I don't mind the AI features per se, but is there a configuration setting to sent the traffic through a local AI Gateway to prevent the AI from receiving private information? At the very least to track what is sent over the wire.
On thing that troubles me is that code reviews are also an educational moment for seniors teaching juniors as well as an opportunity for people who know a system to point out otherwise undocumented constraints of the system. If people slack on reviews with the agent it means these other externalities suffer.
Are being handling this at all? Is it no longer needed because it gets rolled into AGENTS.md?
I find working with Ai a lot like working with a junior employee... with the junior employee they learn and get better (skill level and at dealing with me) but with Ai the mentoring lessons reset once you type /clear
Skills are a positive development for task preferences, agents.md for high level context, but a lot of the time its just easier to do things the way your Ai wants.
You see, this is no longer necessary - companies are firing all the non-seniors, are not hiring any juniors, and delegating everything to AI. This is the future apparently!
> the AA-Omniscience Hallucination Rate Benchmark which puts 3.0 Pro among the higher hallucinating models. 3.1 seems to be a noticeable improvement though.
As sibling comment says, AA-Omniscience Hallucination Rate Benchmark puts Gemini 3.0 as the best performing aside from Gemini 3.1 preview.
> Stop citing single studies as definitive. They are not. Check if the ones you are reading or citing have been replicated.
And from the comments:
> From my experience in social science, including some experience in managment studies specifically, researchers regularly belief things – and will even give policy advice based on those beliefs – that have not even been seriously tested, or have straight up been refuted.
Sometimes people use fewer than one non replicatable studies. They invent studies and use that! An example is the "Harvard Goal Study" that is often trotted out at self-review time at companies. The supposed study suggests that people who write down their goals are more likely to achieve them than people who do not. However, Harvard itself cannot find such a study existing:
Definitely ignore single studies, no matter how prestigious the journal or numerous the citations.
Straight-up replications are rare, but if a finding is real, other PIs will partially replicate and build upon it, typically as a smaller step in a related study. (E.g., a new finding about memory comes out, my field is emotion, I might do a new study looking at how emotion and your memory finding interact.)
If the effect is replicable, it will end up used in other studies (subject to randomness and the file drawer effect, anyway). But if an effect is rarely mentioned in the literature afterwards...run far, FAR away, and don't base your research off it.
A good advisor will be able to warn you off lost causes like this.
Also look at Hedgehog Lisp. The bytecode compiler (runs on a PC) is separate from the interpreter, i.e. there is no REPL. But it means that the interpreter is only about 20KB of code. It's quite practical. It's not Scheme but rather is a functional Lisp (immutable data including AVL trees as the main lookup structure) and it is tail recursive. https://github.com/sbp/hedgehog
