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

531–540 of 1001 posts

Re: Grok3 Launch [video]

#531

Post by theshackleford »

jagermo wrote: post #207
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…

For me, getting summaries of meetings is my favorite use case. Saves me from taking notes and I can extract next steps. It also helps me getting started with new content, kind of building the scaffolding of, say, a blog or social post. It still needs adaption and fine-tuning, but getting rid of a white page is a great help for me. And I use LLMs to play through ideas and headlines. I would normally do this with other…

> although the AI not being able to really give criticism is a bit annoying.

I’ve managed to get ChatGPT to a good place in this regard recently and it’s better for it. In fact, it’s been doing such a good job recently that it almost seems like…human like.

I’ll have to look at my prompts, but somehow I got it from licking my ass and telling me everything I say and do is amazing to a point now where it almost seems eager to tell me I’m wrong.

Which is fantastic, huge improvement. I don’t really use it for coding though, because I am not a programmer. I would have no means today to correctly evaluate 90% of what it would return me.

Re: Grok3 Launch [video]

#532

Post by sanxiyn »

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…

Many people replied with anecdotes, but recently Anthropic published analysis of claude.ai production traffic. As far as I can tell this is the single best currently existing source of "how people actually use LLMs". Everything else you can't be sure whether it is representative.

https://www.anthropic.com/news/the-anthropic-economic-index

Re: Grok3 Launch [video]

#533

Post by jsheard »

moduspol wrote: post #461
Ray20 wrote: post #418

Earlier quoted context omitted.

I think business model there pretty simple: to be in the front line when AI will go into the category of landscape-changing trillion dollar technologies. And investors keep pouring their billions exactly for that business model. >This hype will burst sooner than later and will trigger yet another global recession. It seems to small of bubble for global recession. I mean if it is a bubble at all, there is all the reas…

IMO that's not really a business model. That's a hope that you can come up with one by being at (or near) the front of the pack if one materializes.

See also: Meta's previous push into VR/AR/Metaverse. They spent a hundred billion to be at the front of the pack when that revolutionary world-changing paradigm-shift took off... which simply didn't happen.

Re: Grok3 Launch [video]

#534

Post by jonlucc »

bobxmax wrote: post #483
aprilthird2021 wrote: post #30

Earlier quoted context omitted.

I think it's already clear that these are going to be commoditized and the free / open source versions will be good enough to capture enough of the value that the remaining players will not be Facebook-level monopolies on the space

The vast majority of people couldn't care less about open source

If you're paying $200/month for something I can do with open source software and $10/month of compute, why wouldn't I offer you the service for $100/month? And then someone offer it for $50?

Not everyone has to know about, understand, or use open source solutions for it to open the field.

Re: Grok3 Launch [video]

#535

Post by namaria »

olalonde wrote: post #453
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…

I predict this comment will age very, very poorly. Bookmarked.

Cool. Invest in it then. That way you get paid instead of saying "I told you so" to some screen name.

Re: Grok3 Launch [video]

#536

Post by esafak »

zozbot234 wrote: post #472
nomilk wrote: post #81

Earlier quoted context omitted.

It was mentioned during the launch that current datacenter requires up to 0.25 gigawatts of power. The datacenter they're currently building will require 1.25 (5x) (for reference, a nuclear powerplant might output about 1 gigawatt). Will be interesting to see if the relationship between power/compute/parameters and performance is exponential, logarithmic or something more linear.

> It was mentioned during the launch that current datacenter requires up to 0.25 gigawatts of power. The datacenter they're currently building will require 1.25 (5x) (for reference, a nuclear powerplant might output about 1 gigawatt). IIRC achieving full AGI requires precisely 1.21 jigawatts of power, since that's when the model begins to learn at a geometric rate. But I think I saw this figure mentioned in a really…

The funny part was that none of his workers recognized the film, which was a blockbuster. A veritable "I must be getting old" moment.

Re: Grok3 Launch [video]

#537

Post by coder543 »

numpad0 wrote: post #446
sigmoid10 wrote: post #385

Earlier quoted context omitted.

The impression seems to be warranted: Grok 3 has directly jumpted to the top of all leaderboard categories in Chatbot Arena: https://lmarena.ai/?leaderboard In math it shares the top spot with o1 and is just a few points behind (well within errors). In creative writing it is basically ex-aequo with the latest ChatGPT 4o and in coding it's actually significantly ahead of everyone else and represents a new SOTA.

