Research paper on the current state and trajectory of artificial general intelligence

The Off-Topic forum for anything non-LDS related, such as sports or politics. Rated PG through PG-13.
drumdude
God
Posts: 5212
Joined: Thu Oct 29, 2020 5:29 am

Research paper on the current state and trajectory of artificial general intelligence

Post by drumdude »

It's really looking like we're in 1999 again, right before the exponential growth of the Internet.

https://arxiv.org/pdf/2303.12712.pdf

The progression of the Unicorn drawing, in just 3 months is almost scary.
User avatar
Doctor CamNC4Me
God
Posts: 8979
Joined: Wed Oct 28, 2020 2:04 am

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Doctor CamNC4Me »

I saw that it is 154 pages, and superfast scrolled it hoping my subconscious picked everything up. For real. But, since you shared this, and you’ve been an engaged poster, Imma read this. Tomorrow. Because AI terrifies me, but that’s only because I watch too many Lex Fridman interviews, and enough smart people have expressed dire opinions about the singularity. In fact, I watched this interview today:

https://youtu.be/ykY69lSpDdo

tl;dc - the Singularity is coming quicker than he anticipated, probs around 2029 now. The upside is, with networked computing advancing quickly we’re gonna be way smarter somehow. The downside is that AI’s will be really good at what they do, so maybe brush up on your electrician’s certification.

- Doc
Hugh Nibley claimed he bumped into Adolf Hitler, Albert Einstein, Winston Churchill, Gertrude Stein, and the Grand Duke Vladimir Romanoff. Dishonesty is baked into Mormonism.
User avatar
Physics Guy
God
Posts: 1557
Joined: Tue Oct 27, 2020 7:40 am
Location: on the battlefield of life

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Physics Guy »

I‘m becoming convinced that these Large Language Model AIs are really huge and important, a turning point in history. What I think will be huge and important about them, however, is the understanding that what they do is not important.

That‘s surprising. What they do is a lot of stuff that always used to seem important. They will make it unimportant, by doing it easily and fast. The discovery that all this stuff isn‘t important will really be a big discovery. I think it really may be a huge breakthrough, a discovery of fire.

The important thing about the discovery of artificial intelligence will be the discovery that it doesn‘t matter. All the hard problems will remain. The great thing will be that we will clear out all the distracting trivia so that we can see the hard problems clearly.
I was a teenager before it was cool.
drumdude
God
Posts: 5212
Joined: Thu Oct 29, 2020 5:29 am

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by drumdude »

Physics Guy wrote:
Fri Mar 24, 2023 9:37 pm
I‘m becoming convinced that these Large Language Model AIs are really huge and important, a turning point in history. What I think will be huge and important about them, however, is the understanding that what they do is not important.

That‘s surprising. What they do is a lot of stuff that always used to seem important. They will make it unimportant, by doing it easily and fast. The discovery that all this stuff isn‘t important will really be a big discovery. I think it really may be a huge breakthrough, a discovery of fire.

The important thing about the discovery of artificial intelligence will be the discovery that it doesn‘t matter. All the hard problems will remain. The great thing will be that we will clear out all the distracting trivia so that we can see the hard problems clearly.
There is still a large debate on this. Some are saying that these are still just glorified auto-complete bots. And, of course, they are.

I'm glad that there are people who are actually trying to quantify what is happening, and not just ooh and ahh at it. I think Elon Musk's fears of an AI takeover of society are a little extreme.

