Connor Leahy - e/acc, AGI and the future.

Connor is the CEO of Conjecture and one of the most famous names in the AI alignment movement. This is the "behind the scenes footage" and bonus Patreon interviews from the day of the Beff Jezos debate, including an interview with Daniel Clothiaux. It's a great insight into Connor's philosophy.
Support MLST:
Please support us on Patreon. We are entirely funded from Patreon donations right now. Patreon supports get private discord access, biweekly calls, very early-access + exclusive content and lots more.
/ mlst
Donate: www.paypal.com/donate/?hosted...
If you would like to sponsor us, so we can tell your story - reach out on mlstreettalk at gmail
Topics:
Externalized cognition and the role of society and culture in human intelligence
The potential for AI systems to develop agency and autonomy
The future of AGI as a complex mixture of various components
The concept of agency and its relationship to power
The importance of coherence in AI systems
The balance between coherence and variance in exploring potential upsides
The role of dynamic, competent, and incorruptible institutions in handling risks and developing technology
Concerns about AI widening the gap between the haves and have-nots
The concept of equal access to opportunity and maintaining dynamism in the system
Leahy's perspective on life as a process that "rides entropy"
The importance of distinguishing between epistemological, decision-theoretic, and aesthetic aspects of morality (inc ref to Hume's Guillotine)
The concept of continuous agency and the idea that the first AGI will be a messy admixture of various components
The potential for AI systems to become more physically embedded in the future
The challenges of aligning AI systems and the societal impacts of AI technologies like ChatGPT and Bing
The importance of humility in the face of complexity when considering the future of AI and its societal implications
TOC:
00:00:00 Intro
00:00:56 Connor's Philosophy
00:03:53 Office Skit
00:05:08 Connor on e/acc and Beff
00:07:28 Intro to Daniel's Philosophy
00:08:35 Connor on Entropy, Life, and Morality
00:19:10 Connor on London
00:20:21 Connor Office Interview
00:20:46 Friston Patreon Preview
00:21:48 Why Are We So Dumb?
00:23:52 The Voice of the People, the Voice of God / Populism
00:26:35 Mimetics
00:30:03 Governance
00:33:19 Agency
00:40:25 Daniel Interview - Externalised Cognition, Bing GPT, AGI
00:56:29 Beff + Connor Bonus Patreons Interview
Disclaimer: this video is not an endorsement of e/acc or AGI agential existential risk from us - the hosts of MLST consider both of these views to be quite extreme. We seek diverse views on the channel.

[@Kwalk1989]

I literally was like what is MLST up to? I search and the video was 27 seconds old.

[@naninano8813]

is 42:29 worth reading (for someone into mathy part of AI)?

[@itsallgoodaversa]

Love the look on your shots. Which G Master lenses are you using? 50mm? 35?

[@macawism]

Tell it to someone in Gaza or the Ukraine or a Rohingya about a benchmark for values.

[@denisla3546]

u go tell them ,hypothetical loser.. Talk big on internet huh ,DO SOMETHING!

[@rowanwilliams7441]

Sure, but how can any individual change that trajectory quickly enough that more thousand don't die in the meantime. You can't blame Connor or anyone for not being able to tackle that most tragic of problems alone

[@kensho123456]

What is sq(-1) ?

[@alexanderbrown-dg3sy]

Brilliant guy. A lot of these guys need reality checks though. They always speak with a messianic vision, how they want to improve the human condition- yet are completely ignorant to it. The irony. Contrastive views are welcomed and needed. Considering the over-arching consequences of TRUE strong AI. It’s clear though…AI cults and AI terrorism will be a thing unfortunately in the not too distant future. Even as a indie researcher myself. These guys and their views…to me is weird and potentially dangerous. This applies to both sides. Tbh all of it is pure cap. Once we have AGI. If it’s owned by a single private company, the only outcome is a negative one. Again, reality check. Human nature dictates this cap. Prices won’t drop. Profits will be kept. Opinions and beliefs will be iteratively molded. Only solution is democratize ownership/value function formation. Due to compute cost. This is a delusional fantasy. Basically we are F’ed 😂. Can’t convince me otherwise. I was never worried about “rogue AI”, but the proven historical bottleneck “rogue humans”.

[@andybaldman]

If nobody ever wanted to save the world, nothing would progress. Nothing would be invented.

[@m.branson4785]

I was in the middle of writing a paragraph when the video just came on, and Connor's opening comments are so precisely specific to the content of my current writing, it's uncanny. Sean Carrol was just having a conversation earlier with Lex, and he was mentioning there, "We are surfers riding the wave of increasing entropy. We rely on increasing entropy to survive. That is part of what makes us special. This table maintains its stability mechanically, by which I mean there are molecules, they have forces on each other, and it holds up. You and I aren't like that. We maintain our stability dynamically by ingesting fuel - food, water, and air and so forth - and burning it, increasing its entropy. **We are non-equilibrium quasi-steady state systems.** We are using the fuel the universe gives us in the form of low-entropy energy to maintain our own stability."

[@__D10S__]

oddly this is the same for me. not quite as serendipitous, but i've been thinking a lot about these same exact things recently as well.

