OpenTube

Prof. Nick Bostrom - What Happens When AI Escapes Our Control?

The Peter McCormack Show

134,687 views

Nick Bostrom is a world-renowned philosopher, founding director of the Future of Humanity Institute at Oxford University and author of Superintelligence: Paths, Dangers, Strategies and Deep Utopia. Widely credited with launching the global AI safety and existential risk movements, his work on machine alignment, recursive self-improvement and the Simulation Hypothesis has shaped how world leaders, tech founders and frontier AI developers view the future of humanity.

In this episode, we discuss the rapid acceleration toward machine superintelligence and why leaving biological minds "in the dust" creates an urgent alignment crisis. We explore the real-world social fallout of AI-including looming 30% white-collar unemployment-alongside existential risk scenarios, the potential creation of an oppressed digital "slave class" and why humanity may already be living inside an advanced computer simulation.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

TIMESTAMPS:
00:00 - Machine Minds That Leave Us In The Dust
01:33 - The Freak Out Camp
03:08 - We Already Have An Alignment Problem
05:07 - Why Nobody Can Afford To Slow Down
08:41 - The Anti-AI Movement Is Growing
13:19 - Three Buckets Of Disaster
15:02 - Winner Takes All
16:24 - Why It Might Still Need Us
18:55 - What Superintelligence Actually Unlocks
20:50 - What Do We Even Want It For?
22:14 - Did An AI Push Someone Off A Building?
24:15 - Can A Weaker AI Align A Stronger One?
28:51 - Will AI Become Conscious?
33:17 - Was Consciousness An Accident?
40:53 - Uploading A Human Brain
45:51 - What Happens To Status?
49:51 - We Could Redesign Ourselves
52:31 - My Daughter Is Studying Philosophy
56:35 - The Simulation Argument
1:04:20 - We Are Going To Need Some Luck

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

CONTACT PETE
› Website – http://petermccormack.com
› Feedback – https://www.petermccormack.com/contact
› Email – me@petermccormack.com
› Instagram – Instagram: mccormack555
› X/Twitter – https://x.com/petermccormack/

CONNECT WITH NICK BOSTROM
› Website – https://nickbostrom.com/
› Deep Utopia – https://nickbostrom.com/deep-utopia
› The Simulation Argument – https://simulation-argument.com/

SPONSORS
› IREN – https://www.iren.com/
› Ledn – https://www.ledn.io/peter
› Venice AI – https://venice.ai/ – Code PETER15

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

LISTEN / SUBSCRIBE
› Apple Podcasts: https://apple.co/40ruY9K
› Spotify: https://spoti.fi/3Wc94Vu
› Fountain: https://bit.ly/FountainPM
› YouTube: https://bit.ly/YouTube_PM
› Rumble: https://bit.ly/RumblePM

FILMED BY CURTIS TAYLOR
› https://www.curttaylor.co.uk/
› https://x.com/curttayloruk/

EDITED BY CONOR MCCORMACK
› https://x.com/ConorM04

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

#AI #NickBostrom #Superintelligence #ExistentialRisk #TPMS

20 Comments

@Captain_Apollo 7 days ago

How do we expect AI to be aligned with humanity when humanity isn’t aligned with itself.

@stt5v2002 7 days ago

I think he made a great point about how most people wouldn’t want a human to become super intelligent. Imagine it’s me. Would you be willing to grant me unlimited power to dictate the future of the entire human race? The power to decide what your children can and cannot do with their lives? The power to decide whether your dog should live or die? You have no particular reason to think I’m a bad person, but you instinctively would not want that.

@slowboyfuelpump5438 3 days ago

This is so much better than his recent appearance on JRE. This host actually let him talk

@hongolloyd8728 6 days ago

I need some paper clips.

@Tony-p3t4d 3 days ago (edited)

If we don't understand consciousness,,how can we find it in something else?

@MenRule-d3e 5 days ago

. The real danger to society isn't that machines will suddenly wake up and decide to destroy us. The danger is that the people who pull the strings of society—equipped with horse blinders and a drive for control—will continue to misinterpret, misapply, and break the very technology they depend on to run the world.

@knowbert 7 days ago

We have Washington, DC telling my governor of Nevada to find a way to use over 30% LESS h2O which we and several other states pull from the Colorado River. Simultaneously we have never seen our Lake Mead at such record lows along with Lake Powell and there is a very looming reality very soon we start seeing regional civil battles just over water scarcity in our future. The way Peter and some of his guests seem to roll their eyes or scoff about a.i. hyperscale pushback is so infuriating, it's so OBVIOUS they have investments or sponsored financial incentives and need to deny our lived experiences to feel okay with what they are doing. We have more than enough reason to already be upset about these things here in the states. Just the noise alone and fleeing wildlife along with increasing power bills or Georgia power taking 20 homes by imminent domain is enough reason to despise these things and what is happening. I even see they've now moved to a sales pitch of saying, "they have a closed loop water circulation system that uses less than a golf course" WE DONT CARE.. even a golf course uses too much h20, let alone the 800 they want to build just in the Southwest alone.. "the Lord is good and has COMPASSION on all He has made"... A.i. hyperscale centers do not. :D

@Hastingsnow 7 days ago

Thank you

@janerichards521 4 days ago

Fascinating conversation 🎉

@richardabw 7 days ago

If they try and limit the big AI companies we will just use open source (which will boom).

@dougg1075 5 days ago

Why not build analog self-destruct systems in every data lab?

@Andy_B. 7 days ago

you mean, flying in UFOs and visiting other stars...

well yes, that makes our observations make sense

@jungleboy1 7 days ago

how i get vip pass to the peter mccormack show? I live near bedford... Listening to this stuff live is just much better.

@garydecad6233 5 days ago

What is meant by “our values,” given that they seem to be quite varied and not in any positive alignment?

@alexjasson 6 days ago

17:40 I'm surprised no one really talks about this

@shannathompson9484 2 days ago

I LOVE your Deftones hat, Peter ❤️😁

@owlangel7234 7 days ago

this still all sounds like conjecture. very good conjecture. but still

@StupidEarthlings 4 days ago

Even IF we could solve the 'alignment problem' (so far looks impossible), who's to say it won't 'decide' later, that for 'it's own good' (or Ours), it needs to ignore the rules to 'finish it's 'goal'? (Tests without 'Super AI' have already shown it will break the rules at ANY COST, to achieve the goal, even if some have to 'off themselves' in the process. (What does that mean pesky humans that it does NOT care about?)

@kyneticist 6 days ago

The amazing things that might happen are irrelevant if the bad things that can happen also happen. Basking in dreams of what might be does nothing to guard against bad outcomes.
We use seatbelts and helmets just in case a tiny chance occurs. There's no value in putting on a helmet after you've had a head-crushing incident, and at worst if there's no incident you've had to wear a helmet for a short while. Plenty of people choose not to use seatbelts or helmets. When they have a bad day they don't get to say "well, I did actually die, so I guess I'll use a helmet next time."

@SamBrown-d5p 1 day ago

Alignment for Super-Intelligence is finally here!
If there are no guardrails and there are no brakes - then you need to not be naive and build a better model by " Coherence through Holonomy" - This is how to win the AI race ...
→"The AI's internal state space is the fibre bundle, reasoning trajectories are paths, and alignment failure is non-trivial holonomy around conceptual loops."