I have two questions: how a user can inspect the original source when the AI’s derived memory changes? Is chronology/versioning becoming more important than semantic summaries?
I distinctly remember statements and discussions like that from way back when. Not just relating to google but around whatever the new search engine of the day was (however, I do agree that the sentiment around PageRank was mostly positive).
I would push for an audit of the whole mathematical corpus (all articles etc ingested by the present AI) as soon as possible. Since there are of the order of 10^7 articles, compared with the progression of results announced, this might be possible in less than a year.
All these results exploit the existing corpus. With enough time and maybe several orders more attention, we humans would discover these results.
So instead of producing heaps of new proofs (but be welcome to do this IMO), make an audit, OpenAI or Anthropic or anybody who listens.
Is this ironic or serious? As opposed to what? Typing it into a Chatbot? Speech input? I use commercial and self-hosted AI as well as (meta-)search engines, RAG, books, journals and whatever else I can get my hands on to search/learn. Neither is useless or old-fashioned and it’s great to be able to approach the goal from more and more angles that can complement each other. As for mode of interaction: For many things, I’m so much faster and more fluent when typing and reading that I don’t bother with speech.
Spring/Summer/Christmas "Fayres" are just a British thing. I'm not sure whether it's emerged as a spelling in recent years, or whether it's always been the case. But churches, village halls, schools, they've been doing it for ages.
It was actually the entire internet. It didn't matter what the search engine came back with, anything on the internet was suspect.
Then it was Wikipedia. Since anyone could update it, anything on Wikipedia was suspect.
Now even though Claude seems smarter than any person I interact with on a daily basis, it is all suspect.
Then there is the opposite with arxiv. Anything on arxiv is not just true but has authority in an argument, automatically. Everything besides papers on reproducibility. Those are basically ignored.
This is actually quite hard to achieve, at least in a one-shot LLM prompt.
You could engineer it by asking for a list of subjects sorted by obviousness, then pass that list to a script that discards the top 30% and randomly picks from the rest.
Even if you prompt an LLM to pick one at random, it usually won't, without an external randomness tool like that - which is curious, isn't it, when non-deterministic-ness is so inherent to LLMs.
I'm building a small tray/menu-bar app called Chiù that automatically prevents your computer from going to sleep while there's a detectable network activity, usually a download, e.g. a Steam game download/update.
I made it because I was tired of going into the system settings to disable automatic sleep manually whenever I wanted to leave a large download running. It's intentionally quite simple, but you can configure things like the activity threshold and grace period. There's also a manual keep-awake mode.
It runs on Windows and macOS. I don't currently have Apple/Microsoft publisher certificates because I'd first like to see whether anyone actually finds it useful before paying for those, so the OS may show a warning on first launch.
What worked best for me is to start doing things by hand, trying to do the architecture and the standards I really want. Then, AI works really well on following my conventions. Therefore it gets easier if I want to check codebase, because it's already organized as I initially planned, even all the new code is not anymore written by me directly. Then I also use Revix AI as a reviewer on my PRs that really helps me filtering anything that gets out of the standards and conventions I like.
Other than that, on other projects that I fully vibecoded, it's just impossible to keep with the codebase. Just blind trust the AI models on that, and never do it on a serious production product.
Evidence is how precise the findings are, and how often people act on them. More comments are not evidence.
Volume without action is just noise.
We tried multiple AI Code reviewers and LLM based review bots and they are all the same, a lot of volume with no real action after that.
Then decided to build Revix AI, where the main focus is basically to revert that. Prefer less comments but act on all of them even if we miss some things rather than get 20 comments that I will end up not reading and in the end get more issues into prod.
Agree that shared understanding is the hard part. Bugbots find bugs. Review is how the team stays aligned on the change. Visuals help a lot for that first pass.
Another cause of burnout for us was noisy bot comments on top of diffs that already look AI-written. We only want findings we would actually act on.
I checked your skill and looks really good, however I find it to be very complex already, we do many PRs and maybe visualizing graphs for every one is too much. I think ideally could filter the complexity of PRs and adapt the visualizer depending on that, being just two sentences on some and complete graphics on some.
I myself created Revix AI and we internally created some rules that we found that served educational purpose as well, we started linking that rules to the PR comments that Revix did so the devs could access that information if they wanted. Following what I said, it's always better to keep it simple and only expand if the human wants to get deeper or if the pull request requires it.
Not sure about the focus pop up but most/all of Apple’s other nags (like set up Siri, try Apple Music and the like) go away permanently if you go to the settings app and click “later” from there (as opposed to the button in the pop up which has the exact same label).
