Connor Leahy on AI Progress, Chimps, Memes, and Markets
Vložit
- čas přidán 9. 07. 2024
- Connor Leahy from Conjecture joins the podcast to discuss AI progress, chimps, memes, and markets. Learn more about Connor's work at conjecture.dev.
Timestamps:
00:00 Introduction
01:00 Defining artificial general intelligence
04:52 What makes humans more powerful than chimps?
17:23 Would AIs have to be social to be intelligent?
20:29 Importing humanity's memes into AIs
23:07 How do we measure progress in AI?
42:39 Gut feelings about AI progress
47:29 Connor's predictions about AGI
52:44 Is predicting AGI soon betting against the market?
57:43 How accurate are prediction markets about AGI?
Social Media Links:
➡️ WEBSITE: futureoflife.org
➡️ TWITTER: / flixrisk
➡️ INSTAGRAM: / futureoflifeinstitute
➡️ META: / futureoflifeinstitute
➡️ LINKEDIN: / future-of-life-institute - Věda a technologie
Fantastic conversation - I could listen to Connor all day and night - amazing.
the most understandable perspective on GAI I ve heard yet. keep up the great content.. superchimps unite!
Well-deserving of three segments🤤
Ah, the glorious digital realm of CZcams, where Connor Leahy enlightens us with his wisdom on AI progress, chimps, memes, and markets. It's not every day you stumble upon a video that manages to conjure a delightful concoction of such diverse topics, and I must say, I'm absolutely intrigued!
While I'm sure our ancestors would have never dreamt of discussing chimpanzees and artificial intelligence in the same sentence, here we are, standing on the precipice of technological advancements as we still ponder over our evolutionary roots. One can almost say that Connor is the veritable Prometheus, delivering the fire of knowledge to both the AI community and chimp enthusiasts alike (a Venn diagram I'd love to see, by the way).
Now, Connor must be a true connoisseur of memes to weave them so seamlessly into his discourse. I presume he understands the profound cultural impact of memes, a true Zeitgeist of our era, transcending national borders and tickling our funny bones while carrying potent social messages. Could AI-generated meme warfare become the new race to the moon? Only time will tell.
Ah, and lastly, we plunge into the world of markets, an enigmatic entity shaped by human decision-making, trends, and the seemingly omnipotent invisible hand. Connor's exploration into this realm only enriches our understanding of the complex interplay between technological progress and the economic undercurrents that pave our collective destiny.
To end my musings on a humorous note, the versatility of this video sparks visions of a future where sophisticated AI-powered chimpanzees, adept in meme warfare, dominate global markets, effectively becoming the puppet masters of human civilization! And what could be more entertaining than picturing the Planet of the Apes reboot that would inevitably follow? Hail Caesar, the meme lord! 😂
AND POST!
Incredible comment and insight. At first, I thought it was generated with ChatGPT. It will be difficult to tell at this point.
Is there a part 2? The ending was quite abrupt.
there is
This guy is at the bleeding age of AI research. I would pay great atention to his thoughts. He has the finger on the pulse of what's happening.
51:48 Back in Nov 2020 he said that gpt 3 was already AGI and was as smart as human beings, I wish he would clarify
A snake biting it's own tail since, knowledge is power.
So what Connor (and others) are suggesting with the "pause", "safety" and "alignment" endeavour basically means, that it should end with savant slaves.
In what way do you envision an ultra-knowledgeable system NOT to be functionally identical to a human regarding allegiance, morals, ethics and so on? According to Connor this must be exactly how our own models of other agents form. Then the question should be: Well, did you raise it in a way that it will feel as loving offspring of humankind?
Although I am highly impressed with Connor's sober attitude towards AGI (and entirely agree regarding the set level of context): Is the whole question about "safety" and "alignment" not completely void, when thinking about hard wired approaches?
Brilliant
rather mindblowing stuff here
i love hearing the distant whipsers of terence mckennas theories readapted/rediscovered. so cool.
The mushroom theory? Makes sense
@@cr-nd8qh yeah and novelty theory, timewave zero, the eschaton, singularity, etc
Just a thought, with the embodiment of AI (humanoid robots) along with mimetic evolution of humanity,could lead to a more social AI . more aligned with humanity.🤔IMO
Yeah, if you think _H. sapiens_ is uniquely violent, I would invite you to learn more about chimps, and to imagine a world with 8 billion chimps (perish the thought).
But I think there is a rather profound difference in raw cognitive horsepower. I strongly suspect you could ask most any 6-year-old human child to improve on a chimp spear design (some chimps hunt bushbabies with shitty spears), and the human child could improve it significantly in a very short period of time -- no research required, just by looking at it, seeing how shitty it is, and understanding in principle how it could be better.
Orangutans outclass chimps in _some_ ways. The nests they build, and their little rain umbrellas, are surprisingly sophisticated. And they watch a human do something and can imitate it perhaps somewhat more quickly than a chimp.
Also, Connor is super wrong about early _H. sapiens_ only having hand axes. Of course they had wooden spears -- chimps make those. Stuff made out of wood doesn't preserve very well. And hafted stone spear points were certainly used as long as 200k years ago, and possibly as early as 500k, with good evidence for their use 460k years ago.
Could you get me some of them future girls how about some of that future food I'd really like to go to one of those future parties it reminds me I need to purchase that singularity book so I can read it in the now. And shave that guys goatee. He's starting to look like Salvador Dali.
There are "signs" we may be able to tell about intelligence 😉.
His take on GDP is wrong. The value of open source resources like Wikipedia IS captured in the GDP implicitly by the productivity enhancement it adds to everything society does that IS captured in the GDP.
A world in which wikipedia was monetized, maybe with side ads, would likely have slightly higher GDP, no? Yet it would be a slightly shittier world. I think that's the point he was trying to get to. And also, there is a point to be made that implicit might not be good enough, can lead one to conclusions like: Windows has been far more impactful than linux, or matlab moreso than python, as the commercial examples will show a direct contribution to GDP while the open source ones will not.
I think you're missing the point which is that it's not directly captured in the GDP. You're right though it would increase the GDP indirectly.
But it's not hard to imagine a productive society built upon entirely free open source software, some free favours, some free goods, and otherwise self-reliance, which has 0 GDP.
It really only makes perfect sense as a measure in a free market libertarian society where everything is owned and monetized. Otherwise you can juice the GDP easily by exchanging things at below market value, or way above. If I give you a rock, you blow on it make a wish and you give it back to me for twice the price, and we keep doing that until we get to billions of dollars, we've created GDP out of nothing. Yet it was never fairly valued, so we distorted the metric.