though.
The other day Codex on Mac gained the ability to control the UI. Will it close itself if instructed though? Maybe test that and make a benchmark. Closebench.
Interesting. When I code, I want a boring tool that just does the work. A hammer. I think we agree on that the tool should complete the assignment reliably, without skipping parts or turning an entirely implementable task into a discussion though.
That is probably the next step, and in practice it is much of what sub-agents already provide: a kind of tabula rasa. Context is not always an advantage. Sometimes it becomes the problem.
In long editing sessions with multiple iterations, the context can accumulate stale information, and that actively hurts model performance. Compaction is one way to deal with that. It strips out material that should be re-read from disk instead of being carried forward.
A concrete example is iterative file editing with Codex. I rewrite parts of a file so they actually work and match the project’s style. Then Codex changes the code back to the version still sitting in its context. It does not stop to consider that, if an external edit was made, that edit is probably important.