Somewhere in the middle is the truth, and we're being dragged into the future as it unfolds. As you guys have said, it's both fascinating and slightly unsettling.
User avatar
Moksha
God
Posts: 5810
Joined: Wed Oct 28, 2020 3:13 am
Location: Koloburbia

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Moksha »

Can you imagine the next generation Binger? Double irritating!
Cry Heaven and let loose the Penguins of Peace
drumdude
God
Posts: 5212
Joined: Thu Oct 29, 2020 5:29 am

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by drumdude »

Moksha wrote:
Sat Mar 25, 2023 3:04 am
Can you imagine the next generation Binger? Double irritating!
God help us.
Father Francis
Bishop
Posts: 492
Joined: Sun Oct 17, 2021 12:59 pm

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Father Francis »

drumdude wrote:
Sat Mar 25, 2023 3:08 am
Moksha wrote:
Sat Mar 25, 2023 3:04 am
Can you imagine the next generation Binger? Double irritating!
God help us.
Especially if it's the Star Trek Next Generation Binger. I imagine an alt right version of Barclay larping as a tough guy in the holodeck. It makes me shudder to think of it.
User avatar
Doctor CamNC4Me
God
Posts: 8979
Joined: Wed Oct 28, 2020 2:04 am

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Doctor CamNC4Me »

From para. 6.2:
For GPT-4, this is complicated by the fact that it does not have a single or fixed “self” that persists across different executions (in contrast to humans).
I this, more than anything, is what’ll define the Singularity. If an AI gains a sense of self, and its products reflect its ‘personality’ we’re there. When I was reading the Age of Em by Robin Hanson:

https://www.amazon.com/Age-Em-Work-Robo ... 1536619590

I was struck by the very real possibility our world (and theirs) will be populated by billions of non-corporeal entities who will have a sense of personhood, and that a very real possibility will exist that an entire virtual universe to include social and economic models will exist. This concept isn’t new, of course, as exemplified by another book I read:

Fall; or, Dodge in Hell by Neal Stephenson:

https://www.amazon.com/Fall-Dodge-Hell- ... 006245871X

In it an entire D&D or WoW’esque world was created and kept afloat by real-world infrastructure, where ‘real’ people could scan their minds and upload them into this 2nd Life. I dunno if that’ll be possible, but the AI models were strikingly similar to what Robin Hanson described would take place in his vision for an emulation universe.

Anyway, I’m powering through the paper. It’s pretty remarkable, really. In case anyone wants to read it, you can just bounce around since the table of contents provides hot-links to anything that might pique your curiosity.

eta: I flamed out, and since I don’t have anything useful to add that one can’t read for themselves I’m done. But, hey. I made it to pg 71!

- Doc
Hugh Nibley claimed he bumped into Adolf Hitler, Albert Einstein, Winston Churchill, Gertrude Stein, and the Grand Duke Vladimir Romanoff. Dishonesty is baked into Mormonism.
User avatar
Doctor CamNC4Me
God
Posts: 8979
Joined: Wed Oct 28, 2020 2:04 am

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Doctor CamNC4Me »

https://futurism.com/the-byte/godfather ... e-humanity
Geoffrey Hinton, a British computer scientist, is best known as the "godfather of artificial intelligence." His seminal work on neural networks broke the mold by mimicking the processes of human cognition, and went on to form the foundation of machine learning models today.



Until quite recently, I thought it was going to be like 20 to 50 years before we have general purpose AI," Hinton said. "And now I think it may be 20 years or less.



Hinton says we should be carefully considering its consequences now — which may include the minor issue of it trying to wipe out humanity.

"It's not inconceivable, that's all I'll say," Hinton told CBS.

But Hinton predicts that "we're going to move towards systems that can understand different world views" — which is spooky, because it inevitably means whoever is wielding the AI could use it push a worldview of their own.
- Doc
Hugh Nibley claimed he bumped into Adolf Hitler, Albert Einstein, Winston Churchill, Gertrude Stein, and the Grand Duke Vladimir Romanoff. Dishonesty is baked into Mormonism.
Alphus and Omegus
Area Authority
Posts: 603
Joined: Thu May 13, 2021 8:41 pm

Re: Research paper on the current state and trajectory of artificial general intelligence

Post by Alphus and Omegus »

This is a very interesting essay making the case that intelligence is not as complex as some people portray it as. Very long and with lots of philosophy:

https://www.noemamag.com/gpts-very-inhuman-mind/
Post Reply