> And before anyone starts thinking about whatabouting, Yandex are based in a country where journalists are openly and publicly assasinated to intimadate
Thankfully Gary Webb is here with us today to laugh at this
What's worrying for them is nothing they're doing is impossible to be recreated, if you bake it in and you don't want to pay rent you can fine tune your own model
The actor is bad, the deed is good. Though, the open sourcing of llama 2 and beyond was pushed by Yann Lecun and others, some of whom have moved on from the company. Zuckerberg supports the AI initiatives but he's hardly leading them himself.
What argument do you have that releasing llama was actually good? They gave AI access to terrorists and criminals as much as they did to independent researchers. Arguably they are partially responsible for escalating blackhat cyber capabilities that are contributing to the rapid rise in attacks. Open source is neither inherently good or bad. The most you could really say is that it was accelerationist
If you can’t explain why releasing llama was a net good then why bother commenting at all? One could surmise you don’t actually have any moral grounding for it and just like that you get to play with it for free
I can answer that question if you'd like, but the question you presented initially was extremely loaded and recycling tired political rhetoric that has already been beaten to a pulp.
Or you can just keep on making unfounded assumptions about my "moral grounding" if you'd like, you didn't leave a good first impression with me and I'm not terribly interested in discussing this with you if this is how it's going to be. Discussions are about finding common ground, but you seem to be looking for an argument.
Good action can also be a way for bad actors to buy good will so they can extend their other bad actions. Plus some of these actions are done on the interest of furthering the interest of the bad actors.
An example is Meta striking deals with telecom operators to offer unmetered access to Meta services making Facebook synonymous with the Internet in some countries.
>Good action can also be a way for bad actors to buy good will so they can extend their other bad actions. Plus some of these actions are done on the interest of furthering the interest of the bad actors.
We talking tech or Hell's Angels' teddy bear runs.
Oh, wait....
The second sentence of the first post was explication of the sentiment described and criticized by the preceding sentence. It is the sentiment I am calling out.
The first post could or should have been more clearly written with an em-dash, but then people would start calling me a clanker. /s
> Actor and action conflation is incredibly common these days — If someone is bad and worth condemnation, all their actions must also be bad.
My position: Bad person can do good thing for bad motives.
example: The evil wife beating doctor who saves a baby just because they get paid for it.
My "autistic" brain can recognize that saving babies is still a good thing. I don't have to like the doctor, or their reasons. For whatever reason, a large portion of people see a comparable situation and instantly start arguing that saving babies is bad.
Adolph Hitler is often credited (with varying accuracy) with the Autobahn, fixing Germany’s unemployment and launching Volkswagen to produce affordable motor vehicles.
None of these make him a good person any more than Meta’s occasional positive acts make it a good company.
Because it won't be in the training directly. It is applied after a model generates its distribution of likely tokens, biasing each token randomly based on a random key and unrelated to any meaning of the words. So half the time, the most likely token becomes more likely and half the time it becomes less likely, and the same for every other token (when temperature is above 0).
You then look at the tokens actually picked to see how closely they follow this pattern that isn't connected to the meaning of the tokens. With enough text, you can then analyze the chance of it happening by chance verses being because the generation of the tokens was done using the algorithm, and you can save a positive result until you are arbitrarily sure. There is a chance of a false positive, but the chance of a false positive approaches the chance that the murderer happened to have fingerprints that matched your and both forensics labs happened to have mixed up the dna tests and the eye witness happened to misremember the face and your phone gps happened to glitch out and put you at the murder scene at the time of the crime all happening. It is theoretically possible only in the same sense that quantum teleporting a cat is theoretically possible.
The real question is how much text do they need for a given level of certainty and what do they check for. If they flag a positive at a p value <.01, that's a problem. If they can reasonably get a p value of < 1e-12 in only a few paragraphs of text, that is effectively no false positives (but a lot of 'too short to analyze' outcomes).
How do you verify in practice then? Wouldn't you need the original prompt so you can reobtain the likely token distribution to validate again the random key(s)?
What about "watermarked long-form code"? I'm having a hard time understanding how a model could watermark not prose, but functional/semantic text like code, that actually has meaning. You can't switch our the characters, you can't use various types of whitespace, you can't add arbitrary code comments, and a lot of other restrictions. Is there any state of the art methods for watermarking code without affecting the quality/correctness?
For straight generated code it'll likely need more text, but it'll still show up.
In cases where one token is extremely likely, it'll randomly be red or green and still be picked in either case as it is simply the best (or only) option. So you'll have more tokens that don't show a pattern either way (half of these cases will match and half won't, just the same as if a human wrote it). Meaning you'll need more instances where multiple tokens were all likely to see if there is a pattern. Given the check algorithm can't identify these cases, it can only judge on the overall text, so the more strict a language, the more the length requirement scales.
