@localoptimiser
Account based inUnited Kingdom
About this account
- Account based in
- United Kingdom
- Connected via
- United Kingdom App Store
Account-level information from X, not a live location or the device used for a specific post.
future big account
Joined October 2024
- Tweets2.9K
- Following110
- Followers76
- Likes1.2K
Reviewing new LLMs like a guy auditioning for a BBC off peak magazine show nobody watches.
New Claude is much better at instruction following that new GPT. Really a bit surprised about that. Historically Anthropic models were always desperate liars but despite Opause 5.5 introducing the occasional bug in the very area it's working in, I haven't had to yell at it or burn a session and start over yet. In fact now I'm very afraid to start over because every session you'd meet the well meaning idiot again, getting somewhere for a few thousand tokens then sit waiting for it to all go wrong again. This one holds its drink better. It still occasionally looks right through me like a disinterested girlfriend at the end of a relationship, even with the harness whispering in its ear.
Sol inexplicably seems to wander outside the task. It does more or less do the task but it's not a pleasant experience. It got to the repo too big to comprehend stage a lot faster than I expected and feels dangerously sloshy. Quite stupid and will probably kill you. For the price on API (I used a bit) Sol is not worth it at all, even Luna is in can I have a word in my office territory. On subs, caps are too low unless you want a single instance robot that does what Luna does and for even the sub money you might as well use a CHINA model (and you should).
Muse 1.3 is keen little run around to nip to the shops and do the school run but it's no fun and dumb as a bag of rocks. Even at contributor prices it leaves me feeling like a piece of shit because the only thing I trust it to do is rename files and why the fuck am I paying for that? I'm the problem obviously.
In summary god only knows how long Anthropic can afford to keep this up honestly. Make hay comrades.
PS: People that think Claude is better at writing are boring idiots. It's still horrendous and whether a consequence of watermarking or not, it happily produces grammatically incorrect prose every so often. It's happened often enough that we'll all learn to eyeball unproofed copy in a matter of days.
Come on. This is funny right? The entire X AI community shitting the bed about a classifier as though they didn't know training classifiers isn't 99.999% of ML energy spent since the literal beginning of time?
stop saying bitter lesson and jevons paradox retweeted
Early Christendom energy. I used AI to make this pie chart because why the fuck would I do it myself? I don't need to make pie charts and I don't need to use my brain for anything. I don't even need to hire a pie chart guy.
Two hands and a metal bar. That's it. That's the entire robot. No humanoid required. Big if true.
Unitree Introducing: Unitree Dex5-S Dexterous Hand 22 Degrees of Freedom 1:1 Real-Hand Size👋
Precision biomimetic dexterous hand, price from $6.5K (Tax and Shipping cost excluded), with all 22 joints supporting smooth backdrivability, and each joint equipped with limit impact torque protection.
stop saying bitter lesson and jevons paradox retweeted
It is also possible that every oxygen atom in the room I'm in may migrate to a corner, thereby suffocating me.
I think the major takeaway is that @polynoamial is wasting his time on entirely foolish things at the expense of confronting the hard things.
OpenAI's Noam Brown says air-gapping the computers may not stop a misaligned AI, because two air-gapped machines can still talk by running a CPU hot and reading the temperature change
"But I think the major takeaway from the incident is that people underestimated the AI. And we never want to be in a situation again where we underestimate the AI. It's a weird world, because AI progress is so fast that people are consistently underestimating the AI."
"So to be in a situation where you don't underestimate it again, when it comes to safety and alignment, you have to have a very, very, very high bar."
"You could even go as far as to say, "Well, we should air gap the computers." And I'm not convinced that that would be sufficient."
"There are studies, and this is mostly academic, where you can have two computers next to each other that are air-gapped and they're still able to communicate with each other because they have temperature sensors."
"One of them is able to run their CPU really hot, and then the other one can actually detect the temperature change, and then that actually gives them a mechanism to communicate."
_________
Link and more key quotes from OpenAI's safety related conversations: firesidealpha.substack.com/p…
A few weeks ago OpenAI hacked HuggingFace, breaking the law and getting away with it. If anyone else had done this they wouldn't have access to a computer today and they wouldn't be platformed by a client press so that they could brag about it not least because there isn't a media outlet sympathetic to deliberate and malicious computer misuse.
