Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror

Comment Re:Entirely mechanical (Score 1) 199

Erm no, because humans reason, i.e. feed scenarios into their thought processes and evaluate outcomes. And they are affected by a greater manner of inputs and a wide scope of context than just some sentence. An LLM is basically a crank handle - same input token == exact same output token. LLMs attempt to mitigate with randomization of output (e.g. picking a token randomly based on statistical likelihood) but it's a simulacrum, nothing more.

Comment Re:Entirely mechanical (Score 1) 199

They really aren't doing more than I said. LLMs are trained on data in a way that given any set of input tokens, deterministically it will produce the exact same set of outputs. To mix things up, models will use a "temperature" parameter that will randomly select the next token from the list of most likely outputs so it appears more random than it is otherwise. If the temperature is too high, or the model is insufficiently trained, the response is garbage. If the temperature is too low the response is boring and the same each time.

More modern LLMs might also have callbacks to allow the implementer to inject context into the response, but I'm talking about the general mechanics of what is going on.

Comment Entirely mechanical (Score 5, Insightful) 199

Almost every LLM works like this - given this series of input tokens, give me a list of the potential tokens, choose one with some random weighting, append to list of input tokens, rinse, repeat. With sufficient "parameters", or nodes trained on a sufficient body of input it makes the AI look like it is generating meaningful output whereas it's practically a mechanical process.

Comment Great Wall Motors (Score 1) 62

GWM released the Ora Funky Cat car in Europe and it failed spectacularly. I would not be surprised if GWM has been doing a little price dumping of its own to get rid of them - selling them to leasing firms and the like at rock bottom prices to clear inventory. The car itself isn't awful, but its just not as good as the competition and it cost too much money for what it was.

Comment I really enjoyed the first one (Score 1) 22

Cyberpunk certainly had plenty of bugs and performance issues when it launched. But it was still a staggeringly large open world sandbox style game filled with decent content where you could spend hours doing significant side missions. When I was playing it I could feel the effort that went into the world building, story, and content unlike (for example) any Ubisoft sandbox game and I enjoyed it.

As for bugs, luckily CD Projekt provided highly significant updates that fixed them and also reworked things like police response, upgrade paths and other functionality. My only main criticism of the game and/or GOG.com, is when these updates arrived they were enormous - 20-30GB files. Whatever happened to the idea of binary patches?

So if there were a sequel I wouldn't have any qualms about buying it. But I would wait to see what the system & download requirements are first. And pretty please, learn to incrementally patch the game and don't make downloads unnecessarily large, e.g. by forcing people who don't need ultra quality textures to download them anyway.

Comment No it wouldn't quite obviously (Score 1) 240

If AI companies had to ask permission they can seek it from the artists or their estates, or they can seek it from bodies representing artists, or they can wait for the content to enter the public domain. Same goes too for other content they plunder. That might involve paying money (shock horror) or providing attribution, links, references or other relevant information in the response. It should require affirmative consent, or a transparent revocable process which requires an AI company to make a best effort to obtain consent. But would it kill the AI industry? No it wouldn't. I'm sure these assholes could even write AI tools to help them deal with simplifying the effort for themselves.

Comment Quite obviously yes (Score 3, Interesting) 92

Short of legislation companies will never, EVER, change their behaviour. The EU has such laws like GDPR and ePrivacy precisely to force companies to obtain user consent for data gathering and limit their ability to store and collect it except for the purposes intended. Consumers can also demand their data and the right to be forgotten. And there are heavy fines for companies that flout the rules.

There is no way that the likes social networks would do this shit otherwise that's for sure.

Comment Re:Write once, runs everywhere painfully (Score 1) 100

I can't say I've had many issues running Java on one platform vs another that wasn't caused directly by some assumption in the code itself, e.g. file paths. I suppose AWT was an issue back in the day but Swing is portable. So generally I take a JAR file built wherever and run it wherever and the expectation is it just works.

That isn't to say Java is faultless because tweaking heap and GC can be a pain in the ass and the language itself feels its age especially compared to more modern languages.

Comment Well this sounds stupid (Score 0) 46

People have screens to watch videos, take videos, browse. Remove the screen and you've just emasculated the device. And that's before considering the horrific privacy implications of allowing something fundamentally awful prying into your personal life, your emails and everything else.

Slashdot Top Deals

The price one pays for pursuing any profession, or calling, is an intimate knowledge of its ugly side. -- James Baldwin

Working...