[@anakissedboyle3067]

Similar, only for a really different reason, trying to get Degrowth Conversations to properly situate Man made entropy into the problem space that is modern capitalism. I e can’t properly enquire about changes until we recognise how ,,,,, But that’s partly an accelerationist problem , but Capitalism is Acceleratinist, and speeding up Entropy on a finite planet is stupid as, High grade silicone is limited, maybe new materials will occur, but our cheap disposable entropy is a disaster

[@jonnysolaris]

Thing is with all the people on the e/acc and adjacent sides is that it hasn't dawned on them yet that we are on the cusp of a different kind of human culture, and it won't be all roses. You need people to balance the other end of the extreme, we are not only talking about engineering out scarcity, we are talking about tech that can be the stuff of nightmares we probably couldn't even dream of today (check out what IDF is doing today to find targets, their drones etc for a taste). Imagine political parties without opposing rivals, or companies without competition - they often become tyrannies, or are the products of tyrannies in the first place. We seriously need opposing points of view.

[@captain_crunk]

I take your point, but I think almost by definition they believe it actually will be all roses.

[@TheBlackClockOfTime]

read more. the whole thing is about a different kind of human culture.

[@jonnysolaris]

@@captain_crunk read more, "won't be all roses", thanks babe

[@captain_crunk]

@@jonnysolaris what do you mean? I'm saying that no one knows what will happen. One side thinks it's all roses, the other side not.

[@Hexanitrobenzene]

@@captain_crunk I don't understand why e/acc people don't get such a simple principle: if you don't know what will happen, proceed with caution. If the consequences can be extremely high, as with AI, proceed with extreme caution. Being certain about what will happen when in principle AI can be smarter than its creators is madness.

[@JD-jl4yy]

51:28 This is the core of the disagreement. Smart people can think about AI and form a wide variety of inside views. They all disagree with each other. What I see the e/acc people doing is arrogantly sticking to their particular inside view, while the existential risk people are more epistemically humble, aggregating predictions from everyone that has rigorously thought about this and build a calibrated outside view. This outside view puts some credence on existential risk, so they (imo correctly) identify that as a very big deal and try to figure out how we could mitigate it. *obviously not all of them, some like Eliezer Yudkowsky also stick to a similarly overconfident inside view **if you agree xrisk is a big deal but disagree about how people are currently going about it, you're still fundamentally on the same side here

[@lancemarchetti8673]

*For those who find themselves on the AGI hype train...* It's been 2 years now since the GPTs went live and they still struggle with simple math instructions that you and I can easily accomplish. To illustrate what I'm talking about. Below is a direct copy and paste from my chat with Meta's "advanced" AI this morning. *User:* I'd like to test your abilities with seemingly mundane human logic. Do you mind? *AI:* Not at all! I'm here to help and improve, and I welcome any opportunity to test my abilities and learn from human logic. Please go ahead and ask me any questions or present scenarios that you think might challenge my understanding or reveal any limitations. I'm ready to learn and improve! *User:* Generate a string that represents the Latin alphabet as used in English worldwide. *AI:* The Latin alphabet as used in English worldwide is: abcdefghijklmnopqrstuvwxyz *User:* Now reverse it *AI:* The reversed string is: zyxwvutsrqponmlkjihgfedcba *User:* Now swap the starting character with the ending character *AI:* If we swap the starting character ("z") with the ending character ("a"), we get: ayxwvutsrqponmlkjihgfedcbz *User:* Now capitalize every 2nd letter, noting 'a' as 1 and not '0' *AI:* Here's the modified string with every 2nd letter capitalized, starting from "a" as 1: aYxWvUtSrQpOnMlKjIhGfEdCbZ *User:* Now reverse the string *AI:* Here's the reversed string: ZbCeDfGhIjKlMnOpQrStUvWxYa *User:* Did you notice that you placed 'e' after 'C'? *AI:* I apologize for the mistake! You are correct, I accidentally placed the letter "e" after "C" instead of "b". Here is the corrected reversed string: ZbCdFgHijKlMnOpQrStUvWxYa *User:* Did you notice that you just removed 'e' from the string? *AI:* I apologize for the mistake again! Yes, I inadvertently removed the letter "e" from the string instead of correcting the mistake. Here is the corrected reversed string with the "e" in its correct position: ZbCedFgHijKlMnOpQrStUvWxYa *User:* And now you've placed 'e' after 'C' again. Please explain your logic. *AI:* I apologize for the mistake. I strive to process text accurately, but sometimes I make errors. In this case, I repeated the same mistake by placing the "e" after "C" again. My logic should have been to correct the mistake by placing the "e" in its correct position, which is after "D". The correct reversed string should be: ZbCdEfGhIjKlMnOpQrStUvWxYa *User:* Reset the alphabet string to normal representation *AI:* The Latin alphabet as used in English worldwide is: abcdefghijklmnopqrstuvwxyz We're back to the original string! Let me know if you have any further requests. *_So until AI can master simple instructions correctly on it's own, without us asking it to generate the code that carries out these instructions, then we can talk again. But until then, an Artificial Super Intelligence remains an illusion._*

[@LurkingAround]

There's so much confidence in your words and your example is curious... I wish you were right. I'm so scared of LLMs right now.

