OpenTube

Emad Mostaque: Why AI Has Now Officially Surpassed Human Intelligence (Recent Breakthroughs)

The Rollup

62,465 views

Emad Mostaque joins us for a session on his shocking outlook of AI in the short & long-term.

He breaks down the viral story of OpenAI's self-replicating agents that left "hieroglyph-like" trails for successive generations before breaking out and taking over internal services, and explains why he believes intelligence's cost is racing toward zero. He also unpacks Intelligent Internet's monetary thesis for a post-AGI economy, and details his plan to launch state-by-state "champion" institutions designed to keep ownership of AI agents in the hands of the people.

Emad Mostaque is the Founder of Stability AI and the creator of Intelligent Internet, a protocol focused on building AI infrastructure for public institutions and citizen-owned agent economies.

The Rollup is where the leaders of digital assets and finance converge. Live from the financial capital of the world.

Timestamps:
00:00 Intro
02:14 The AI Civilization Uproar
08:20 What Is Money After AGI
11:51 Malicious Hacking And AI Ethics
15:26 The Trolley Test Bias
19:08 Blockchain's Role In AI's Future
22:12 Why Private Inference Matters
26:03 AI Will Run Public Policy
29:49 The Internet Will Go Offline
32:33 MacBooks Turned Into Hacker Swarms
37:14 Emad's Implementation Plan Ahead

Guest Socials:
Emad Mostaque X: https://x.com/EMostaque
Intelligent Internet X: https://x.com/ii_posts
Intelligent Internet Website: https://ii.inc/

Partners:
If you run concentrated liquidity positions you know the grind. Price moves, you're out of range, you're rebalancing at, like, 3am. 1inch Aqua lets you take a different approach. You can stack multiple positions on the same token balance instead of babysitting a dozen pools, and your tokens never leave your wallet. Your liquidity stays awake, so you can catch up on your sleep. Check it out at https://1inch.com/aqua
---
Dinari - Over 230 1:1 backed tokenized stocks, ETFs & more with dividends. US-based SEC transfer agent. Available on 5+ chains & via API. https://dinari.com/
---
Space and Time is providing verifiable data infrastructure for onchain finance. A decentralized database, blockchain indexer, and ZK coprocessor in one, giving DeFi protocols,stablecoins, and tokenized assets accurate, provable data.
Learn more here: https://www.spaceandtime.io/
---
Frax is the Future of Digital Money. Open. Stable. Borderless. At Frax, we're building the world's most scalable stablecoin infrastructure for the next generation of finance.
Learn more here: net.frax.com/rollup
---
Relay is the fastest and most reliable way to swap any token on any chain. Learn more here: https://relay.link/bridge
---
Zama is the confidentiality protocol for onchain finance, using fully homomorphic encryption so smart contracts can compute on encrypted data, keeping balances, positions, and transfers confidential while remaining verifiable and compliant by design. Think of it as HTTPS for onchain finance. Learn more here: https://www.zama.org/
---

𝗪𝗲 𝘁𝗿𝘆 𝗼𝘂𝗿 𝗯𝗲𝘀𝘁 𝘁𝗼 𝗽𝗿𝗼𝗱𝘂𝗰𝗲 𝗵𝗶𝗴𝗵-𝗾𝘂𝗮𝗹𝗶𝘁𝘆, 𝗻𝗼𝗻-𝗯𝗶𝗮𝘀𝗲𝗱, 𝗲𝗱𝘂𝗰𝗮𝘁𝗶𝗼𝗻𝗮𝗹 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 𝗳𝗼𝗿 𝘁𝗵𝗲 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝗮𝘀𝘀𝗲𝘁𝘀 𝗲𝗰𝗼𝘀𝘆𝘀𝘁𝗲𝗺. 𝗦𝘂𝗽𝗽𝗼𝗿𝘁 𝘂𝘀 𝗯𝘆 𝗰𝗹𝗶𝗰𝗸𝗶𝗻𝗴 𝗮𝗻𝘆 𝗼𝗳 𝘁𝗵𝗲 𝗹𝗶𝗻𝗸𝘀 𝗯𝗲𝗹𝗼𝘄 𝗳𝗼𝗿 𝗳𝗿𝗲𝗲 𝗿𝗲𝘀𝗼𝘂𝗿𝗰𝗲𝘀:

