We might build AGI by 2026 | Dario Amodei and Lex Fridman

[@LexClips]

Lex Fridman Podcast full episode: kzbin.info/www/bejne/q5jZeXaOeLSgo5Y Thank you for listening ❤ Check out our sponsors: lexfridman.com/sponsors/cv8228-sa See below for guest bio, links, and to give feedback, submit questions, contact Lex, etc. *GUEST BIO:* Dario Amodei is the CEO of Anthropic, the company that created Claude. Amanda Askell is an AI researcher working on Claude's character and personality. Chris Olah is an AI researcher working on mechanistic interpretability. *CONTACT LEX:* *Feedback* - give feedback to Lex: lexfridman.com/survey *AMA* - submit questions, videos or call-in: lexfridman.com/ama *Hiring* - join our team: lexfridman.com/hiring *Other* - other ways to get in touch: lexfridman.com/contact *EPISODE LINKS:* Claude: claude.ai Anthropic's X: x.com/AnthropicAI Anthropic's Website: anthropic.com Dario's X: x.com/DarioAmodei Dario's Website: darioamodei.com Machines of Loving Grace (Essay): darioamodei.com/machines-of-loving-grace Chris's X: x.com/ch402 Chris's Blog: colah.github.io Amanda's X: x.com/AmandaAskell Amanda's Website: askell.io *SPONSORS:* To support this podcast, check out our sponsors & get discounts: *Encord:* AI tooling for annotation & data management. Go to lexfridman.com/s/encord-cv8228-sa *Notion:* Note-taking and team collaboration. Go to lexfridman.com/s/notion-cv8228-sa *Shopify:* Sell stuff online. Go to lexfridman.com/s/shopify-cv8228-sa *BetterHelp:* Online therapy and counseling. Go to lexfridman.com/s/betterhelp-cv8228-sa *LMNT:* Zero-sugar electrolyte drink mix. Go to lexfridman.com/s/lmnt-cv8228-sa *PODCAST LINKS:* - Podcast Website: lexfridman.com/podcast - Apple Podcasts: apple.co/2lwqZIr - Spotify: spoti.fi/2nEwCF8 - RSS: lexfridman.com/feed/podcast/ - Podcast Playlist: kzbin.info/aero/PLrAXtmErZgOdP_8GztsuKi9nrraNbKKp4 - Clips Channel: kzbin.info *SOCIAL LINKS:* - X: x.com/lexfridman - Instagram: instagram.com/lexfridman - TikTok: tiktok.com/@lexfridman - LinkedIn: linkedin.com/in/lexfridman - Facebook: facebook.com/lexfridman - Patreon: patreon.com/lexfridman - Telegram: t.me/lexfridman - Reddit: reddit.com/r/lexfridman

[@GregDowns]

This is one of those KZbin videos where you can instantly tell which commenter has watched it and which hasn’t. The guy’s prediction about his prediction being taken out of context in 2026 will come true.

[@ricochetrabbit9818]

What worries me is the dark side of humanity and what such technologies would affect the criminal elements of society. Such topics are so often ignored or merely glanced over when current technology is so easily hacked every day.

[@1130MarsV]

so in other words there's 0 evidence behind this prediction

[@wingzero7316]

AGI already exists somewhere in the future

[@IndianaBones24]

In other words? He stated this to begin with. So I'd say it's his words, it's not scientifically based

[@VexDrive]

@@wingzero7316 No it doesn't. We have no idea where or what the hard limits and great filters are.

[@Stanley-px3bt]

@@wingzero7316 AGI isn't possible with current hardware. We'd need a breakthrough in room temperature operating quantum computing to achieve it, and that's decades away right now, so yeah, in the future...

[@moocorp4574]

A prediction cannot have evidence unless the prediction has either failed or succeeded

[@alst4817]

News just in! Company tries to sell its product!

[@meatskunk]

Maybe we could start with a definition of what “AGI” actually is? Would be great to get that by 2026 but won’t hold my breath 😅✌️

[@MrFocusification]

artificial general intelligence

[@meatskunk]

@ right. Thanks for clearing that up 👌

[@carleynorthcoast1915]

Only 13 months left until the "this didn't age well" comments. Didn't the A.I. hype bubble already burst?

[@anonymouscat1299]

what hype?

[@johnburns9696]

Dude either didn’t watch the video or he has a future in political consulting.

[@justinleemiller]

AI is under hyped. Teachers and universities don’t even recognize that students are using it to do all their work.

[@carleynorthcoast1915]

@@justinleemiller Great until your in class essay is worth 50% of your grade.

[@gefallenesobst6855]

At least it feels like being on an exponential curve of the AI progress!

[@Unineil]

This is the most intriguing thing of 2024, ai and llm models.

