All the methods and data are not public. We don't know what unpublished methods they're using. You can get most of the pre-training data publicly but they've probably spent a ton of money curating it and are now doing things like buying rare books. The RL training data is all (/mostly) proprietary though, and that's the real secret sauce part.
All the RL data are exactly public. There are huge amount of distilled data freely available, and that amount is more than enough to train a ~10T model.
There are 'transfer stations' and that's how exactly I use GPT and Claude in China. OpenAI and Anthropic do not sell in China, so we use their AI with a much lower price like 1% of the official API price. The largest transfer stations have TBs of traffic every day, and the traffic is eventually possessed by the open source community.
Subscription engineering is a deep field. Neither OpenAI nor Anthropic have any technical advantage in this field.
I think a big issue is that the companies pushing AI want it to replace people entirely. So the software around it is designed with that mindset.
I'd really like something that works more like pair programming. Where you share an editor session with the AI, and it's much more interactive. eg. It says "I'm thinking about doing X here, what do you think?". Then you give a response and it continues. Or you can see it doing something silly and immediately stop it and correct something either with a prompt or by manually editing yourself, before letting it rip again. Or just ask it "why are you doing it that way?". Maybe with some control over the speed it's going, for when you feel confident about what it's doing.
Claude has made some positive changes over time where it asks you more when there are different approaches it can take. But I feel like I'm not being brought along with my mental model as much as I would like. A lot of the time it spits out a whole pile of code and then I have to go and build up the mental model after the fact, and correct a lot of what it's done.
The current approach is good a lot of the time though, when you're not really changing anything architectural and just want it to bash out code while you do somehthing else.
It's the exact same training process for both of your examples. I don't really see how you can claim books are not replicated, but that output from other LLMs is.
Process is the same, but intent is not. One intent is to extract information from the book for better general eloquence and overall awareness - not for replicating the book itself (ability to recall verbatim fragments is a side effect, not the goal). Another intent is to replicate the behavior, carry it over using training.
Again - IANAL - but in my understanding (and I spend some time reading on this), the legal concept of fair use is all about the intent how copyrighted material is used. It's all copying or distribution, but law does make distinction about what and why.
They have no choice but to train their own model to try and survive. They're paying API pricing for the top tier models but competing against subsidized subscriptions.
The problem is, what's ambiguous or precise is subjective. Your devil's advocate needs to reflect all of the possible readers, and that isn't possible.
There's a good reason we use jargon in professions, or more constrained and less ambiguous languages for maths/coding
While I understand the spirit of this comment, if you look at the fossil record you’ll see that’s objectively not true.
Roughly half of the shifts in the last 11 evolutionary periods, over the last 500 million years, were caused by changes that occurred in a-few-hours-to-a-few-thousand-years with 75%-90% species lost.
You are tautologically saying that massive shifts resulted from massive changes, but that doesn't contradict the statement about evolution--which is about far more than such "shifts" (not an aspect of nature but rather changes large enough for humans to perceive)--operating over long time periods. Every single instance of offspring is a "shift" from its progenitors.
Also talking about evolution failing to work is a category mistake--evolution is an ongoing process that is the inevitable result of imperfectly replicating biological mechanisms and there's no "succeed" or "fail" about it.
Only if our explicit goal is to preserve the exact environment that was in place when humans showed up and gained enough knowledge to decide change wasn't allowed anymore
>With RL, models no longer just learn what sounds correct based on patterns they've seen. They learn what words to output to be correct. RL is the process of forcing the pre-trained weights to be logically consistent.
How does Reinforcement Learning force the weights to be logically consistent? Isn't it just about training using a coarser/more-fuzzy granularity of fitness?
More generally, is it really solving the task if it's given a large number of attempts and an oracle to say whether it's correct? Humans can answer the questions in one shot and self-check the answer, whereas this is like trial and error with an external expert who tells you to try again.
The underlying feature of FIAT money creation is debt. And debt is a very natural thing (existing before money) that will just manifest in the crypto system instead.
How do you know it's safe to redeploy? If your entire operation may be compromised, how can you trust the code hasn't been modified, that some information the attackers have doesn't present a further threat, or that flaws that allowed the attack aren't still present in your services? It's a large company so likely has a mess of microservices and outsourced development where no-one really understands parts of it. Also, if they get compromised again it would be a PR disaster.
They're probably having to audit everything, invest a lot of effort in additional hardening, and re-architect things to try and minimise the impact of any future attack. And via some bureaucratic organisational structure/outsourcing contract.
You literally have some of your team buy new laptops and hang out in a temporary wework to set it up on entirely new infra, air-gapped from your ongoing forensic exercise. You just need to make sure none of the people you send are dumb enough to reuse their password. You need to take the domain name, but they will be using one of the high end domain companies so that can be handled.
Bear in mind that this is a company which still sells physically and has retail and warehouse staff. All that the e-commerce side needs to do is issue orders of what skus to send to what addresses, and pause items that are out of stock. M&S is not Amazon and doesn't have that many SKUs, 5 people could probably walk round the store in a few days and photograph all of them for the new shopping site.
Sure, customers will need to make a new account or buy as a guest. But this stuff is not hard on the technical side. There is no interaction between customers like a social media site, so horizontal scaling is easy.
Now I get that there are loads of refinements that go into maximising profit, like analytics, price optimization, etc. But to get in revenue these guys don't even need to set up advertising on day one because they have customers that have been buying from them for decades. The time to set up all that stuff is when your revenue is nonzero
> M&S is not Amazon and doesn't have that many SKUs, 5 people could probably walk round the store in a few days and photograph all of them for the new shopping site.
I can't speak about M&S buy all big physical retail brand which started selling online are exactly operating as Amazon with SKUs coming from various third party entities. The offering is much bigger than what is sold at the physical shops.
I had the impression that M&S wasn't, but if that's the case then yeah, that would invalidate my analysis. Especially if even their retail stock goes through that route when bought online.
reply