Where I wonder if this keeps working is in tool calls. Often, you don't take code straight from the llm, you take the results of a tool call to edit already existing code. It might be that the result of this leads to far too few signals to pick up, meaning that this only works when one does significant generation with a single model (even swapping between different models, at least by different companies, breaks this just as much as having a human write parts of the code).
Think of it like finding a loaded dice. A dice that has a slight bias in a few dozen roles is just random chance. If that bias continues after hundreds of thousands of roles, the dice is loaded. But will a code base have enough samples, especially when edits made from tool calls? I could see this being unable to detect things at the size of a reasonable PR and only being useful for massive sets of changes and only if the person behind them didn't structure their AI usage to avoid detection.
As anybody who has put together a coding standard knows, there are a lot of options for individual expression, meaning a lot of room for things like watermarking. And of course you can add arbitrary comments; my Claude-generated code is very verbose.
> there are a lot of options for individual expression, meaning a lot of room for things like watermarking
The way I use LLMs (and I'd advice everyone to do the same) there really isn't, the agent implements things exactly how I want them, or I use the agent to massage it into the exact bit-by-bit version I imagined when I first sent the prompt afterwards. I honestly don't know what the point would be to let the agents compose worse code than what I'd do manually, although I know it's a popular approach taken by many.
> And of course you can add arbitrary comments; my Claude-generated code is very verbose.
So watermarking for all users who allow code comments from agents, no watermarking for us who force the agents to never write a single code comment? Alright, I'd be fine with that.
From what I've seen, your approach to LLMs is exceedingly rare, so I suspect it's one the people who care about watermarking aren't very concerned with.
And the reason to let Claude make worse code than a professional would by hand is basically suppressed demand. Since programmers are expensive, previously code mostly got written when a large number of dollars were on the line, or when an individual programmer did something not economically optimum (e.g., hobby project).
That left a whole lot of somewhat less valuable software unwritten. It's the economic space that no-code tools have been nibbling on for years. One way to think of things like Claude Code is as effectively no-code tools. Pre-LLM no-code tools would produce data structures that got executed by special environments without ever being seen or tuned by a human. Claude Code can be used just like that, with text as the input and python as the intermediate representation that nobody ever looks at.
That approach probably isn't sustainable for what we professional programmers would call a serious project. Claude can easily get in over its head and I expect that its code decays over time, in a fashion similar to how many human teams get in a state where they just have to rewrite everything. But faster, I'd expect.
But there are a lot of unserious projects that previously would have never been created. E.g., a quick app to manage your little league team, or a bit of in-house business stuff in the "a little hard to do with a spreadsheet" range.
> I honestly don't know what the point would be to let the agents compose worse code than what I'd do manually
You never generate throwaway code used to test an external service? or try out an interface idea? There's a lot of code that's only meant to be ran once. I often dont even care what language it's written in.
> You never generate throwaway code used to test an external service? or try out an interface idea?
And save/persist it? No, most of any experimental stuff goes into /tmp which gets cleared out on reboot, nothing I care to save in any repository. Or just "show me how this would look like" and then it's only in the session itself (and the logs/state I suppose, technically...).
I think they mean it like this: imagine you ask me a random number sequence. I give you a random number sequence. Little did you know, I used a very specific PRNG to generate it, so later I can prove with certainty that your number was generated by me, and you can't say you came up with it yourself.
There is no room for false positive here in the same way you can't randomly find a collision in a hash function if it's strong enough. Like the rate is so infinitesimal that it is effectively zero.
Now replace random number sequence with prompted string of words. And instead of using the PRNG on every word I use it every n words. If the generated text is sufficiently long I can tell by matching the expected deterministic pattern.
You can defeat it by changing the words yourself and triggering a false negative but there isn't really any room for a false positive if the text is long enough and the pattern matches perfectly. If the pattern doesn't match then I can compute a probability.
Sorry you're getting downvoted, this interpretation doesn't seem that far fetched to me.
Here's the strawman: The text-based watermarking is going to be done procedurally instead of generatively. Maybe they add some sequence of zero-width Unicode characters to all generated text at certain intervals. Then, there is effectively no false positive possible (because humans would [effectively] never type such sequences of unicode naturally). It may survive some editing (depending on how you select/edit the characters), and it's possible to be stripped (false negatives).
You can access streaming (not torrenting) sites without a VPN and just watch whatever movies you want and there's no hastle or malware? If you like something particularly you can download it and then storage becomes a factor, and if all streaming sites are blocked a VPN subscription (reputable just mulvad) costs less than any streaming service and gives you access to every single movie and show, impossible to debate that it's not more convenient
reply