Guillaume Verdon: Should We Have a 'Second Amendment for AI'?

Based Beff Jezos, co-founder of Extropic, discusses AI safety, decentralization, and going analog.

1x 1.1x 1.25x 1.5x 2x 3x

:15 :15

Download

Guillaume Verdon: Should We Have a 'Second Amendment for AI'?

HD Video Download

Must we accelerate AI innovation?

You've probably heard of "effective altruism," but how about "effective accelerationism," or e/acc? "You claim to be building an artificial god in the human image. We're building the conduit for the thermodynamic god that created us. We are not the same." 

Those are the words of Based Beff Jezos, a pseudonymous X account devoted to spreading the e/acc message far and wide, once described by venture capitalist Marc Andreessen as a "patron saint of techno-optimism" and by its detractors as "unhinged" and "absolutely toxic for the AI discourse." 

The man behind Based Beff Jezos was unmasked in Forbes magazine last December as Guillaume Verdon, a Google quantum computing engineer-turned-co-founder of the AI startup Extropic, which has received about $14 million in seed funding to develop a new kind of chip for running AI models. 

He joins us today to talk about effective accelerationism, the politics of AI, and what his company is doing to make sure that Artificial General Intelligence (AGI) arrives quickly and remains beyond the control of a single corporation or government. 

Sources Referenced:

• "Who Is @BasedBeffJezos, The Leader Of The Tech Elite's 'E/Acc' Movement?" by Emily Baker-White in Forbes
• what the f* is e/acc, the e/acc newsletter
• "SITUATIONAL AWARENESS: The Decade Ahead," by Leopold Aschenbrenner
• Gavin Newsom's veto of California's AI bill
• "What is Moore's Law?" by Our World in Data
• "Microsoft, Google and Amazon turn to nuclear energy to fuel the AI boom," by Mehek Mazhar in CBC Radio
• Extropic: Ushering in the Thermodynamic Future
• Just Asking Questions with Bryan Johnson

Chapters

• 00:00 Coming up…
• 00:28 Introductions
• 02:13 What is effective accelerationism?
• 04:33 Building a conduit for the thermodynamic god that created us?
• 09:54 Is AGI inevitable?
• 17:01 Why open source AI doesn't need regulation
• 29:18 A Second Amendment for AI?
• 31:58 Philosophical foundations of e/acc
• 37:16 An AI arms race between the U.S. and China?
• 55:27 Criticism of OpenAI
• 01:01:08 AI under the second Trump administration
• 01:07:22 Designing a biologically inspired chip
• 01:15:46 How the American chip industry will be impacted by China and Taiwan
• 01:18:44 Will AI make humans irrelevant?
• 01:23:53 How do we reimagine governance?

• Producer: John Osterhoudt

Comments

[charliehall]

ChatGPT still makes up stuff. There is even a technical term for that: Hallucinations. Until this gets fixed, AI will never be reliable.

[SQRLSY]

Try "Perplexity". I've had very-very little trouble with it... No hallucinations that I have noticed. Also it provides web sites (links) to back up what it writes.

[Zeb]

Have you checked all the links? AIs are known for making up references when asked as well.

[VULGAR MADMAN]

It’s super cool that we’ve invented schizophrenic machines, don’t you think?

I can’t see anything bad coming from this.

[JesseAz (mean girl ambassador)]

Consider the training data.

https://www.nytimes.com/2023/04/18/technology/reddit-ai-openai-google.html

[Longtobefree]

How will you know it is fixed? Ask the AI?

[Zeb]

Then it's a lot like natural intelligence in that way.

[Rev Arthur L kuckland (5-30-24 banana republic day)]

I will support this as soon as all the fag ags that go after the 2a are personally liable

[mad.casual]

The 20th Century, for mankind, was the deadliest Century of any in history and it was because mankind killed more of itself than any other conspicuous or preventable/avoidable cause of death. The 21st has had a similarly hellacious start.

As I pointed out in the other BTC thread, any one of the top 10 crypto heists in the 5 yrs. have exceeded both the lump sum and the per capita theft of any cash or credit scheme by almost an order of magnitude despite the fact that it performs and order of magnitude fewer transactions for an order of magnitude fewer people.