Considering that OpenAI subscription is $200 per month, and "Premium Plus" subscription that includes this thing is only $40 per month, does that mean instantaneous "Elon factor" is now at least -$160 per month per user, or is it supposed to be added up to more than -$240 per month? How would the math change after factoring in that OpenAI isn't even covering entirety of opex with the sub anyway, and/or people finding…

No... sigmoid10 was comparing with o1 (not o1-pro), which is accessible for $20/mo, not $200/mo. So, the "Elon factor" in your math is +$20/user/month (2x) for barely any difference in performance (a hard sell), not -$160/user/month, and while we have no clear answer to whether either of them are making a profit at that price, it would be surprising if OpenAI Plus users were not profitable, given the reasonable rate limits OpenAI imposes on o1 access, and the fact that most Plus users probably aren't maxing out their rate limits anyways. o1-pro requires vastly more compute than o1 for each query, and OpenAI was providing effectively unlimited access to o1-pro to Pro users, with users who want tons of queries gravitating to that subscription. The combination of those factors is certainly why Sam Altman claimed they weren't making money on Pro users.

lmarena has also become less and less useful over time for comparing frontier models as all frontier models are able to saturate the performance needed for the kind of casual questions typically asked there. For the harder questions, o1 (not even o1-pro) still appears to be tied for 1st place with several other models... which is yet another indication of just how saturated that benchmark is.

Re: Grok3 Launch [video]

#538

Post by wyclif »

butifnot0701 wrote: post #191
lionkor wrote: post #144

I don't understand how and why Grok would be related to "understanding the nature of the universe", as Musk puts it. Please correct me if I'm wrong, but they basically just burned more cash than any human should have to buy Nvidia GPUs and make them predict natural language, right? So, they are somewhat on-par with all the other companies that did the same. This is not innovation, this is baseless hype over a mediocr…

I feel like that's part of what Elon is flexing. Teslabot was late comer compared to competitors like BD. Elon is showing off he can marshal enough resources and talents to be on par (kinda) with state of the art products in crazy time. That's been most of his superpower so far - not breakthrough tech that didnt exist before. We've had rockets before.

We've had rockets before

Yeah, but we didn't have reusable orbital rockets, and that's a distinction with a big difference.

Re: Grok3 Launch [video]

#539

Post by mobiuscog »

sebzim4500 wrote: post #360

Controversial opinion but I think the AI game studio idea is a very good one. Not because I think they will make any money off the games, but dogfooding will lead to so much more improvement than relying on feedback from external customers.

Because 'dogfooding' has worked so well for other products...

If you don't get feedback from the people actually playing your game (or using your product), you will never get the improvement you need to help them.

You can have the most talented passionate people there are developing a product, but if it's not working for the people you want to sell it to, it's the wrong product.

Most tech products are terrible because those paying for them are not those that have to use them every day, or because they solve a corporate problem (compliance) and not a usability problem which is the actual need from the people on the shop floor.

Many big games/products are already built mostly on metrics, and that has proven to be a terrible way to work out what people 'want'. It's a great way to justify money decisions though, so it keeps happening (and games/products from big companies keep getting worse).

Re: Grok3 Launch [video]

#540

Post by voidUpdate »

InsideOutSanta wrote: post #185
lionkor wrote: post #144

I don't understand how and why Grok would be related to "understanding the nature of the universe", as Musk puts it. Please correct me if I'm wrong, but they basically just burned more cash than any human should have to buy Nvidia GPUs and make them predict natural language, right? So, they are somewhat on-par with all the other companies that did the same. This is not innovation, this is baseless hype over a mediocr…

There are two answers to this. Answer 1: Some people think that LLMs are a path to the singularity, a self-improving intelligent program that will vastly exceed human intelligence and will be able to increase its knowledge exponentially, quickly answering all answerable scientific questions. Answer 2: LLM companies need to keep the hype train rolling. I didn't watch the whole clip; I jumped around a bit, but I notice…

I'm pretty sure it doesn't contain the whole of human knowledge... I doubt grok knows my youtube password or my bank PIN
Post reply on HN