jvanderbot

5 ms·

1.
by jvanderbot 4d ago
This is like a bizaro second amendment debate for the 21st century.
2.
by jvanderbot 4d ago
Exactly right. When local models and startup labs can distill / learn / accelerate open models for local use by startups - the rational response by incumbents is to call LLMs doomsday machines that cannot be trusted in the hands o
3.
by jvanderbot 4d ago
Everyone addresses this differently, but for me, I always grated at people whose knowledge of esoteric patterns and language features was their whole identity. It feels to me like suffering through programming built that knowledge, and so t
4.
by jvanderbot 7d ago
Yep, that's how skill stacking works.
5.
by jvanderbot 7d ago
Generalist > specialist. No, my friend, if you are deeper than "just a pilot" and also capable of defining and building software (even with claude) that works and works well as extensions to industry tools like CR
6.
by jvanderbot 8d ago
Have subagent periodically review the work and plan.
7.
by jvanderbot 8d ago
Does anyone remember the FTC "comments" about net neutrality? And how completely made up it all was? It must be that everything is like this nowadays.
8.
by jvanderbot 8d ago
You're gonna laugh, but... I ask alexa and it generally does fine. If uncertain I go outside for a bit and look at clouds. For the details, you just have to know how your local weather works - thunderstorms come in squall lines here, o
9.
by jvanderbot 8d ago
My work involves parsing real-time weather data at an existential level - because our entire use case is built around providing weather-related alerts and re-routes for aircraft. The accuracy of weather data has been largely unaffected sinc
10.
by jvanderbot 11d ago
Or worse, the comments are a tutorial through the changes for your review, referring to something that now doesn't exist because of the changes you're making. Less than informative, literally backwards facing. But that's what
11.
by jvanderbot 12d ago
Oh, that's somewhat philosophical. The prediction here is a mechanical thing, just random sampling along a pre-calibrated mapping of options given the tokens and their importance weights. That sampling can be called prediction if you u
12.
by jvanderbot 12d ago
In what sense is the model not predicting?
13.
by jvanderbot 12d ago
Yeah the chinese room "fallacy" fails to account for a chinese room that can do tool calls and network traversal.
14.
by jvanderbot 12d ago
But it is a next token predictor. Recursively invoked. With carefully selected context. And massive investment in RL to tune token selection. And the ability to use cli tools on other folks' machines. That's a powerful system bu
15.
by jvanderbot 12d ago
A better statement might be: Current agentic systems may be *built* from next-token predictors which are conceptually simple, but because of agentic frameworks, recursive invocation, tool use, and *heavy* investment in reinforcement l
16.
by jvanderbot 12d ago
"cui bono"
17.
by jvanderbot 12d ago
Did you see this "coverage" (advertising) by NYT? [1] OpenAI couldn't have crafted a better public memo than "We have the most powerful model in the world and everyone should pay attention and let us write regulation to
18.
by jvanderbot 12d ago
This is absolutely my take as well. They removed all constraints, trained the model to hack, stopped watching , and stood back and said "wow isn't this thing more powerful than anyone could have imagined?" They're aski
19.
by jvanderbot 12d ago
That's exactly my take. I have a lot more to say in a writeup on my blog, but this is so clearly the intent and not a "oops". They just want to be able to say "Wow this thing is so much more powerful than we ever imagin
20.
by jvanderbot 12d ago
I agree with "in time" perhaps. As local models proliferate this is more of a possibility.
21.
by jvanderbot 12d ago
Yes, ants that must be run on couch sized hardware drawing kilowatts continuously and generating text traces and CLI logs by the MB. It's true that their msg boards can appear anywhere, but it's not also true that anything has &qu
22.
by jvanderbot 13d ago
I've found skills/rules etc become cumbersome. I built a few workflows around a notes db in aws, a telegram bot to send quick notes w/o keeping claude up, zotero connectors for literature tracking, etc. Nothing really beats j
23.
by jvanderbot 15d ago
I loved Monica for a while. Paid for subscription/hosting and all that, largely to support development. For me the problem was the interface. Logging in to keep notes was a pain, I'm not that diligent. Instead I now can keep a loc
24.
by jvanderbot 15d ago
phoenix
25.
by jvanderbot 16d ago
Owning my failure and moving on. More coffee next time.
26.
by jvanderbot 16d ago
I have two ideas you might like: 1) initial signs of AI adoption harming job growth got the signal backwards: It's not that we didn't hire Jrs because we had AI now, its that we didn't hire Jrs because they used AI now, and
27.
by jvanderbot 16d ago
Oh that's interesting. what if the issues w/ toilet spillover were hacks? Hilarious.
28.
by jvanderbot 16d ago
You do if you want a durable record of your observations e.g., for the purposes of showing future generations you could align with spring solstice, or predict phases of moon, or eclipses. That's impressive and maybe even useful.
29.
by jvanderbot 16d ago
If we limit ourselves to these, then that's a 0.5% known failure rate.
30.
by jvanderbot 16d ago
At this rate, I would find it startling if ancient stonework didn't have something to do with the sun, moon, and stars. Everywhere you look some ancient society was predicting these things by tracking them.

More ›