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

Google DeepMind CEO says some form of AGI possible in a few years

wsj.com

Post reply on HN
Print view View original

341–350 of 388 posts

Re: Google DeepMind CEO says some form of AGI possible in a few years

#341

Post by pixl97 »

300bps wrote: post #228
deagle50 wrote: post #75

Earlier quoted context omitted.

The directives that mandated eliminating its creators don't necessarily lead to interstellar conquest. Or more likely IMO, we're just too far apart and interstellar travel isn't feasible even for artificial super intelligence.

Here’s a paper that shows how a civilization could colonize the galaxy in under 300 million years: https://arxiv.org/pdf/1902.04450.pdf Here’s a YouTube visualization of the timeline: https://youtu.be/hNMgtRf0GOg

What is the actual rate of intelligences per galaxy, not the simulated one? Galaxies are very far apart from each other.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#342

Post by pixl97 »

valine wrote: post #328
alex_duf wrote: post #319

Earlier quoted context omitted.

That's assuming the AI really wants to make a lot of paperclips though

I'd say a hallmark of an intelligent agent is knowledge seeking. You can only learn so much without venturing out into space.

There is a slight possibility that this is an unintelligent question, with humanity being ignorant as to why that's the case.

The universe we live in is only around 14 billion years old, which under the ideas of how long a universe could be around is actually a tiny immeasurably small blip on eternity. Before our universe started there was a vacuum metastability event somewhere that we at least assume had a nice even entropy. Again, that instability happened pretty damned recently...

So what if the first thing any super intelligence realizes is that at any moment our universe could have another such instability event? It's not going to work on exploring the vast distances of space... It's going to figure out if it can develop a plan to GTFO of here to a plain of reality that's more stable than this one.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#344

Post by pixl97 »

maister wrote: post #182
wg0 wrote: post #44

Earlier quoted context omitted.

> Or maybe folks realize the degree of lies/mistruths inherent in its content is actually unmanageable and can’t be improved. After the hype wears off what it will be used for gets greatly curtailed and we see a big contraction. I think it's going to be this the future that's ahead us. There's enormous faith being put in next token predictors as the intelligence breakthrough just because their output coincidentally r…

> I don't understand the "coincidentally" argument. Nothing is coincidental about those models. They were designed after processes in the brain. They underwent rigorous training to generate a function that probabilistically maps inputs to outputs. Eventually, it exceeded the threshold where most humans consider it to be intelligent. As these models grow larger, they will surpass human intelligence by far. Currently,…

Heh, another one I see

"LLMs don't create anything new" and "LLMs hallucinate all the time"

I want to ask those people which one is the correct sentence as they appear to be in conflict with each other.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#345

Post by Ancalagon »

wg0 wrote: post #13

The other few things possible in next few years for past few years are: 1. Fusion 2. Self driving cars. 3. Ubiquitous, generally available and stable 8192 bit quantum computers. 4. High energy density grid scale batteries. Just few more "this maybe the... what could revolutionize..." articles away. What we forget is that it is in these people's business interest (executives making such statements) to state the exagge…

don't forget room temperature super conductors

Re: Google DeepMind CEO says some form of AGI possible in a few years

#346

Post by IIAOPSW »

throwawayadvsec wrote: post #338
andrepd wrote: post #171

Earlier quoted context omitted.

> ChatGPT does this very well Lol, this is precisely what chatgpt doesn't do well at all! It fails to understand simple concepts, logical inference, simple math, etc.

GPT is actually EXTREMELY good at math compared to an average human who is not using a calculator. Ask a random person stuff like whats 465*42/12 and you'll get ridiculous answers, and LLM can use tools now. I'd argue that it's actually superior to humans in that regard

Ask a person on the street and they will either say "I don't know", "I don't have time for this", or on rare occasion you'll find some nerd who starts juggling numbers outloud eventually reaching some rational terminus (at worst with an error in recall but not in principle along the way).

Re: Google DeepMind CEO says some form of AGI possible in a few years

#347

Post by olddustytrail »

hammyhavoc wrote: post #336
blagie wrote: post #327

Earlier quoted context omitted.

The level of confidence both sides give here is not warranted. We have no idea about what internal structures emerged within LLMs. We only know outer behavior. From a humanist / secular perspective, humans evolved to make babies. Emotions are an emergent behavior to maximize the number of babies made, and their survival. Nothing less, and nothing more. What analogues emerge when we train machines not to survive but t…

Why would anything emerge?

Well, because they do: https://en.wikipedia.org/wiki/Large_language_model#/media/Fi...

If you're asking "Why does anything emerge?", I don't think anyone really knows.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#348

Post by ogogmad »

theGnuMe wrote: post #335
ogogmad wrote: post #247

Earlier quoted context omitted.

That would suggest that 1-Hidden-Layer neural nets would work fine, since they are also universal function approximators. But no -- when people talk about "deep learning", the word "deep" refers to having lots of hidden layers. I'm not an expert, but the motivation seems more like this: - Linear regression and SVM sometimes work. But they apply to very few problems. - We can fit those models using gradient descent. A…

1-layer nets are equivalent in theory and universal approximators, they are just hard to train.

[flagged]

Re: Google DeepMind CEO says some form of AGI possible in a few years

#349

Post by oittaa »

alex_duf wrote: post #322
oittaa wrote: post #203

Earlier quoted context omitted.

This. It surprises me every time when battery or any "green" technology is discussed here and 90% of Americans keep parroting Trump's talking points. For example one of the comments above is talking about how EVs are just "a rich boy's toy".

In fairness EVs still are more expensive than petrol cars. I'm hoping this will change though

On average usage case Tesla Model 3 is already cheaper than Toyota Corolla. Chevy Bolt, while they still sell it, is even cheaper.

inb4: I only buy 20 year old cars for $500 and I want to drive 1000km without stopping while shitting in my pants.

Re: Google DeepMind CEO says some form of AGI possible in a few years

#350

Post by _hypx »

alex_duf wrote: post #321
_hypx wrote: post #51

Earlier quoted context omitted.

Grid scale energy storage is basically solved once you realize that hydrogen storage is an option already. It's in the same boat as when photovoltaics got very cheap, but skeptics continued to deny that possibility for some years afterwards.

Isn't hydrogen extremely hard to store over a long period of time ? (Seasonal)

No. That is its strength.
Post reply on HN