Hacker News Reader

Comment by ArtTimeInvestor | Hacker News Reader

It is a step into the right direction.

Over time, more and more work is going to be done by AI though. At some point, it will be unthinkably slow and expensive to let humans work on anything.

To do *that* locally, you need GPUs and LLMs.

How will Europe solve these two?

View on HN

The EU chips act is subsidizing new fab construction in Europe.

Meanwhile the french Mistral is partnering with Nvidia to build an AI data center near Paris on which their LLMs will run.

But I agree this is not enough to make the EU a contender in the race with the US and China. The EU still has not seriously considered decoupling from American big tech.

Do you people have to squeeze a comment about AI into every post?

Not all AI uses LLM, and for some common LLM applications like summarization and translation you can already use CPU only models. The government, or even your average employer, is not going to need a lot of AI video generation or other really GPU intensive tasks. Prompt processing is currently more GPU oriented, but I don't see it as an impossible challenge given, say, 10-15 years.

Also, CPU-only doesn't necessarily mean "on your own computer". You can easily have 100 TB RAM in a couple of racks.

I think it depends on how strong the compression advancements are going to be, such that much can be done locally in the future. I'd be interested in experiences of others here in using Gemma4, which is at the forefront of "intelligence per gigabyte" atm. (according to benches).

No-one needs LLMs.

AI has no value.

The chariot was superior! Who needs them darn cars

My grandpa was doing just fine before those newfangled chariots became all the rage. What's wrong with walking?

I, for one, have never needed AI for anything ever in my life.

AI has, however, made my life noticably worse. Especially when dealing with braindead robot driven customer "support". But also in making it financially impossible to buy more RAM or upgrade a GPU.

I think we'd be better off without yet another bubble.

Were you born yesterday? Phone AIs being dumb didn't take LLMs at all. They were always stupid and frustrating to deal with substitutes for customer support.

Except now they also pretend to respond to email. Basically email support became /dev/null with cream and icing on top.

At this point in the broader dialogue your position is roughly as interesting as flat earth. Only bored people are going to bother replying and no one is taking you seriously. Don't do yourself a disservice by clinging to this.

Okay, give me one example of what AI might be useful for.

As a learning tool to quiz you.

Okay, and what value would that provide?

I'm not interested in games.

The value is that you can have an effective tool to learn something new. I'm not quite sure I understand your question.

I think your wasting your time arguing with him bro

Oh I know :)

Okay, I don't think it would be all that effective and I don't see how it could be.

I learn things by doing them, not by playing guessing games.

That's fine. We all learn different. But it's still effective for other people :) So it's useful for them.

Fair enough I guess. Everything I've seen that's presented as something great about AI just looks like something I'd pay quite a lot of money to avoid.

Im skeptical of the AGI claims but this is a bit too far in the toher direction. I use it to turn designs to code all the time

> I use it to turn designs to code all the time

A human can however do the same job. Turning designs into code isn't a fundamentally new capability unlocked by GenAI, it's just a shuffling of costs from employing humans -> renting GPUs

Turning designs into code is the bit I enjoy. I write all my best code when I'm driving my car. It's great - no-one talking to me, no-one phoning me, nothing to distract me, I can just sit and think about stuff.

Show me something that will somehow turn the code in my head into code on a computer, and I'll be interested.

Reply on news.ycombinator.com