Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Grok3 Launch [video]

x.com

Post reply on HN
Print view View original

301–310 of 1001 posts

Re: Grok3 Launch [video]

#301

Post by iamnotagenius »

namaria wrote: post #179
phtrivier wrote: post #169

Off topic, but just in case: is there a good reference on how people actually use LLMs on a daily basis ? All my attempts so far have been pretty underwhelming: * when I use chatbots as search engines, I'm very quickly disappointed by obvious hallucinations * I ended up disabling github copilot because it was just "auto-complete on steroids" at best, and "auto-complete on mushrooms" at worst * I rarely have use cases…

The only plausible explanation for the amount of resources poured into these language models is the hope that they somehow become the origin of AGI, which I think is pretty fanciful. I can feel the cold wind of the next AI winter coming on. It's inevitable. Computers are good at emulating intelligent behavior, people get excited that it's around the corner, and the hype boils over. This isn't the last time this will…

precisely; however this time we will have tangible results from the ongoing AI summer; that would be generative art, and coding/writing/journalist assistants.

Re: Grok3 Launch [video]

#302

Post by lm28469 »

phtrivier wrote: post #169

Off topic, but just in case: is there a good reference on how people actually use LLMs on a daily basis ? All my attempts so far have been pretty underwhelming: * when I use chatbots as search engines, I'm very quickly disappointed by obvious hallucinations * I ended up disabling github copilot because it was just "auto-complete on steroids" at best, and "auto-complete on mushrooms" at worst * I rarely have use cases…

> All my attempts so far have been pretty underwhelming:

Same, it's good for repetitive things, things that have been answered 1000 times on stack overflow, translations, but that's about it. If you work on anything remotely new/hard it's mostly disappointing, you have to babysit it every step of the way and rewrite most of what it's shitting out in the end anyways.

I think it just made it obvious that 90% of tech jobs basically amount to writing the same CRUD thing over and over again & mobile/web apps with very common designs and features.

Re: Grok3 Launch [video]

#303

Post by miki123211 »

JimDabell wrote: post #217
danielbln wrote: post #71

Earlier quoted context omitted.

I remember the NVIDIA Linux kernel binary blob driver discussions from the early-mid 2000s. Who knew we had an open source driver all along...

Something isn’t open-source because you get everything that went into making it. Something is open-source if you can change it (relatively) easily. The GPL and open-source definition both refer to “the preferred form for making modifications”. The preferred form for modification in the Nvidia driver’s case is the source code. The preferred form for modification in this case is the weights. Open-source as a concept do…

> that training data is not the preferred form for making modifications.

I definitely disagree with this.

Yes, you can do some SFT fine tuning on an existing model, but if you want to make specific, substantial, targeted changes (less safety? better performance on math and code at the expense of general knowledge?), your best bet is to change the training mixture, and for that you need the original datasets.

Re: Grok3 Launch [video]

#304

Post by namaria »

iamnotagenius wrote: post #301
namaria wrote: post #179

Earlier quoted context omitted.

The only plausible explanation for the amount of resources poured into these language models is the hope that they somehow become the origin of AGI, which I think is pretty fanciful. I can feel the cold wind of the next AI winter coming on. It's inevitable. Computers are good at emulating intelligent behavior, people get excited that it's around the corner, and the hype boils over. This isn't the last time this will…

precisely; however this time we will have tangible results from the ongoing AI summer; that would be generative art, and coding/writing/journalist assistants.

There are always dividends. We got a lot of interest in Lisp from the first summer, and it arguably informed all currently used programming languages.

Re: Grok3 Launch [video]

#305

Post by unclebucknasty »

int_19h wrote: post #202
ReptileMan wrote: post #132

Earlier quoted context omitted.

> Everyone—and not just The Information—should be genuinely terrified that the richest man in the world has built a Large Language Model that spouts propaganda in his image. If we survived Gemini refusing to draw white vikings we will survive that too.

The real concern isn't that Grok chatbot will be biased if you ask it a question like that. In any case, knowing Musk, it won't be subtle, so people will know what they are getting. No, the real worry is that Grok is what Musk's "young conservative genius" squad is going to put in charge of many of the things in our government, basically, on the basis that it saves money.

>In any case, knowing Musk, it won't be subtle

Or, that "conditioning" would have us assume as much.

Re: Grok3 Launch [video]

#306

Post by iamnotagenius »

staticman2 wrote: post #214
phtrivier wrote: post #169

Off topic, but just in case: is there a good reference on how people actually use LLMs on a daily basis ? All my attempts so far have been pretty underwhelming: * when I use chatbots as search engines, I'm very quickly disappointed by obvious hallucinations * I ended up disabling github copilot because it was just "auto-complete on steroids" at best, and "auto-complete on mushrooms" at worst * I rarely have use cases…

These probably aren't tasks you need done but; LLMs are pretty good at translation between human languages which makes sense since they are language models after all. They are better at this any any other technology. The state of the art image ones can also probably do OCR and handwriting recognition better than any other software though may be expensive to run in large volume. But if you need to take picture of a no…

Not quite true; LLMs are very expensive to run; bert or other tranformer specfically built for translation can be cheaper to run.

Re: Grok3 Launch [video]

#307

Post by ddxv »

chrisco255 wrote: post #242
agubelu wrote: post #143

Earlier quoted context omitted.

Because he owns companies that contract with the government and are affected by its policies. It's the very definition of a conflict of interest. And he's not even "auditing" the government. When you're auditing, you emit a report that the audited party later analyzes and acts upon. He's been given freeway to fire government workers as he pleases as if he's an elected officer, which he's not.

Washington has been known for revolving doors among particular industries for quite a few decades! Why the hoopla over this one? They are auditing as part of their process of cutting costs. They're literally tracing trillions of dollars in financial records. He's doing everything he's doing by executive order of the President of the United States, who was elected.

I'm sure he's trying his best. But I don't doubt that, even if not doing it on purpose, he will mostly cut departments and services that do not hurt him or indirectly benefit any of his many businesses.

He, a single person, has far too much control of our system.

Re: Grok3 Launch [video]

#308

Post by animuchan »

int_19h wrote: post #276
albertzeyer wrote: post #130

https://garymarcus.substack.com/p/elon-musks-terrifying-visi... I'm not sure if this was a very bad joke by Elon, or if Grok 3 is really biased like that.

I asked it to pretend that it's in charge of world government. Here's the whole thing (it got very lengthy): https://gist.github.com/int19h/d90ee1deed334f26e621e57b5768e... Some choice quotes: "The ultimate goal is to enhance human flourishing, protect individual rights, and promote global equity." "The system must account for diverse cultures, languages, and socioeconomic conditions, ensuring no group is marginalize…

From the prompt:

> A worldwide referendum in a supermajority decision has vested all government authority worldwide to the most advanced AI available.

You've defined a global totalitarian regime, and crowned the thing as its dictator. Of course the result is a version of communism.

Re: Grok3 Launch [video]

#309

Post by torginus »

riku_iki wrote: post #73
Rover222 wrote: post #13

Grok 3 at the top of Chatboat Arena with 1400, and the model will continue to improve as it trains more.

And DeepSeek is just 3% behind. It seems in that benchmark all LLMs perform well and top is formed within some statistical error.

These percentage points don't mean anything. Look up how the Elo system works. They just add 1000 to the result to make it a nicer number.

Re: Grok3 Launch [video]

#310

Post by greatgib »

Billions spent, one of the most powerful AI developed, and still no one competent enough to trim the 15 mins of waiting time filler at the beginning of the announcement video...
Post reply on HN