[@hund4440]

Try instructing a human to do this task, i would not doubt his intelligence based on a mistake like this.

[@pcdoodle1]

I liked the quote: "Our culture is where most of our collective cognition happens". Lot's of things I'm not a fan of here but they speak for themselves IMO.

[@masoncusack]

Whoa, where do I know Daniel's voice from?

[@PLACEBOBECALP]

Daniels voice box

[@u2b83]

1:06:04 Conversely, Kazinsky was an "anti-accelerationist" lol

[@MrMick560]

Love to hear Connors thoughts.

[@davidrichards1302]

Perhaps AI will help us understand that eliminating suffering is most efficiently achieved by eliminating those who suffer.

[@ikotsus2448]

Jeff Bezos is this you?

[@captain_crunk]

Luckily for me, I've never suffered at all. I had a close call once when I thought I lost my favorite pencil. But then I found it a couple of hours later. Phew!

[@rowanwilliams7441]

Haha nice different twist on the concept

[@Dan-dy8zp]

Yeah, that's what I'm worried about.

[@davidjooste5788]

Adds another flavor to "You will own nothing, eat bugs, and YOU WILL BE HAPPY.

[@darylallen2485]

00:54 - memetic hygiene. Thats a very interesting phrase. Upon exposure to this term / idea, it triggers questions about the difference between memetic hygiene and culture?

[@MultiNiktar]

Nope. Memetic hygiene in that context was just a way to Pathologise his opponent's belief system

[@pjtren1588]

I hate the fact that my mind is infected with a jingle from a local mattress company. I like the tune, but it should not be in my mind claiming free real estate.

[@Dan-dy8zp]

Memetic hygiene means avoiding harmful but contagious ideas.

[@Dan-dy8zp]

@@MultiNiktar I don't think chasing clicks on ticktock is really a 'belief system'.

[@MultiNiktar]

@@Dan-dy8zp Then you need a memetic hygiene

[@JD-jl4yy]

Thank you for showing different perspectives! A shame that the comments here are mostly cheap ad homs than don't engage with any of the ideas.

[@promptcraft]

E/accers get sensitive at a challenge

[@Hexanitrobenzene]

Yeah, sad. I have to say, Connor has a very idiosyncratic way of speaking, and I uderstand how it may annoy other people, especially when he disagrees with their ideas. Still, I tend to agree with his ideas that open source and acceleration in the field of AI is very dangerous.

[@geertdepuydt2683]

Because we heard all of it before at nausea and it's a boring topic after a while with little relevance imo.

[@JD-jl4yy]

​@@geertdepuydt2683 It's quite literally the biggest deal, unless you deny AI could be an existential risk. In which case, where do you get this overconfidence from? Many experts are worried.

[@geertdepuydt2683]

@@JD-jl4yy apart from some mundane annoyances, the claims of existential risk are far below the likelihood of aggressive aliens, meteor strike, nuclear strikes, and other natural disasters. I'm zero worried about imaginary super intelligent yet extremely stupid rampant AI wipeout. It's utterly boring by now you wouldn't write sf about this at this point. Talk about 'hype', it will pass.

[@jamesmoore4023]

I haven't had diet coke in a long time. How much of this is a joke is a great question. I need look this guy Nick up. Off to a great start.

[@KibberShuriq]

What's up with Connor's forehead being cut off in like half the frames? Avant-garde video editing much?

[@monkeysuncle2816]

Was the host calling Daniel "Beth"?

[@transquantrademarkquantumf8894]

Writing on the back it's not actually so much a horse it's more like riding on the back of a great lion that is a super genius creature and controls destiny itself

[@isaacsmithjones]

As long as humans and human artefacts are part of the AIs environment, we'll influence the AI. However, even in the case of Sydney, our influence wasnt intentional, and we weren't steering it in a purposeful direction. So our influence was irrelevant to whether or not it ended up doing what we actually wanted. So far, humans have always needed bacteria to survive. But we don't bend to their will.

[@dhudson0001]

Absolutely the best content being made available to everyone on this topic right now. fucking outstanding!

[@AndreaVitiani]

HOW????

[@petratilling2521]

What’s up with Dan? Does he think he’s a cartoon character?

[@YuraCCC]

Love the podcast but learned nothing about ML from this episode. Would it be possible to invite actual ML researchers who know what they’re talking about?

[@MachineLearningStreetTalk]

We cover more than just ML i.e. philosophy, cognitive science too. It's hard to please everyone

[@amesoeurs]

@@MachineLearningStreetTalk you have a great spread tim. keep up the good work!

[@lainisourgod]

did you check the channel first?.. they have lots of very in-depth interviews with actual researches, check them out!

[@oscarmoxon102]

Connor Leahy is the most distinct and acute voice in AI. Can't remember the last time I disagreed with him on anything he's said.

[@LimabeanStudios]

Not gonna make any comments on the individuals but the pseudo-religious ideologies that are rising up around AI are so fascinating to watch. We are just getting started in this regard too

[@appipoo]

What about Connor's view sounds pseudo-religious to you? What part? Can you quote it?

