RA_Fisher

Ah okay, can it work on a whole repo in an agentic way?

Claude Code will write the whole thing for you. Whereas doesn’t Copilot require input along the way of coding? ie- it doesn’t do all the programming for you

Do people bring their own then (considering work doesn’t pay for it)?

AI gives us a means of leverage. We can do more with less. production = f(labor, capital, technology) + eps

Anthropic and Claude are running circles around Google / Gemini for me these days. Anthropic was quite helpful for a while but strange limit issues started popping up. The final thread was a bug that essentially broke my ability to develop. I moved over to Claude Code full time and haven't looked back. Opus 4.6 is awesome for accelerating probabilistic programming!

They’ll probably receive most if not all of Iran’s focus now.

I limit the risk and insist on payment upfront.

Location: Brighton, MI, USA Remote: Yes Willing to relocate: Yes

Technologies: Bayesian statistics, econometrics, causal inference, experimental design, machine learning, AI systems, R (expert), Python (expert), SQL (expert), Stan/brms, PyMC, scikit-learn, tidyverse, PostgreSQL/Redshift, AWS, Airflow, Ansible, Terraform

Résumé/CV: https://statwonk.com/about/

That’s good, reducing healthcare costs will increase access and boost the our health.

Agree that AI should replace CEOs. They’re often biased in unhelpful ways that AI isn’t and it costs people wellbeing.

True, though I don’t think they’re too different on that dimension.

AI makes it easier to catch up. :)

I use AI to shop and it seems easy for an AI to understand when that’s the case (they can do statistics and physics after all).

I think they'll keep the status quo of quarterly, bc if any announce switching I'd expect their stock value to fall (bc in my mind the decision would transmit more bad potential than positive potential for future earnings). ie- I don't buy that quarterly reporting drives too short-term decision-making (or that it's generally too short).

I built this after noticing how much variation there is in AI output depending on the quality of the prompt. Models matter a lot — but even the best model underperforms when the prompt is vague. Enriching the input is one of the highest-leverage things you can do.

Most prompts are drafted in 10 seconds and sent. They're missing context, constraints, a clear objective, examples. The model fills in the gaps with whatever it wants, which is why the output often feels generic or off.

Prompt Enricher scores your prompt across 5 dimensions (the YIELD framework: Your Objective, Input Context, Expectations & Constraints, Layout of Output, Demonstrations/Examples), then rewrites it with the missing pieces filled in. It also produces 4 variants — Packed, Quality, Concise, and Reliable — so you can match the enriched prompt to your use case.

The "Show Changes" diff view that highlights exactly what was added (green) and removed (red) — so it's immediately clear what enrichment actually did to your prompt.

Does that extra work bring in more revenue? I think that’s the key question.

Progress can be reviewed over time, and I'd think that'd take a lot of the risk out.

I’m a fan of using a variety of methods to teach, no issue with that. My issue is with teachers that don’t admit how the world is changing. Dinosaurs.

What seems false?

I use them to probabilistically program. They’re better than me and I’ve been at it for 16 years now. So I wouldn’t say they’re superficial at all.

What have you tried to use them for?

Claude, Codex and Gemini can read code much faster than we can. I still read snippets, but mostly I have them read the code.

Agree, if Bezos hadn’t alienated the readership, they’d probably be doing well.

I used to look up to him before he became an obsequious traitor.

Very fair. I think even more we underestimate our own sentimentalities. eg- the teacher that believes adding or multiplication has to be done a particular way (like the standard algorithm vs. partial products).

I think learning arithmetic is a good idea, but it’s only a part of computation. I don’t think we should get too hung up on a particular method of computation (bc there’s so many ways).

It’s possible to run up to 4 agents at once vs. Claude Code’s single thread. Sometimes I’ll find meaningful quality differences between what agents produce.

That reminds me of when teachers would say: what if you're without a calculator? And yet we all have smartphones in our pockets today with calculators.

Infinite. :) Chess is strictly unbounded.

Very true. The queues need to be improved.

Exactly, air security has actually done a really good job over the last 25 years. I hope they keep improving it.

Econometrics allows us to understand and measure these things. This result isn’t surprising at all. Treating employees well has a similar effect in improving productivity.

If they made it clear when skills were being used / monitored that, it'd seem to mitigate a lot of the problem.