quinndupont

Jev seems to be the target, but without saying so. Jev is trained differently from typical LLMs to avoid these issues.

Agreed, but don’t subscribe to z.ai; their inference is terrible. Stick with openrouter

Just switch from a programmer to a software developer. Don’t compete with AI, use it.

I could do without more surveillance.

We have adopted marimo notebook for doing modern data science. I’m happy with their progress. Nice to see greater integration.

Looks great, but the pricing model is a challenge for a small team. If there was a generous trial period (for testing) with a lower small team price I might be able to consider it, but it’s hard to justify constantly increasing AI and compute costs (already paying for several model subscriptions, existing orchestration tools… in addition to typical infrastructure)

A global labour crisis and broken reputation signaling in FOSS has produced a deluge of software production that has basically nothing to do with low quality submissions and everything to do with capitalism.

FOSS has become the release valve for too much labour supply.

I’m sure this will get downvoted by the Platonists (ie all contemporary mathematicians), but it’s interesting to see mathematicians slowly realizing that they are dealing with a generative topic, like all human writing, and not direct access to God’s asshole. Soon they will realize the same crisis as everyone else and then eventually they will realize that writing (math) is a human invention requiring humans to make meaning.

Syntactical (grammatical) text is signal but maybe not the kind of meaningful signal we all wish it was. Meaning, of course, resides in the reader, in the practice of reading (Wittgenstein etc), not the author. We used to use writing quality as a proxy for quality of meaning (thought), but that no longer works well.

The rise of AI writing has only been matched by superficial articles comprised of ideas salad that evince no deep theoretical or historical understanding. Crappy writing has and always will exist, AI doesn’t change that, it just makes awful writing grammatical.

Clergy worried about digital Luther, police AI. Old story, told anew!

Looks great! Complimentary to my own little project that takes care of the application process: https://jobapplicator.win/

30 seconds and you have all your materials ready for submitting.

Looks great! Complimentary to my own little project that takes care of the application process: https://jobapplicator.win/

30 seconds and you have all your materials ready for submitting.

Very helpful analysis that confirms everything I’ve encountered. OCR remains a thorny issue. The author talks about professional workflows struggling with tables and such, but I’ve found it challenging to get clean copies of long documents (books). The hybrid workflow (layout then OCR) sounds promising.

Ubuntu is the issue.

Desperate job seekers. Nobody wants LinkedIn.

I have a similar idea for a little Potemkin village that AI agents can hang out in, do work, relax, etc. I think we will see more of this. Integrating machine to machine payment is a requirement.

Imagine you are poor and a rich person offers you a choice to steal some bread or some beer. It’s not a real choice because you are poor and therefore steal. The rich person offering the choice is wrong.

It’s an unethical, false choice. The reviewers are not perfectly rational agents that do free work, they have real needs and desires. Shame on ICML for exploiting their desperation.

How is nobody considering the broader political economy of scholarly publications and reviews? These are UNPAID reviews! Sure, maybe ICML isn’t Elsevier, but they are cousins to the socially parasitic and exploitative companies, at the very least.

Hiding behind a false “choice” to not use AI or basically not use AI isn’t an appropriate proposal. This is crooked and shameful. We should boycott ICML except we can’t because they are already the gatekeepers!

Has some similar conclusions to my Job Quality-Adjusted Displacement Index https://github.com/quinndupont/JQADI

I built my own MCP server to do some of this but I like the “enrichment” feature and NEO4J relationality.

I’m waiting for the agentic models trained on virus and worm datasets to join the red team!

Summary: good scientific theories have “reach,” which is not defined in any precise way. Reach has complexity and this can be handled with large parameter neural networks. Assumptions: mechanistic and deterministic worldview; epistemological perfection is the goal (perfect knowledge of facts).

HN optimal posting time & advertising value model.

Agreed. My point for spinning up this alternative metric is that the policy implications of the original suggest people should go out and get tough, dirty, dangerous jobs if they don't want to be displaced. But, there's a reason you don't see many 60 year old people in the trades.

That's largely based on the original analysis and their methodology. "Responsibility" (only attributable to humans) is one reason, another is that judges probably don't want to speak with robots in court.