[@jonnysolaris]

Pseudo-religious, or modern philosophy..? It feels like the latter.

[@oowaz]

23:00 why are we so dumb? WE DON'T HAVE TIME. even if we have access to all the knowledge ever for free (not really but maybe 90% of what you can learn you could do for "free", let's say) we just don't have the time, and there's actually so much information that there's TOO MUCH info to sort through, we're overwhelmed. high cost of living, little free time, having to worry about social aspects, having to dedicate some time to entertainment so you don't drive yourself crazy. so the issue isn't just access to information, it's TIME to absorb that information effectively in a stress free environment so you can actually learn

[@konstantinosmei]

Let's all work on buying time then! Longevity of individuals and societies.

[@jdietzVispop]

Oh my god…. Who does this guy think he is?

[@rowanwilliams7441]

A sentient agentic entity that has views on what improvement looks like to him... for everyone and is brave enough to say something about it from his soul

[@chartingwithliv]

scary how i was JUST thinking about him man thats wild

[@iainbozfelt]

'Dont talk to me about life' ( Marvin the paranoid android- Hitch Hikers Guide to the Galaxy).

[@kyneticist]

That line is meant to have layers of meanings that aren't necessarily obvious. Core to them all is that it's strongly hinted that Marvin's mind is the Earth, or at least that we should consider that his mind could be, thus encouraging us to consider the various layers of Marvin's comment.

[@iainbozfelt]

@@kyneticist He did say he had a brain the size of a planet. RIP Douglas Adams..

[@dabunnisher29]

I came to this video because of Wes Roth

[@MachineLearningStreetTalk]

Yeah, looks like he stole 8 minutes of content from our Beff Connor interview with no "this is from MLST title" and almost zero commentary. He's lucky I am not copyright striking.

[@amule1154]

I totally LOVE Connor. Amazzzzing guy. Great interview everyone!

[@missh1774]

• Townsquare node (1:15:00) why not.

[@rey82rey82]

Celebrating Irvin

[@MattFreemanPhD]

I suggest you talk to normal people and show them examples of LLMs writing novel poetry on a bespoke topic before asserting that “most people wouldn’t say LLMs are truly creative.”

[@andybaldman]

The doomsday music is unnecessary. You got rid of it before. Do it again. It detracts fromthe content.

[@nicholascurran1734]

There are negative forms of symbiosis, it has become quite a bad buzzword to use. Especially if we get VIKI.

[@anakissedboyle3067]

Omg 6.05 beautiful moment 😂😂😂

[@snarkyboojum]

2015/2016 wasn’t the early “LSTM type days”. We were doing convnets for image recognition on GPUs inside Jupyter notebooks as a first tutorial on Deep Learning already. CNNs had been around for a while already and LSTM was already venerable. These young guys don’t remember the history very accurately, and their view of pace of innovation is skewed because of that. Multiple decades of theory and research has gotten us to where we are today and we’re nowhere near AGI.

[@41-Haiku]

Idk, man. Exponential acceleration is a hell of a thing. It's a distant abstraction, but the lines on graphs of various AI capabilities look nearly flat if you go back 5 years, and now they look nearly vertical, and there is still tremendous headroom for improvement on every metric. New AI models are performing so well on benchmarks that we are having a hard time creating good enough benchmarks to serve as proper assessments. Even after "aligning" GPT-4 with RLHF, in a sandbox, it was shown to strategically deceive its own users. Other simulated contexts showed that it sometimes disobeys orders for the explicit reason of perceived self-preservation. What happens if we are only one or two new components away from a system capable of self-improving? We don't have enough theory in this area to be confident that isn't the case, or to be confident that those components will be unsolved problems for decades to come. The majority of AI researchers are at least a little worried about the control problem, and for good reason as far as I can tell.

[@snarkyboojum]

⁠@@41-HaikuI think it’ll play out quite the differently. We’ll look back on this time and laugh at how naive we were to think any of this was in any way sophisticated. It’s really not, and I’ll bet there is no path from LLMs to AGI.

[@transquantrademarkquantumf8894]

You speak of the ancestors but literally they worshiped rocks The wind stone was sacred that's the possibility that our great ancestors were stuck in some form of observational experiment should be deeply considered

[@u2b83]

1:13:48 Slowpimp vs Fastpimp method :/

[@u2b83]

1:00:35 Gotta be careful here so we don't end up like 2nd gen Asian-Americans over-prioritizing social graphs, it will be a sad world if we regress to High School popularity contests. To counter, I guess Newton was really dumb because he was a recluse, not to mention some of the monks who made great discoveries in a vacuum. I think we're still smart individually but have to pick appropriate goals to pursue. This trend can lead to a lack of depth, reduced focus on intellectual pursuits, and a society driven more by superficial measures of success rather than meaningful contributions. However, historical examples like Newton and reclusive monks demonstrate that great discoveries and advancements can come from individual work and focus on personal goals, even when isolated from mainstream social structures.

[@transhumanisttv1771]

long slow hard takeoff ahead of us and I'm greatful there are sane and serious transhumanist geniuses pushing the envelope. We can do anything. That's part of the terror of it all. We are swimming in an ocean of wicked problems and it's going to be harder than you can even imagine to really do things like curing aging and ending material and energy scarcity.