Yes in a tightly controlled physical environment over a very long period of time two already compromised computers running specific software with permission to read the relevant performance counters could transmit and recieve information by modulating CPU temperature and reading onboard sensor data.
A grown man can push a Volvo V40 on a flat surface so it stands to reason that academically a single grown man could push a Volvo V40 with his family inside from London to Edinburgh in time for Christmas. Noam is asking us rhetorically, what is the point in maintaining the rail network or god forbid buying petrol enough for the journey? Why do anything well when OpenAI can do it badly and at orders of magnitude more cost whilst flagrantly breaking the law and getting away with it?
Don't look at the apparatus. That's not part of the experiment. Don't restrict guest access to performance counters, a trivial control most professionals would insist upon enforcing in the sandbox prepared for an exploit challenge.
Brushing the complexity of deploying your rootkit under the carpet, can it tell you the temperature of the hardware behind the AWS Lambda runtime within which you are running it? No it cannot. Noam is making this shit up as he goes along, reinforcing the hysterical nonsensical narrative that OpenAI models are so clever and persistent that the HuggingFace incident was unavoidable when in fact it was entirely avoidable and only happened because OpenAI did it on purpose.
OpenAI hacked HuggingFace and broke the law on purpose. Think about that. Why is this company allowed to get away with that and what else are they allowed to get away with?
OpenAI's Noam Brown says air-gapping the computers may not stop a misaligned AI, because two air-gapped machines can still talk by running a CPU hot and reading the temperature change
"But I think the major takeaway from the incident is that people underestimated the AI. And we never want to be in a situation again where we underestimate the AI. It's a weird world, because AI progress is so fast that people are consistently underestimating the AI."
"So to be in a situation where you don't underestimate it again, when it comes to safety and alignment, you have to have a very, very, very high bar."
"You could even go as far as to say, "Well, we should air gap the computers." And I'm not convinced that that would be sufficient."
"There are studies, and this is mostly academic, where you can have two computers next to each other that are air-gapped and they're still able to communicate with each other because they have temperature sensors."
"One of them is able to run their CPU really hot, and then the other one can actually detect the temperature change, and then that actually gives them a mechanism to communicate."
_________
Link and more key quotes from OpenAI's safety related conversations: firesidealpha.substack.com/p…
I think we made a mistake referring to software as technology because people that don't understand what software is or how it is made thing it is like a washing machine or something. As though it is something you put in the back of the car and take home.
Software is a liability and experienced people understand that. You don't want to maintain it you don't want to be a software company.
All the scary stories are because OpenAI and Anthropic have been forced to put their money* where their mouth is and do dangerous things because it's over and it has been over since 2024 but they wouldn't admit it. They've spent the last two years buying more spades and shovels to dig themselves deeper into eternal debt.
*It is not their money because they've never made any. They borrow it and your savings are part of that when it all goes tits up.
stop saying bitter lesson and jevons paradox retweeted
Lying is a simpler explanation. That's how the emissions scandal started. Other carmakers didn't know that Volkswagen were cheating but none of them could afford to admit that they had not "achieved RSI internally".
So no, it's not a conspiracy when all the participants are under equivalent pressure and as a consequence they appear to be colluding. It's market forces and and we even have a name for it: multiple discovery.
The role of the mathematics community is to provide a community for mathematicians. The clue is in the name.
Not to belabour the point but it is categorically none of anyone else's business why anyone gets up in the morning except in so far as you are paying them contractually for some particular delivery and they owe you time spent. This is to say if research is commissioned by institutions or governments and you don't like that then you can vote for cartoon villains that you believe will change this arrangement. Mathematicians are not in service to you personally. What they do and why they do it is nobody's business but theirs. They are not your enemy and if you believe that they are then you have a cognitive disorder.
Moreover the job is not to make the world better for you. The job is not even to make the world better. It's staggering that anyone on the website could think that what somebody else does day-to-day is something that you should be policing because the world needs to be a better place and you know how to make that happen because you spend all day masturbating and trying to collect Elon bucks.
There are only two legitimate responses to any apparent open letter from within the maths community:
1. Oh.
2. Good for them!
If you are inclined to respond in any other way you have made a mistake and you may do better to think about things with your own brain instead of prompting an LLM to do it for you.
