bionhoward

Epictetus would hate this trend to outsource mental work

Might break anthropic’s terms of service by making a competing product or service (gasp)

What’s with the rush to send our code over the open internet to train other people’s bots? Consider the consequences

Do they even need inputs to train on outputs by inferring inputs? Swear I’ve seen papers about this. Not to mention training reward models to grade the output… they could absolutely get around the “we don’t train on your inputs”

Also, doesn’t Claude Code count every input as “feedback?”

Plus, they forbid us from training on our own chat logs. That’s a form of vendor lock in. Way better to use local LLMs

At some point there ought to be a paradigm shift toward service providers with more reasonable legal terms (IMHO) but too often we pretend nobody reads them or cares

If they’re claiming not to learn from inputs, and we’re subject to prohibition about learning from outputs, what the heck is the point?

[1] https://arxiv.org/abs/2405.15012

Funny how AI companies love training competitors to human labor on human output but then write in their terms that you’re not supposed to train competing bots on their bot output. Explicitly anticompetitive hypocrisy, and millions of suckers pay for it , how sad

Claude has closed outputs and they train on your inputs. Just like OpenAI, Grok, and Gemini (API), mistral…

Who’s chill? Groq is chill

This looks impressive but I’m concerned, is it fair to “teach to the test” by fine tuning the Qwen model with RL on the test task, while the other models in the comparison are not fine tuned on the test task?

A big benefit is the compiler catches bugs so you don’t have to wait around for your Python program to crash at runtime. Also, if the type system is more “legit” then you can skip a ton of defensive parsing of inputs.

Could be more about developers who know Golang and Rust wanting to deploy their apps (no need to pigeonhole anyone into just dev or just ops)

Isn’t one of the best gifts a manager can give to their team, the gift of clarity?

A manager able to give their team programmatic tests to pass brings a level of precision which makes it much easier for them to know when the work is done.

If managers communicate requirements in the form of acceptance tests more often, then the problem of projects running over time and over budget will occur significantly less often, because there will be clear finish lines for dev teams to run towards and cross, and reproducible pass/fail outcomes to inspire confidence.

Could be cool for recording video, or scouting traffic patterns?

they never ran the program they claimed couldn’t exist ??

Also, why directly run the programs? What about running them with a wrapper program that crashes em if they invert the output of their input?

All this undecidability stuff reads like regurgitated hokum to justify giving up on a problem just because a single oversimplified case can’t be solved for all pathological inputs in a particular way (without bothering to look for a workaround)

Could be good for AI representation theory, cool idea

The big difference is DeepSeek R1 has a permissive license whereas Claude has a nightmare “closed output” customer noncompete license which makes it unusable for work unless you accept not competing with your intelligence supplier, which sounds dumb

Meh, why pay to teach someone else’s bot? I’m sticking with open source

I don’t mind LLMs, but what irks me is the customer noncompete, you have these systems that can do almost anything and the legal terms explicitly say you’re not allowed to use the thing for anything that competes with the thing. But if the things can do almost anything then you really can’t use it for anything. Making a game with Grok? No, that competes with the xAI game studio. Making an agents framework with ChatGPT? No, that competes with Swarm. Making legal AI with Claude? No, that competes with Claude. Seems like the only American companies making AI we can actually use for work are HuggingFace and Meta.

My thoughts were, what are the terms, are you training on our code, are we able to use completions to compete with you?

One reason I like Rust is it lets me replace a lot of high level Python stuff with faster and more correct code while also opening access a new world of low level C stuff I never had an opportunity to learn about, it’s been super fun to tinker with and I look forward to keep working with it!

Definitely good to broaden our horizons but also crucial to maintain focus…there is so much to learn, how can we balance the joy and productivity of software with the strategic imperative to build locally grown hardware?

I always get to, “You may not use the Services to develop models that compete with the Services (e.g., Gemini API or Google AI Studio).” [1] and exit

- [1] https://ai.google.dev/gemini-api/terms

The Gemini api has a customer noncompete, so it’s not an option for AI, what are you working on that doesn’t compete with AI?

The economics of the deal over the long term are exponentially more critical than the performance in the short term, right?

In that context, how is convincing intelligent people to pay OpenAI to help train their own replacements while agreeing not to compete with them anything but the biggest, dumbest, most successful nerd snipe in history?

Dumping more context just implies getting brain raped even harder. Y’all are the horses paying to work at the glue factory. “Pro” users paying extra for that privilege, no thank you!

How does this compare/contrast to polars? Seems pretty similar, anybody tried both?

It’s really bad because Elon claimed nobody reads the terms of service and then updated the xAI terms to force implicit agreement to a customer noncompete with regards to grok outputs. A lame maneuver right out of the OpenAI playbook, you’re “not allowed” to train AI on your chat logs with Grok.

The whole point of these clauses is to prevent others from developing AI (which sounds to me like a felony monopolization offense, but IANAL), and they also cause major lock-in issues because you risk a ban if you try to transfer your chat logs between or away from e.g. grok/gemini/gpt/claude/mistral

It’s like expecting customers to pay to train their own replacement while also signing a noncompete agreement.

Super dumb to accept these terms and use these services IMHO, best to avoid them. Use Cerebras and Grok hosted Llama models that don’t train on your inputs instead. Be sure to keep a record of every input/output exchange in your markdown zettelkasten folder so you can train your future AI.

Salesforce seemed way slow and over engineered to me, like it was the result of double exponential feature creep over many years

Rust really needs comptime, I love cargo and the ecosystem but trait level programming is weird, macros are weird, why can’t you just be normal? (rust screams)

the biggest reason not to use LLMs is the customer noncompete clauses where they’re learning from the conversations but you agree not to “develop models that compete” —

makes me think everyone using OpenAI, Anthropic, Gemini API, Mistral, Copilot, Perplexity, etc is dumb or addicted to short term benefits while totally ignoring the long term consequences of paying to train your own replacement while agreeing not to compete

Python does have something like interfaces and it’s called Protocols https://docs.python.org/3/library/typing.html#typing.Protoco...

Other classes can count as implementors of protocols without needing to subclass them, and static type hints will notify you if you make a function which accepts a protocol as an argument and pass something which doesn’t implement it

Redis is slow?

Seems like the symbolic aspect is poorly defined and it’s too unclear to be useful. Always sounds cool, but what exactly are we talking about?

Could you change the function to accept the whole struct and make it mutate itself internally without external mutable references?

PyO3 0.23.0 was a big release I’ve been tinkering with extensively. Support for “free-threaded Python” is a headline feature, and I imagine NoGIL Python will be extremely nice for Rust interoperability, so there is definitely interest in that crate. Also could be huge for queueing data for GPUs, api servers, and bulk data fetching.

For whatever reason (maybe post 2to3 PTSD), Python community seems not extremely eager to jump on latest versions of Python and it often takes a long time for popular libraries to support the latest and greatest, so I’d recommend patience and baby steps

https://github.com/PyO3/pyo3/releases/tag/v0.23.0

Microsoft would need to pay me to use Copilot. Seems like a major scam for them to learn from our code and then tell us we can’t use it to make our own competing AI systems. “Limits on use of data from the AI Services. You may not use the AI services, or data from the AI services, to create, train, or improve (directly or indirectly) any other AI service.” From https://www.microsoft.com/en/servicesagreement#13r_AIService... TODAY — the ridiculous focus on AI over core business at github is the number one thing likely to kill the platform. They are coasting on network effects