scripod.com
Should we slow down AI progress? | MOONSHOTS #288

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Should we slow down AI progress? | MOONSHOTS #288

Moonshots with Peter Diamandis

2 DAYS AGO
Moonshots with Peter Diamandis

Moonshots with Peter Diamandis

2 DAYS AGO
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

The mates sit down with Emad Mostaque to discuss mounting warnings from AI labs, a researcher’s claim that we’re “gambling with our lives,” calls to slow AI development, and the accelerating race toward superintelligence. Get access to metatrends 10+ year...

Highlights

This episode examines whether AI progress should be slowed, how data and compute are reshaping competition, and what accelerating intelligence could mean for science, work, health, governance, and humanity.
08:12
Data beats algorithms
24:43
The story smells wrong
33:08
Slowing down could waste time instead of solving the problem
44:36
Alignment should steer progress, not stop it
1:00:16
Science is thoroughly cooked
1:05:15
Plan for what breakthroughs unlock
1:13:48
Compute must be secured now
1:19:50
Robotics is the real end game
1:33:07
Transformer attention can be beaten and improved
1:41:25
Model weights become national-security assets
1:58:51
GDP may be the wrong measure
2:08:12
Abundance should come with an obligation to protect workers
2:12:57
Longevity escape velocity may already be here in spikes
2:16:21
AlphaGenome as a genomic periodic table
2:25:00
We are building living products
2:29:34
AI progress does not end with AGI
2:33:02
Robotic manipulation may be close to saturation
2:35:13
Ancestor simulation is not a trade-off
2:39:26
Recursive AI can spiral in any direction

Chapters

The AI race and the case for building through crisis
00:00
Data becomes the strategic AI advantage
07:35
From data valuation to catastrophic AI warnings
15:35
Acceleration makes ordinary life misleading
48:54
Compute, world models, and the next production bottlenecks
1:11:42
Low-cost models reshape competition and control
1:33:25
The economic meaning of a possible AI singularity
1:49:23
AI turns biology into an addressable design space
2:08:47
From genomic lookup tables to a biology economy
2:17:51
After AGI: engineering, personhood, and suffering
2:25:25
Embodied AI, simulations, and the accelerating finish
2:30:47

Transcript

Peter H. Diamandis: Jacob Coxon, a researcher who spent three years on pre-training both OpenAI and Anthropic, resigned this week. Neither company is acting responsibly in the race towards self-improving superintelligence. He's called the competition gambl...