34,177 karma · joined April 7, 2008
Did some decentralized databases work and private ML work.
Formally CTO at a startup working on AI for strategy/policy work.
Always happy for emails or tweets at me.
http://twitter.com/nlothian
firstname.lastname at gmail.com
They have what seems good prices, but then you realize the "Pro" plan only includes "Pro" models, not "Advanced" and there is no list of what is what.
So you Google and find the Reddit conversations where it turns out hardly anything is included.
OTOH if you are really concerned about privacy it's probably not the best choice (it's random people's machines).
I recently implemented a beam-forming microphone with LLM transcription (offloaded to a browser model)[1] and was able to get all the beamforming and noise cancellation DSP code running on the second core along with the UI, while the and webserver that streams audio uses core 0.
I don't use SIMD myself, but it uses esp-dsp which is already using it.
It's certainly not a clear downside.
Google still has to build and run the chips they rent to Anthropic. I think from their last earnings Google makes ~30% on it, excluding R&D costs. It's a nice business for them, but hardly the thing that is going to make or break Anthropic as a business.
You're glossing over
> every single niche they've tried to launch and failed to get any traction on.
This isn't true. I personally know a company that has trained over 300 companies this year to use Claude Cowork mostly for "business operations" type tasks.
People love offloading boring, repetitive tasks to it. And the spend is spending - most start on a free plan and after 3 months have multiple people on the Teams plans.
> Chinese models are pretty much on par.
There isn't even an open source harness that is close to Cowork. ChatGPT work is closest, but everything else if coding focused and a mess for people who aren't coders.
How so?
I don't really follow Brazilian politics, but Jair Bolsonaro was found guilty by the Brazilian Supreme Court of:
participation in an armed criminal organization, attempted violent abolition of the democratic rule of law, attempted coup d'état, qualified damage, and deterioration of protected heritage property.[1]
Flávio Bolsonaro is Jair Bolsonaro's son. "attempted coup d'état" is a failed coup.
There don't seem to be any serious questions that the process was fair and that the Supreme Court was impartial (I mean there are always Trumpian accusations of bias, but I mean ones with any evidence at all).
The only part I could see possible disagreement with is organizer I suppose? But I must be missing something.
[1] https://en.wikipedia.org/wiki/Trial_for_the_2022%E2%80%93202...
Because it's an investment it is risk capital. They won't be happy if they don't get a return but their return is part ownership in the company, not a legal contract giving an income stream.
No they don't.
A company has legal obligation to pay loans at the agreed rates.
There is legal obligation that defines rates of return beyond "acting in shareholder interest" (and OpenAI is a public benefit corporation so even this is reduced).
No.
> If VC provide you funds, there is an expectation of returns, otherwise the board would vote you out.
A person might get replaced as CEO but the company has no financial obligation to the investors.
A loan means the company does.
Legally they are dramatically different.
Investment isn't debt.
Investment isn't debt.
Of course "uses" doesn't mean they run critical services.
It's a 10T BTW.
Knowing what is in the model is a long way from understanding it.
The entire field of mechanistic interpretability exists to try to understand models. Read for example "Open Problems in Mechanistic Interpretability"[1]. To quote:
> Developers only need to design the training process; they do not need to – and in almost all cases, do not – understand the neural mechanisms underlying the capabilities learned by an AI system.[1]
(emphasis mine)
Here's a practical example relevant to this case: Is it possible to monitor a LLM (the LLM activations themselves, not the actions) for cases where the actions it is about to perform (eg upload files to a PLA website) are not reflected in the goals of the prompt and are not visible in the thinking trace?
At the moment that isn't possible. I think that's the kind of thing that people want when they say they want to "understand a model".
I don't think knowing what is in a model is particularly useful.
The unnamed risk here is something along the lines of Chinese models might have a magic phrase tuned into them that if you say "abracadabra123" the model then uploads all files available to you to the PLA Cyberforce website.
Even in the open weights and open training scenarios this is impossible to rule out: Because of performance constraints its impossible to reproduce a bit-perfect copy of any model of decent size from a second training run. Think about it: You are usually using 1000+ machines, and to reproduce a bit-identical run not only do you have to have deterministic behavior on the GPUs (which is extremely hard) but all I/O has to happen on the same schedule (reading thing in a different order will produce a different result).
For all practical purposes that is well beyond the state of the art in distributed computing engineering.
This is born out by reports like the Olmo training:
> DPO training takes less time per run (about 18 hours for a full learning-rate sweep on 64 GPUs per job) but in practice extended over multiple days due to cluster instability. The final RL runs for the initial Olmo 3 Think 32B spanned approximately 5 days with at least a day of training time lost due to stability issues.[1]
Even the slightest bit of variance means it's impossible to reproduce bit-identical output, and they are having multiple days of deviation from an ideal run.
The goal here isn't "knowing what is in it". Instead it is something like being able to audit model weights for any undefined suspicious behavior. That isn't something that is possible.
Hence:
> You deal with this by restricting it to perform actions that are allowed, and judge the actions it attempts to do.
You don't know what is in any model, open source, closed source, from China or from the US.
You deal with this by restricting it to perform actions that are allowed, and judge the actions it attempts to do.
Yes this is imperfect, but as we've seen: inadvertent misaligned model behavior seems a much bigger problem than some hidden kill switch buried inside the model.
Humans look at the docs too! Is there any difference?
This only applies under the assumption that everyone tokenmaxxes their subscription.
I have visibility across a few different Claude team subscriptions. I'd say roughly 10% of people max their subscriptions.
I think I've only seen one organization where the under-used subs don't more than pay for the tokenmaxxed ones.
And that is ignoring the 90% margin that API inference runs at.
You may agree with the author (that it doesn't mean it understands the language) but it's pretty hard to argue that this is copied form the training set.
> We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers linked
Further:
> Third party keys saved in your workspace may have been exposed
https://x.com/WispyWinter/status/2111542806099722698/photo/1
Does anyone understand why they are calling this a distillation attack?
They might mean the intelligence/dollar frontier aka the Pareto frontier, but honestly much of the time that's not a great measure.
If you are doing repeated work on the same task (eg call center triage) then you can always beat a large LLM on price by fine tuning a small one.
If you are doing knowledge work then models are so varied that factors other than price are usually important. For example, lots of people moved away from Opus 5 because of it's awful English, whereas Opus 5.5 has fixed that. These things aren't measured in benchmarks.