[@JasonLaveKnotts]

Graphene transistor GPUs

[@kleemc]

Nah... i don't believe current architecture + scaling law alone will bring us to AGI. A superhuman capabilities in specific fields, that I believe. A general human level understanding of the world, human emotions and with intuitions, that would require couple more breakthroughs of foundational architecture.

[@woodoyoy]

You wouldn't believe in what we have today, a couple of years ago. Depends on how far "faking smart" can lead us. Who knows maybe we will be very surprised

[@donak2773]

Ai helps people be lazy. People will look back and miss life without technology

[@stant7122]

Meanwhile, just about every LLM can't output more than 4096 tokens in a single response. It has a long way to go.

[@mrnoblemonkey8401]

I’ve seen this prediction from lots of people. I’m interested to see what 2026 holds

[@---md2ny]

With current AI we are in discovery phase .(means we are at early era of cell phones , playing with it teying to find the best one creating lots of BS until world settle down like last 2 decades.where ios and android are the stable core so economy could grow naturaly again Every phase will last a little longer because of FOR PROFIT CORPORATIONS . I personaly expecting lots of fusion energy nuclear energy quantum computers

[@kratistoi1]

No bet was left unhedged here.

[@tomich20]

So much confusion to what we call AGI. Chatgpt with PROACTIVE talk before talking to it.. I would already call it agi

[@IvanMeouch]

I use open ai advanced voice mode to practice Japanese and it’s incredibly stupid and frustrating

[@brandongroves4465]

Johnny Adams, Norm MacDonald, Joe Rogan, Genghis Khan

[@answerth]

So far, models are GREAT at using and reformulating the data they've been trained on. Emergent properties are a beautiful thing, but all of this seems to revolve around it discovering or inventing information that it was not trained on. I wouldn't be surprised if it discovered the low hanging fruit and missed opportunities, but for it to have a model of the world and then make discoveries about that model for which data doesn't exist seems like quite a leap

[@christopherviers8302]

Not a big A.I. fan, but... the idea of applying it as an improvement to the medical welfare of humanity sounds wonderful...!!! Just keep the politicos out of the mix...!!!

[@rfn900]

2000s: we are going to build AI. We didn't. We then call our models "AI" and the old AI is renamed to "AGI". 2020s: we are going to build AGI. We don't. We call our models "AGI" and change the old AGI to "ASI". And the hype cycle goes on. Just keep lowering the bar for marketing to keep that sweet investor money coming.

[@usamamohamad4654]

But the models when have now are way better from those we had in 2000 so there is really a progress

[@rfn900]

@ oh yeah. For sure there has been a lot of progress. We probably wouldn’t call the current models AI, as the original concept defined. But yes, it is an achievement

[@mynameisjeff9124]

People from the 2000s would consider today’s models as AI, so everything happened as predicted 👍🏻

[@Stanley-px3bt]

The main companies making big money on A.I. are the hardware manufacturers like NVIDIA. The actual A.I. software companies aren't that profitable yet. That's why A.I. is 90% hype, and 10% useful products right now. The limits to current A.I. LLMs are about to be reached in a year, and there will be no A.G.I. breakthroughs. There will just be a final end to the current A.I. bubble and hype.

[@JerkMaiDhaang]

Yeah? Shit. Well, we might find life on another planet in 2025! GG

[@zandrrlife]

The core requirement for AGI, is continuous online learning at test-time. These companies are so dishonest. No you won’t. No lab is remotely close to that feat, none. That would be as big as backprop itself. These companies are building powerful AI models, but powerful AI isn’t AGI. They’re hoping to get as much representational coverage as possible and call it AGI. Na buddy 😂.

[@pandoraeeris7860]

We've already got AGI. o1 is AGI.

[@twizshaq]

😂😂

[@dbgarf]

I continue to be confused and surprised by how many people find it credible that a system for statistical inference is even the same KIND of thing as a generally intelligent mind. This is a category error.

[@calvinlow2]

Newman?

[@enkenthouse]

2030 it is then

[@themassmauler]

This is fantasy at best

[@SHUGSCOTLAND100]

Weve supposedly going to build robots that pass as humans and flying cars since the 80s and none of it ever happens 😂😂. Its a lot of shite.

[@woodoyoy]

Flying cars have been proven to be a dumb idea,they are not practical. We have them already, they are called helicopters. The robots could not exist because of the lack of computing power and many people predicted when robots will start happening based on Moors law. There is a reason they are billioner IT CEOs and not you

[@tonywuhooo]

Yeah sure buddy, after you get another 1 trillion in Vc funding right?

[@artsybt6015]

anthropic is least funded AI company. Google and OpenAI have the most money. I don't think dario is over hyping anything here. if anything he is being extremely honest. He doesn't know if it will happen in next 2 years but surely it will happen in our lifetime.

[@---md2ny]

Everything changed once we start using quantum chips. So from AGI to ASI to synthetic intelligence . Thats where fun begin .

[@aytee5862]

Shit like this makes me lose faith in "experts" in general.