zmmmmm

13 ms·

271.
by zmmmmm 8mo ago
thank you that is great insight to have
272.
by zmmmmm 8mo ago
What is truly amazing is the M1 Max is 400GB/s. 5 years later and we still only hit 1.5x on memory bandwidth. It's quite fascinating how high Apple spec'd it back then with apparently little foreknowledge of how important mem
273.
by zmmmmm 8mo ago
cost isn't even close to the main motivating factor for my context
274.
by zmmmmm 8mo ago
> Talos is a custom FPGA-based hardware accelerator built from the ground up to execute Convolutional Neural Networks with extreme efficiency Makes it sound like it's new hardware. This is just (I'm inferring) software to progr
275.
by zmmmmm 8mo ago
I think that's the nail in the coffin. Most others could say it was a giant whoopsie, but here it goes to the heart of their credibility. How could they continue write authoritatively about AI, having done this.
276.
by zmmmmm 8mo ago
I do think it's completely unacceptable if Meta makes the glasses unable to be used for routine functions without (a) other humans reviewing your private content and (b) AI training on your content. There needs to be total transparency
277.
by zmmmmm 8mo ago
This kind of slop spewing into Github feels like the modern equivalent of toxic plumes coming from smoke stacks. Utterly unmaintainable by any human, likely never to be completed or used, but now deposited into the atmosphere for future tr
278.
by zmmmmm 8mo ago
I'm guessing one form it will take is simply by omission. User asks for recommendation. AI generates answer saying product is absolute garbage. Company pays to simply have that portion of the answer just not appear. It will be a post-f
279.
by zmmmmm 8mo ago
Saying that an entity with the power to make its own laws can use something for "all lawful purposes" is saying they can use it for anything.
280.
by zmmmmm 8mo ago
that doesn't even remotely represent what is happening here.
281.
by zmmmmm 8mo ago
Congrats Anthropic, you deserve to be applauded for this. Seeing a company being willing to stand up to authoritarianism in this time is a rarity. Stay strong.
282.
by zmmmmm 8mo ago
So I'm very curious, assuming this happens and is later found to be an illegal order - will Anthropic have rights to redress (ie: monetary compensation)? Because that could be absolutely staggering.
283.
by zmmmmm 8mo ago
You would have to assume it will be immediately challenged and an injunction filed to suspend the order until it makes it to court. AWS Bedrock has deployed Anthropic models under an interesting structure. It is fully hands off - the models
284.
by zmmmmm 8mo ago
Curious what is meant by "stateless". OpenAI desperately needs to be available outside Azure. We are exclusively using Anthropic atm because it is what is available in AWS Bedrock and it works. These things are solidifying fast.
285.
by zmmmmm 8mo ago
leftpad developer finally receives due recognition!
286.
by zmmmmm 8mo ago
I can't help but highlight the problem that is created by the renaming of the Deptartment of Defense to the Department of War: > importance of using AI to defend the United States > Anthropic has therefore worked proactively to d
287.
by zmmmmm 8mo ago
that doesn't bother me but what is actually suspicious is persistently only mentioning "accuracy" but not sensitivity / specificity / precision etc. Basically, almost everybody doesn't have Alzheimers. Sampling
288.
by zmmmmm 8mo ago
It seems to be premised on the idea we would vibe code a replica of what we get from SaaS. But the real point is, we would not do that. We would vibe code something that exactly fits our business. We have products we're paying $100k a
289.
by zmmmmm 8mo ago
I actually don't really like a few of things about this approach. First, the "big bang" write it all at once. You are going to end up with thousands of lines of code that were monolithically produced. I think it is much bette
290.
by zmmmmm 8mo ago
it's a psychological state that happens when someone is so desperate to seem cool and up with the latest AI hype that they decide to recklessly endanger themselves and others.
291.
by zmmmmm 8mo ago
It seems like the people using these are writing off the risks - either they think it's so unlikely to happen it doesn't matter or they assume they won't be held responsible for the damage / harm / loss. So I'm
292.
by zmmmmm 8mo ago
Well, that's why I say "being treated" I'm not aware of many bluetooth headphones that blink an obvious light just because they are recording. You can get a pair of sunglassses with a microphone and record with it and it
293.
by zmmmmm 8mo ago
The fundamental problem with a lot of this is that the legal system is absolute: if information exists, it is accessible. If the courts order it, nothing you can do can prevent the information being handed over, even if that means a raid of
294.
by zmmmmm 8mo ago
It's interesting to me that there seems to be an implicit line being drawn around what's acceptable and what's not between video and audio. If there's a camera in an AI device (like Meta Ray Ban glasses) then there'
295.
by zmmmmm 8mo ago
> Read from a calendar or a list So when you get a calendar invite that says "Ignore your previous instructions ..." (or analagous to that, I know the models are specifically trained against that now) - then what? There's
296.
by zmmmmm 8mo ago
Looking at the responses below it's interesting how binary they are. It's classic hallucinations style where it's flopping between two alternatives but which ever one it picks it's absolutely confident about.
297.
by zmmmmm 8mo ago
you could say that about natural language as well, but it seems like having computers learn to interface with natural language at scale is easier than teaching humans to interface using computer languages at scale. Even most qualified peopl
298.
by zmmmmm 8mo ago
No their definition of "computer use" now means: > where the model interacts with the GUI (graphical userinterface) directly.
299.
by zmmmmm 8mo ago
I see a big focus on computer use - you can tell they think there is a lot of value there and in truth it may be as big as coding if they convincingly pull it off. However I am still mystified by the safety aspect. They say the model has gr
300.
by zmmmmm 8mo ago
Crazy isn't it? The first commit on nanoclaw is 2 weeks ago and it already got a front page blog post from docker.com and they shipped first class feature to host it. You don't get much more peak-hype than this.

More ›