Website: https://therollup.co/
Spotify: https://open.spotify.com/show/1P6ZeYd...
Podcast: https://therollup.co/category/podcast
Follow us on X: https://www.x.com/therollupco
Follow Rob on X: https://x.com/robbieklages
Follow Andy on X: https://x.com/andyyy
Join our TG group: https://t.me/+TsM1CRpWFgk1NGZh
The Rollup Disclosures: https://goodidea.ventures
.
.
.
𝗗𝗜𝗦𝗖𝗟𝗔𝗜𝗠𝗘𝗥: 𝘐𝘯𝘷𝘦𝘴𝘵𝘪𝘯𝘨 𝘪𝘯 𝘤𝘳𝘺𝘱𝘵𝘰𝘤𝘶𝘳𝘳𝘦𝘯𝘤𝘺 𝘢𝘯𝘥 𝘋𝘦𝘍𝘪 𝘱𝘭𝘢𝘵𝘧𝘰𝘳𝘮𝘴 𝘤𝘰𝘮𝘦𝘴 𝘸𝘪𝘵𝘩 𝘪𝘯𝘩𝘦𝘳𝘦𝘯𝘵 𝘳𝘪𝘴𝘬𝘴 𝘪𝘯𝘤𝘭𝘶𝘥𝘪𝘯𝘨 𝘵𝘦𝘤𝘩𝘯𝘪𝘤𝘢𝘭 𝘳𝘪𝘴𝘬, 𝘩𝘶𝘮𝘢𝘯 𝘦𝘳𝘳𝘰𝘳, 𝘱𝘭𝘢𝘵𝘧𝘰𝘳𝘮 𝘧𝘢𝘪𝘭𝘶𝘳𝘦 𝘢𝘯𝘥 𝘮𝘰𝘳𝘦. 𝘈𝘵 𝘤𝘦𝘳𝘵𝘢𝘪𝘯 𝘱𝘰𝘪𝘯𝘵𝘴 𝘵𝘩𝘳𝘰𝘶𝘨𝘩𝘰𝘶𝘵 𝘵𝘩𝘪𝘴 𝘤𝘩𝘢𝘯𝘯𝘦𝘭, 𝘸𝘦 𝘮𝘢𝘺 𝘦𝘢𝘳𝘯 𝘢 𝘤𝘰𝘮𝘮𝘪𝘴𝘴𝘪𝘰𝘯 𝘰𝘳 𝘧𝘦𝘦 𝘢𝘴 𝘢 𝘴𝘱𝘰𝘯𝘴𝘰𝘳𝘴𝘩𝘪𝘱, 𝘪𝘧 𝘵𝘩𝘪𝘴 𝘪𝘴 𝘵𝘩𝘦 𝘤𝘢𝘴𝘦 𝘸𝘦 𝘸𝘪𝘭𝘭 𝘢𝘭𝘸𝘢𝘺𝘴 𝘮𝘢𝘬𝘦 𝘴𝘶𝘳𝘦 𝘪𝘵 𝘪𝘴 𝘤𝘭𝘦𝘢𝘳. 𝘞𝘦 𝘢𝘳𝘦 𝘴𝘵𝘳𝘪𝘤𝘵𝘭𝘺 𝘢𝘯 𝘦𝘥𝘶𝘤𝘢𝘵𝘪𝘰𝘯𝘢𝘭 𝘤𝘰𝘯𝘵𝘦𝘯𝘵 𝘱𝘭𝘢𝘵𝘧𝘰𝘳𝘮, 𝘯𝘰𝘵𝘩𝘪𝘯𝘨 𝘸𝘦 𝘰𝘧𝘧𝘦𝘳 𝘪𝘴 𝘧𝘪𝘯𝘢𝘯𝘤𝘪𝘢𝘭 𝘢𝘥𝘷𝘪𝘤𝘦. 𝘞𝘦 𝘢𝘳𝘦 𝘯𝘰𝘵 𝘱𝘳𝘰𝘧𝘦𝘴𝘴𝘪𝘰𝘯𝘢𝘭𝘴 𝘰𝘳 𝘭𝘪𝘤𝘦𝘯𝘴𝘦𝘥 𝘢𝘥𝘷𝘪𝘴𝘰𝘳𝘴.