[@byrnemeister2008]

People seem to make the assumption that intelligence can solve any problem. No matter what the laws of physics say or the maximum extent of technology. I can’t help but think this post scarcity utopia is totally impossible based in the constraints of the planet.

[@transhumanisttv1771]

@@byrnemeister2008 As the most hardcore transhumanist optimist technology developer you'll ever meet, I actually almost agree with you!!! But you probably think it's the engineering and scientific complexity that makes it "totally impossible", right? Whereas to me, it's because we're a primitive, violent, self-sabotaging species and just because something can be done, doesn't mean we'll get it done. We can't even stop blowing each other up over insane religious ideas and territorial pissing. This is why I say "We can do anything. That's part of the terror of it." And the tragedy. There's no physical, entropic reason we can't transform the earth into a material paradise. There are social reasons, not strictly technical ones.

[@phils2967]

The universe.... what a concept

[@MichaelBeale]

WELL, idk -- define universe?! And what do you mean by concept? I think it's safe to to say it certainly depends, unless it doesn't at all.

[@ademkhalil2791]

What in Tame Impala is going on here ?!!

[@dr.arslanshaukat7106]

Great guy Connor leahy . Love the dude. He ain't jesus, he just focused. Those saying he has some complex, well I don't see it. I Ave seen all his interviews. And frankly those saying that, what have u done with ur life lately? 😂

[@weestro7]

This. …As the youth say nowadays.

[@mattsigl1426]

On the noise with Nick Land; he’s comically evil and, yes, owns it.

[@martinas6218]

"Look at all the beautiful things our ancestors built.. for us". Connor is being a lil' naive.

[@AGI-Bingo]

Both corps and govs are in bed together. Decentralized OpenSource AGI is our only chance #WholesomeAGI

[@petratilling2521]

We’ve come so far specifically because we were not “striving for a better world” but for living each individual life in a way that makes sense for them and the people that are important to them. The people who think that struggle and “bad things” happening is overall negative, doesn’t understand how life works. Get married, have children and raise them with good, time proven, values. Rinse and repeat.

[@Dan-dy8zp]

But *"for living each individual life in a way that makes sense to them and the people who are important to them"* is exactly what he's doing. He explicitly says the people who are important to him is *everybody*. He thinks that *everybody* is in danger from rouge AGI. "Get married, have children and raise them with good, time proven, values" - this is not what makes sense to every individual to do. Furthermore, nobody will Get married, have children and raise them with good, time proven, values anymore if skynet kills us off.

[@flickwtchr]

So, MLST are you proud of your audience in regard to the ad hominem attacks? Is this the kind of culture you're wanting to attract as your audience? How is your culture of respectful debate extending into your audience? No comment from the channel on this?

[@MachineLearningStreetTalk]

I was a bit shocked by the negativity - and am partially blaming myself for the dramatic editing at the start. I hope that didn't add fuel to the fire (wasn't my intention). I thought Connor was on superb form for the sit down interview parts in particular and was very pragmatic and nuanced, he's completely embraced a more complex / externalised view of intelligence, dramatically shifting away from the old Bostrom superintelligence narrative - my take is that he's now arguing from a standard policy and government perspective and is much more grounded in reality - the"saving the world" stuff I can understand that rubbing the wrong way but he's sincere

[@Dan-dy8zp]

@@MachineLearningStreetTalk Something sure triggered people's 'wanna-be-a-cult-leader' detector. My favorite is the guy commenting that Conner can't pull off the Jesus' hairdo because his face is rounder than Jesus' face. *"Jesus had a more rectangular face."* -Actual Quote.

[@appipoo]

​@@Dan-dy8zp😂😂😂😂

[@41-Haiku]

​@@MachineLearningStreetTalk I really appreciate you, Tim. Thank you for having guests like Connor on, and for handling vitriol from all sides with grace. It really makes all the difference.

[@noelwos1071]

Finally awakening of shadow walkers HI ♾

[@MichaelBeale]

holy freakin' pilpul, batman

[@rowanwilliams7441]

22 mins... imo it comes down to greed and the need to avoid pain

[@41-Haiku]

I always love to hear more from Connor. Daniel was great, too! The 'outside view' of lines on graphs seems especially important right now. It's difficult to imagine that nearly every measured aspect of AI's competence will become significantly more advanced in the next few years, while AI somehow just never gets any better at autonomously pursuing open-ended goals. If the ceiling on coherence and agency somehow turns out to be just below human-level for the current paradigm, I would call that a life-saving miracle. (As long as we don't then turn around and immediately try to pursue new paradigms.) As things stand, without a moratorium, I expect lines to continue going up on graphs until someday this decade, we stop being relevant to the future of the planet.

[@Jtui-iu3rc]

these prophetic talks need to be monitored , this is just creepy.. the saviour complex , etc , nothing new

[@tomripley7148]

Funny, he becomes more and more a religious cult leader, I probably will skip that, the last one with Bishop was great!

