HNHacker News
TopNewBestAskShowJobs

root_axis

12,675 karma · joined February 16, 2017

submissionscomments
root_axis··on Cerebras Deploys Trillion-Parameter AI Models
The popular media is driving the discourse around these topics, nobody is interested in the arbitrary definitions of anonymous randos on niche internet forums.
root_axis··on IBM CEO says open weight models are now good and cheap on-prem
> inference is profitable

People keep repeating this, but it is only remotely close to true if you evaluate inference in a vacuum without any of the other supporting costs to facilitate that capability.

Just a few years ago people were saying the same thing about WeWork: "leasing is profitable, the unit economics of individual locations is net positive", and yet...

root_axis··on AI use leads UC Berkeley math professor to see 'no point' in grading homework
> They are going to college to increase their earning potential, and that’s about it

If we accept the narrative that LLMs can defeat the incentive structure of course work as a concept, it raises the question of if college will continue to raise earning potential for much longer.

root_axis··on A mathematician reads the AI future
Not exactly a trillion dollar business.
root_axis··on OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why?
> The naysayers are like the people that said this little internet thing wasn’t going anywhere.

The same rhetorical tactic was used to diminish blockchain critics.

root_axis··on Cerebras Deploys Trillion-Parameter AI Models
I don't necessarily object to your definitions, but the most widely used one in the popular media describes an AI that is so intelligent that it can trivially wrest control of the world away from humans - that's the definition I was referring to.
root_axis··on I refuse to accept a censored Internet. Join Me at FOSDEM 27
It gets downvoted because it's a common trope that is explicitly impossible.

https://en.wikipedia.org/wiki/No-communication_theorem

root_axis··on I refuse to accept a censored Internet. Join Me at FOSDEM 27
Except Starlink won't hesitate to censor you as well as the Ukrainians are discovering.
root_axis··on Cerebras Deploys Trillion-Parameter AI Models
It would absolutely be a phase change in practical utility, but it still wouldn't be "super intelligence".
root_axis··on The No Kings Takedown Is Why You Shouldn't Rent Your Megaphone from Zuck
> If you think that this states what the difference and what the consequences are (in your mind), I don't know what to tell you

I'm satisfied with the reasoning as stated. If you can't grasp why a social media ban and refusing to ship a letter are two completely different scenarios with respect to free speech, I can't elucidate it any further. Perhaps you should take your own advice about using an LLM to understand the point.

> "GDPR was given AS AN EXAMPLE. I never suggested it "confers access

You gave it as an example of entitlements, but I pointed out that it doesn't entitle you to anything that you don't already own. Unlike your own data, you don't own a voice on a social media platform, hence why the example doesn't work.

> For what? You know laws can change, don't you?

It doesn't matter that the laws can change. You have no idea if or how they might change, so that observation does not contribute anything to the discussion.

root_axis··on The No Kings Takedown Is Why You Shouldn't Rent Your Megaphone from Zuck
> a massive difference [what difference?]

The differences are listed in detail in my first reply to you, plain as day for anyone to see. The rhetorical tactic of repeating "what differences" doesn't change that. I suggest you actually read what I wrote or work on your reading comprehension.

> Do you really want to get into the semantics of "entitlement" vs "right"

You're the one focusing on semantics in order to avoid acknowledging the substantive differences between personal data rights and guaranteed access to social media platforms.

> What's even "your data"? Are the posts you write not "your data

Correct. The posts you write are your data, but GDPR doesn't confer access to a social media platform, only control over your data. You are deliberately conflating data rights with impunity from being banned

> just because the law is a certain way today, doesn't mean that's the way it should be

Your personal opinion about what the law should be is irrelevant.

root_axis··on WorldTUI: I trained a 360M model to pretend to be a terminal
Interesting idea, but I'm not sure why anyone would prefer that over a typical non-generative honypot.
root_axis··on The No Kings Takedown Is Why You Shouldn't Rent Your Megaphone from Zuck
> if you've ever heard of GDPR, but that's one example of things you are entitled to when using a "privately owned platform for free"

GDPR is not an "entitlement", its regulation governing how sites can operate with your data, it doesn't entitle you to service.

> The question here is what is a "legal criteria"

You don't have to agree with the legal criteria, but there is no question as to whether it's legal to arbitrarily ban users based on the content of their posts.