This, vaguely, tracks with long-held almost primordial precepts of Iron Law of Bureaucracies and 'Power corrupts and absolute power corrupting absolutely.'

I'm not opposed to any/all progress, but blindly running head first into traffic or off a cliff because you think it's the way forward isn't smart and even plunging ahead while obliterating any awareness of your past nullifies the very conception of progress and makes it impossible to even know if you're going in circles or not.

[Peter Kreeft taught me LOGIC, BEWARE !!!!!!!]

I have followed AI since the early 60's -- YIKES-- and all this modern talk is not about AI it's good old Expert Systems (no intelligence whatsoever and nobody claimed there was) and Natural Language Processors, glorified sentence diagrammers, again with not an ounce of intelligence and no one ever claimed that. That covers 90% of what is being talked about.

But the bullshit does cover the whole 60 years. Read the book
What Computers still can't do by Dreyfus and realize you are being scammed by PR folks who want to get into Biden's wallet.
======================
Dreyfus was rather contentious but even more eirenically and convincinly the case is made against AI by the great scientist and philosoher Michael Polanyi

Take your pick

"…a formal system of symbols and operations can be said to function as a deductive system only by virtue of unformalized supplements, to which the operator of the system accedes:
symbols must be identifi able and their meaning known, axioms must be understood to assert something, proofs must be acknowledged to demonstrate something, and this
identifying, knowing, understanding, acknowledging, are unformalized operations on which the working of the formal system depends"

[Peter Kreeft taught me LOGIC, BEWARE !!!!!!!]

This will be like the Climate Change nonsense. We'll spend our ass off (Janet Yellen says we need at least THREE TRILLION A YEAR) and people will make a living off of it, it will eventually fail (cf the many wrong predictions of Al Gore) and we will be off on another mad chase.

[mad.casual]

Guillaume Verdon: Should We Have a 'Second Amendment for AI'?

Did Guillaume choose this title? Because for all the high-brow, techno-optimism something tells me Guillaume Verdon is not the *based* 2A scholar he thinks he is.

[Kungpowderfinger]

Fucking Google engineers are going to “…make sure that Artificial General Intelligence (AGI) arrives quickly and remains beyond the control of a single corporation or government”?

https://en.wikipedia.org/wiki/Dragonfly_(search_engine)

[Winston in Wonderland]

Until somebody is accountable for AI's mistakes, it needs to be regulated.

In other words, if your AI defames me or defrauds me, I want to know whom I can sue.

Until then, keep your AI out of my life.

[Zeb]

And how is this regulation going to work? The source code is out there. And a lot of clever people. You're not getting that toothpaste back into the tube.

[mad.casual]

On the topic of techno-optimism: Colonization of Mars will make an entire planet "The Puerto Rico of Earth" for the next 500 yrs. Despite less than half the solar radiance of Earth, minutes in round-trip EM communication, and relatively less abundance of pretty much everything we have here on Earth, the Reason Magazine of 500 yrs. in the future will declare states that forbid the launching of interplanetary rockets from within their borders to be backwards, irrelevant, anachronistic idiots afraid of War Of The Worlds broadcasts (because being 140 million miles further from the Sun is just an abstract social construct).

Elon wants to spend a trillion of "his own dollars" to die on the rust ball, fine (though there is a question about what happens to the cult once he dies). But there's no serious colonization of Mars happening without a serious colonization of the other 75% of the Earth's surface that humanity doesn't currently occupy even year round. As long as The Jones Act is plausibly responsible for making PR a shithole (whether that's because PR is a shithole or people are stupid enough to believe that The Jones Act is the sine que non cause), Mars will be a very expensive and technologically complex shithole.

[Archibald Tuttle]

I listened to "02:13 What is effective accelerationism?" What I heard was a bunch of word salad. Even Kamala Harris could not have done better at talking without saying anything.

[Rick James]

I suggest you go and watch every season of HBO's Silicon Valley. That will help you better understand tech-incubator speak.