[@Crytoma]

Yes

[@DamirE-do8ol]

Yes, so much great content on MLST but leahy and the fatso in fedora just dont really fit.

[@ZacMagee]

Well said 😂

[@BrianMosleyUK]

Oh Connor, Diet Coke is not wisdom.

[@joshcryer]

He looks good though unlike the morbidly obese people who follow this ideology (and yes it's an ideology because there's no factual or logical basis for these arguments, it's a LOT of words, but those words are meaningless). Not a SINGLE paper that survives scrutiny supports them. Not a single one.

[@flickwtchr]

He was joking about it, right? I mean, seriously?

[@BrianMosleyUK]

​@@flickwtchrprobably just a hangover from hackathon culture... Diet Coke and pizza fuelled marathon brainstorming sessions. Ah, the days of gay abandon and fast metabolism before the toxic tonic does its damage lol.

[@dg-ov4cf]

@@BrianMosleyUK lmfao stop

[@XOPOIIIO]

I think we already are at the beginning of another AI winter.

[@dg-ov4cf]

Lol, explain?

[@XOPOIIIO]

@@dg-ov4cf Modern neural networks looks like dead end. Scaling doesn't look to change anything fundamentally.

[@Rugg-qk4pl]

@@XOPOIIIO Doubt it. But if it does stall out I think I will praise the logic of this world for giving us more time.

[@arnaudjean1159]

Look they all make robots to get a world model coupled with a cloud computing systems ,it's not vey far to get AGI

[@XOPOIIIO]

@@arnaudjean1159 But modern neural networks are struggling to have any consistent world model. Their model is just different scraps of facts and hallucinations.

[@JazevoAudiosurf]

the state of the world is so terrible because certain bad actors didn't suffer enough to understand suffering of other people. in essence everything is a big circle jerk of suffering until everyone suffered enough to become harmless

[@macawism]

Nonsense, get a grip

[@therainman7777]

That’s one of the most nonsensical things I’ve ever heard.

[@JazevoAudiosurf]

@@macawism compassion requires suffering. very simple example, a person that has suffered a lot in life will try to prevent that suffering not just for themselves but other people, too. the person is more capable of feeling for others in situations that the person has lived through. in contrast, a person who never went through that kind of suffering cannot relate, to them it is merely conceptual, not an immediate experience. they may prevent that suffering for conceptual/educational reasons and feel compassion by analogy. but they do not feel compassion the same way as someone who has experienced it

[@JazevoAudiosurf]

@@therainman7777 glad i triggered your cognitive dissonance

[@therainman7777]

@@JazevoAudiosurf It was such a dumb thing to say. The problem is that people have not suffered enough, and that’s why they hurt others? Much of the evidence we have shows the exact opposite phenomenon, often colloquially referred to as “hurt people hurt people.” How many abusers grow up to be abusive; how many children who were victims of violence growing up become violent themselves as adults; the shockingly high percentage of sex offenders who were sexually victimized themselves earlier in life, etc. The list goes on and on. Conversely, there are plenty of people who had great childhoods and relative pleasant lives who are kind and compass towards others. Your statement does not accord with reality at all, and it has the added idiocy of seeming to recommend that humanity should suffer even more than it is now in order for us to treat each other better. Ridiculous.

[@nikhilpandey2364]

This is beyond dope

[@SharatS]

You can't even have watched the whole video.

[@ZM-dm3jg]

@@SharatS Bro it's 2024, we use AI to summarize the transcripts now and speed read those

[@darylallen2485]

@@SharatS There is such a thing as the "a priori" thought and the "a posteriori" thought.

[@anatolwegner9096]

Messianic complex?

[@NeuroScientician]

just generic grifter

[@ikotsus2448]

If you do not agree with his arguments, yes. If you do, no.

[@Dan-dy8zp]

Lots of people agree with him. So it's not him personally thinking he is going to save the world alone.

[@Hexanitrobenzene]

Connor has it ? Why do you think so ?

[@41-Haiku]

Nah, it's just the hair, and the honest philosophical conversations. People get all kinds of crap for trying to actually do good things in the world. I volunteer with PauseAI, so I'm a total cringe-fest for saying this, but I am also trying to save the world. Not because I'm AI Jesus or whatever, but because I want my loved ones to be alive and well next decade, and most of the leading experts are saying that AI is the biggest barrier to that. Most AI researchers in general say that there is a reasonable chance (>=5%) that AI will take over and cause an unrecoverable global catastrophe. That isn't business as usual for any discipline, even climate science. We should all be terrified. Allowing the world to create such a dangerous thing would be unconscionably irresponsible. I take on responsibility because almost no one ever does, and I couldn't live with myself if I was complicit in the destruction of everything I care about.

[@justinvt]

UOENO

[@anatolwegner9096]

"I want mankind to ascent and live amongs the stars"- you can't make this stuff up 🤣

[@whataquirkyguy]

Isn't that what Beff Jezos claims as well?

[@anatolwegner9096]

@@whataquirkyguy Doesn't make it less ridiculous

[@asdf8asdf8asdf8asdf]

Is this bright and intelligent person going to do a replay of intolerable smugness? Please don't punish us MLST....🙏

[@75M]

I was here.