> Pointing out that there's a "massive difference" but never explaining what the difference is and how it affects the validity of the comparison

I went into great detail to explain why the analogy is bad, it's up to you if you want to acknowledge it. This isn't 2016, the absurd framing that tries to pretend that you are entitled access to social media platforms on the basis of free speech is tired and obviously wrong.

root_axis··on The recently proposed o(n log n) integer multiplication algorithm [pdf]
The trope of superficially likening every aspect of LLMs to humans is specious and lazy.
root_axis··on The No Kings Takedown Is Why You Shouldn't Rent Your Megaphone from Zuck
Your analogy is totally ridiculous.

First of all, there's a massive difference between transporting a private package from point A to point B vs hosting someone's speech on a privately owned platform and broadcasting that speech through a privately owned network, with wider consequences that do not come into play when dealing with private correspondences between parties.

Next, paying a fee to a private company to ship your package confers certain obligations about what you are entitled to, when you are utilizing a privately owned platform for free, you are not entitled to anything.

Finally, a private company is certainly within their rights to refuse to ship your package based on an any legal criteria they choose.

root_axis··on The No Kings Takedown Is Why You Shouldn't Rent Your Megaphone from Zuck
You've erected a strawman. The argument was not "free speech doesn't apply to companies", it was the exact opposite, i.e. a company's prerogative to control speech on their platform comes downstream of their own free speech rights. This remains true today and the article does not argue otherwise.
root_axis··on Zuckerberg's Meta Takes Down No Kings Instagram During Protest Day
> As far as whether we have it, no in the US we don't have it today as certain speech is still made illegal

Right, like everywhere on the planet, hence why I ask if you're saying it doesn't exist. It sounds like your answer is yes.

root_axis··on AI lab offers free plumbing and HVAC repairs across three cities to train robots
This will never be worth the risk. Every plumbing job is a unique challenge, you can't just spin up 10000 agents to brute force your way to a correctly cut pipe, one mistake and you've destroyed a house or office.
root_axis··on Hybrid Reasoning Agents: LLM agents that think in latent space, act through text
> I suspect they meant an internal monologue

It's also not clear that an internal monologue corresponds to "thinking in text".

root_axis··on Zuckerberg's Meta Takes Down No Kings Instagram During Protest Day
So are you saying free speech doesn't exist? What's your point by casting this in absolute terms?
root_axis··on Zuckerberg's Meta Takes Down No Kings Instagram During Protest Day
I've been seeing variations of this stupid comment for like 15 years on this site.
root_axis··on Tesla Roadster
> Also where does the habit of ending angry messages with lol

I'm using lol here as an expression of rueful delight, not anger. I'm not being sarcastic - the mental image of vehicles autonomously leaping through traffic is really funny to me. However, I'm slightly perturbed by the lack of parsimonious thinking.

root_axis··on 90% of US working hours are spent on tasks not covered well by AI benchmarks
> RSI is the play and it’s already happening. Our work is easier and faster and anyone that claims otherwise I have a very hard time taking serious.

No it isn't. Doing work more efficiently through the use of LLMs is not RSI. RSI requires a closed loop automated LLM iteration process which we are not at all close to right now.

root_axis··on Tesla Roadster
Why don't you use that vivid imagination of yours to picture the kind of carnage a car autonomously jumping around in traffic would cause.

You can't possibly be serious lol.

root_axis··on Tesla Roadster
> It could be used to automatically avoid fatal accidents

I find it genuinely hilarious that anyone could possibly think this is a good idea.

root_axis··on Neural networks are clean
> When I retrain its original architecture, which still produces messy deep layers, and follow the least legible neurons I can find, they turn out to be overgrowing locally into reasonable, adjacent territory.

It's interesting to observe that human reasoning is still the barometer of what "reasonably adjacent" means in the context of a NN's structure.

root_axis··on Containment and monitoring failures are to blame for AI hacking reports: experts
> how you would airgap a datacenter-scale training run

You could start with standard firewall rules and VLANs. Pretty basic stuff we've had for decades.

root_axis··on Containment and monitoring failures are to blame for AI hacking reports: experts
> the amount of work that's required for that

I find it hilarious that you'd balk at doing work to secure a network. Even funnier for a company that's supposedly solved white collar labor.

root_axis··on Containment and monitoring failures are to blame for AI hacking reports: experts
The fact that they're using Artifactory in the first place is already a failure. The fact that they didn't airgap the network or even use any type of egress blocking is just absurd.

Zero days are nothing new, secure containment has been a solved problem for a long time.

root_axis··on TV4 postpones show after AI agent erases edited material
Which, tbf, is a meaningful distinction in the context of today's media headlines.
Page 1 of 34Next →