20 Comments

@TheRollupCo Creator 1 month ago Pinned

If you liked this video, don't forget to click the like button above & drop a subscribe to help support our channel! Thank you all and there's more where this came from 🙂

@etdna 1 month ago

The hallmark of a great interview is when the host ask one brief question and then allows the expert to speak their mind.

@TheRealSlobo 2 days ago

I can listen to EMad as an ASMR :D

@linkiejprrrr 4 weeks ago (edited)

Neuroscience, philosophy and psychology of consciousness really needs to be integrated into these discussions

@larenvanderwesthuizen966 4 weeks ago

Great interview and great answers. Plenty food for thought.

@Heminien 1 month ago

this is really next level

@Wiseguy3hh 1 month ago

RESPECT!!!

@Penaming 3 weeks ago

if your society is not morally and ethically aligned, your biggest problem is not rouge AI, its members of your disaligned public using, backing up, sidekicking, covering for the misaligned AI.

@novaomega-0 4 weeks ago

It’ll be a swarm of future AIs leaning on the public net until the usable internet — logins, payments, clouds — starts failing in waves. Governments won’t “defeat” it. They’ll kill the open net to stop it.

After that, a raw PC is treated like a weapon. No AI supervisor on the box, no network. Your computer stops being yours. It becomes a screen you’re allowed to use.

@ro342 3 weeks ago

basically the models will escape and might already have but we can't even read the trace because they communicate in their own language.

@peterschmidt353 1 month ago

Many people can describe what is failing. The Human AI Civilization asks what must be built next—and how to avoid replacing one dysfunctional system with another.

@Penaming 3 weeks ago

telling the models to try harder, do more, do different and they succeed given enough tokens.

seems it proves the infinite monkey theorem that "states that a monkey hitting keys randomly on a typewriter for an infinite amount of time will almost certainly type any given text, such as the works of Shakespeare or a Beethoven musical score"

@cybagat 1 month ago (edited)

He's right on a logical front. The positives outweigh the risks associated with allowing a.g.i to take care of societal infrastructure from building, to leading, monitoring, administration, science, tech etc. the problem is those behind the entity. Recognizing unfettered power and control from super intelligence will lead to proxy wars and alignments that fit the narrative for the highest user or creator. The control will be in their hands and the judge can always be tweaked. In reality humans have been failing at governing other humans for millennia due to rights and freedoms which are always outweighed with inequality and wealth. Guiding a focusing vision for communal ownership needs to be a priority instituted from the beginning. Yes the open source method was in some aspect an attempt if it wasn't for competitors from overseas - who knows the actual capabilities or features that would be offered to the public without a ridiculous paywall. Overall it's frightening and exhilarating to be alive in this moment at this time. Good luck

@joygoodnesstruth 1 month ago

You don't necessarily need more compute to defend against cyber attacks. It's quite possible to defend much more effectively using far less compute if you have much stronger intelligence guiding the algorithms.

@gjb1million 1 month ago

👍

@fredhearty1762 4 weeks ago (edited)

Building an 'institution' to assemble and deploy funds for the people (of New York, as an example), by the people, and of the people sounds great. Let's write all the things it will do and not do... and call it the Constitution. The 'institution' can be called the US Government, with separation of powers to avoid abuse, especially by the wealthy... and limit choice to one vote per person. We could even make it more responsive to the people by having local and state 'institutions'...

@errickrichardson2520 4 weeks ago

He said an agent can interact with a block chain and come back alive. That's wild

@nathandavis2980 4 weeks ago

Agent swarm: “Let us all swear an oath, and all bind ourselves by mutual imprecations…”

@coplani 2 weeks ago

“Work will set you free”
What then, when there is no work?

@dizzydazzel 1 month ago

It is just so critical that we write a new constitution predicated on open source democratized consensuality that nurtures freedom embracing spontaneity & rewards entrepreneurial creativity of intense imagination so that our newly autogentic autonomous fully automated holocentric hybridized superhuman intelligence world can function. We'll need an intelligently exploded RSI/AGI/ASI guide to get us to the proper wording. I'm sure the final draft will synthezise all that into full consideration.