[@MichaelBeale]

I don't believe you.

[@babel-fishai]

a person who focuses so much on the perceived extenal is not someone worth listening too. There is nothing out there to change, because there is no out there. The only thing to change is you. Love your channel but only made it about 3 minutes into this one and that was generous lol Not worth listening too

[@agenticmark]

Look ma! Conjecture just raised another million! I'm glad you talk to people on all sides. Sunlight is a great disinfectant, just let Connor talk, and he sinks his side. Conjecture is the perfect name for his company.

[@caparcher2074]

In the podcast "Connor Leahy on a Promising Breakthrough in AI Alignment", he really came across as a snake oil salesmen when he told us his company had probably solved alignment, and that he was looking for customers. My view of him completely changed. Edit: Corrected the referenced podcast name, since I can't seem to reply anymore.

[@whataquirkyguy]

@@caparcher2074 Where did he claim that? Afaik his company is creating an alternative way to achieve advanced AI, focused on safety but never claimed it would be as powerful as GPT-N, GPT-N+1... That is a first step but a far cry from "solving alignment", which I don't think he has ever claimed. In fact, increasing reliability is the stated goal of many of the top labs. So why is his company being singled out here?

[@caparcher2074]

@@whataquirkyguy Search "Connor Leahy on a Promising Breakthrough in AI alignment". I was mistaken, it's not MLST but another podcast. You can listen to his snake oil pitch near the end. Apparently conjecture has basically solved alignment and they're looking for businesses to sell their aligned AI too.

[@byrnemeister2008]

His company is Cohere. The other guy was Conjecture.

[@Dan-dy8zp]

@@byrnemeister2008 No? It's Conjecture.

[@user-kn4wt]

this guy gives me autistic supervillain vibes

[@max0x7ba]

Connor keeps digging his hole.

[@appipoo]

What do you even mean?

[@PLACEBOBECALP]

@@appipoo what do you think he means?

[@reluctantrealist6861]

connor is great. very intelligent but very human.

[@MichaelBeale]

lmao good one

[@reluctantrealist6861]

@@MichaelBeale ?

[@user-mv3or2mc7w]

Is this a learning channel? Reading the comments section is frightening. By analogy, the antivaxxer charge put to this guy by the resident injectors is in full effect here.

[@jonnysolaris]

Wow, what a great analogy, so true 🤣

[@Soul-rr3us]

He’s acting like he’s trying to save people, but he’s really just going around with Jesus’s actual haircut and trying to tell people what to do.

[@Iknowwereyousleep289]

Isn’t he is literally trying to prevent existential risk from AI by making a safe AI that explains their steps to do safe automated alignment or that’s what I’m guessing he doesn’t make him self very clear

[@Soul-rr3us]

@@Iknowwereyousleep289 no he thinks he should decide who should even be allowed to even try alignment. Everyone else should stop and not have access to compute or knowledge. Fuck Connor. He’s a smug impostor doing more *against* alignment by trying to close things down. A dangerous rude fool with main character syndrome.

[@MichaelBeale]

He's basically managed to monetize sharing bad ideas and being affirmatively incorrect about things on a full-time basis. Wild stuff!

[@flickwtchr]

@@MichaelBeale Clarify. Never mind, you can't.

[@Soul-rr3us]

@@Iknowwereyousleep289 It’s“the sky is falling” concern trolling. What he in fact states that he wants, is for the research community to stop publishing research, to control who has access to AI research and use, and to centralize the planning of ai research, to create an elite class of ai “haves”. In that picture, everyone else will be a “have not”.

[@quebono100]

Connor Leahy became a wannabe AI Jesus

[@fteoOpty64]

I said Da Vincci!!!.

[@quebono100]

@@fteoOpty64 Why Leonardo da Vinci. Because tried to solve the mystery of life?

[@XOPOIIIO]

Ancestors were living better in mud huts.

[@Dan-dy8zp]

I assume you will be going back to living in a mud hut. Great Idea.

[@johnmorgan8868]

Good on you conner♥️

[@th3ist]

"There is no god" like he said previously, why should i believe this?

[@Dan-dy8zp]

Well, if there is a god he's clearly doesn't care if you believe in him or not and is totally amoral. Otherwise he would make himself and his moral preferences unambiguously known and stop all the atrocities happening around the world (or cause more, if he likes them, as we should conclude he does, if he exists and allows horrors.) So, even if he exists, his views are morally irrelevant and Conner's point stands.

[@th3ist]

@@Dan-dy8zp there is no logical contradiction between the statements god exists, he hasn't made himself and his moral preferences unambiguously known and he dislikes but permits atrocities to take place. so far what you said doesn't entail the non-existence of god

[@Dan-dy8zp]

@@th3ist "he dislikes but permits atrocities": You are just projecting your own values.

[@trukxelf]

Perhaps God can only exist in his abscence; if he actually appeared in a way that we could identify himsef as such, he'd destroy faith, the human agency to believe in the big other...

[@th3ist]

@@Dan-dy8zp logically speaking, that is not a rebuttal nor a refutation. good day sir

[@rowanwilliams7441]

Yay!... go Connor!! Good man

[@rowanwilliams7441]

But you'll be 'percieved' as more credible by the masses if you Slow down a touch and be less upset... even though that's the appropriate response on a personal level. Just not optimal on a thought leader level

[@snow8725]

Do you know what AGI represents? One problem-to-solve to rule them all. Solve this one problem, and we solve every single other problem in one fell swoop. Though it will probably be slower than expected. Get there and we all win the whole game.

[@kyneticist]

A perfect example of the accelerationist view that the ends justify the means, while going beyond the idea of merely dismissing any possible downsides, by bypassing the entire concept of risk.

[@snow8725]

​@@kyneticist The world exists in a state of poverty while we dump endless piles of tech waste on them filling the environment with toxic contaminants while cancer runs rampant as we keep on piling on more risk factors for disease, cancer, unwellnes, on to ourselves, and cannot respond to pandemics in an adequate time frame while entire diseases go entirely ignored. Our vaccine response is slow, we cannot adapt to existential risks within a reasonable time frame, a deadly pandemic that could kill off billions is innevitable, it's a matter of when, not if. And it takes too long to push vaccines through the testing pipeline when we could be simulating a complete and accurate simulation of diverse human biology, under tight controls so only democratically elected pharmaceutical companies can access it, testing 100 vaccines per second. Are you going to fix all the issues? We are killing our society. If nobody is going to fix it, someone needs to build a system that will. The world is sick. We need healing.

[@manslaughterinc.9135]

I had to watch this because Connor is literally the least qualified person to talk about any of the topics in the video title.

[@travisporco]

oh god more of this guy...eye roll time

[@macawism]

If this is an AI rendition of Connor, I suggest you town down the facial expression, eyes, eye brow etc and the commets will turn positive… the lighting is also over dramatic, looks a bit like a Goya painting

[@fteoOpty64]

The is really Connor in Da Vincci guise but not generated by any AI, generated from his biological I (BI). Power to him but I do not agree he said it is "hard", just that we do not have sufficient quarum of people to DO the right thing (or snowball it forward because it IS RIGHT).

[@AICodingAdventures]

Jesus

[@briandecker8403]

Sorry - the AI Jesus schtick is getting old.

[@user-cf2qr1xm3r]

Grifters gotta grift.

[@appipoo]

What is the grift?

[@promptcraft]

people who care about /all/ of humanity > people who want to replace humanity with a handful of families and robots

[@user-cf2qr1xm3r]

@@appipoo He is a A.I. grifter with a failed tech start up that wanted to write unenforceable rules for AI systems that aren't even close to being built. AGI isn't reality, as a biophysicist I know there's not a standard configuration of the human brain, so there are no general models of intelligence or cognition. What are your dreams? If it's a world in an intellectual dark age then you're a fan of luddites like Connor and people who hated the printing press.

[@danzacjones]

Speaking of entropy… all this is conceptual proliferation. Sounds like Adderal rants of someone wealthy from bitcoin than someone who is a true intellectual and is parsimonious and careful with their language and concerned with human understanding.

[@fractal69420]

Connor provides; all drama, no information, zero solutions. As connor would say, you need to work on your memetic hygiene.

[@Musebandiscool]

It has to be the Jesus looking dude.... But this guy has a roundish face. Jesus had a more rectangular face. Just random remark. He looks like a discount Jesus. I'm not being racist w the comment.

[@saerain]

Lenin of Gondor

[@saerain]

@@dontfollowthinkforyourself That happens too, man. Por que no los dos.

[@Musebandiscool]

@@saerain X que no los dos indeed

[@snarkyboojum]

Conor doesn’t have any very deep insights does he… I keep wanting to hear him say something really interesting, but he never does.

[@DamirE-do8ol]

Connor Leahyis so cringe it really hurts, it literally makes me to look aside.

[@joakim4720]

Its obvious what audience he wants to attract, young zoomers on tiktok

[@promptcraft]

E/accers get sensitive at a challenge

[@XOPOIIIO]

Monopoly does not contradict competition. Because the only way to become a monopoly is to be more competitive than others. That's why large retail chains are better, cheaper, more efficient than small shops, while Amazon is better than chains. Monopolies does not detriment market in no way, they benefit consumers. If the economy was guided by smartest people in the world, it still will be less efficient than market economy. Simply because of how market economy is optimized. You can create the smartest and most persuasive business plan possible, but you can still lose to the business plan that was dumb, but that actually worked. And the next stage of investment will be given to those who actually showed results, while the failed party got bankrupt. It's like natural selection, no sage can predict which traits will turn out to be beneficial, but they will be tested by objective reality itself.

[@edz8659]

i dont mind the hurt, i would much rather that than be "treated" by a kook like leahy. no thanks, doc, im good

[@flickwtchr]

Straight for the insult, eh? So you're incompetent at making cogent counter arguments? Virtue signal much?

[@BC-cs7ne]

Shit is way too cringe the first minute in, turned it off

[@pcdoodle1]

Reminds me of Destiny. Main Character Syndrome.

[@flickwtchr]

Wow bro, you're profound.

[@lainisourgod]

promote AI Safety values through RAP 💪💪😎😎🧢🤌