Yes. Regulate end product deployments.
Don't regulate R&D with an arbitrary compute or mode size threshold.
@ylecun
-
Regulate AI Deployments Not R&D Compute Thresholds
By
–
-
Sanity as a barrier to dangerous AI development
By
–
Sanity is what would prevent us from doing that.
-
Two-layer neural networks achieve universal learning capabilities
By
–
Two-layer neural nets are universal learners, too.
How is that relevant? -
AI Existential Risk and Open Source AI Debate
By
–
Another comment to a tweet from @tegmark asking me (again) why I think AI won't kill us all, and claiming that the question of existential risk is disconnected from the question of open source AI.
-
AI Autonomy Claims Debunked: Understanding System Limitations
By
–
This idea that we don't really understand AI systems and that they "can become autonomous and modify their own goals" is complete bullshit.
-
Hollywood’s Lone Genius Myth: How AI Innovation Really Works
By
–
The idea of lone geniuses that figure it all out by themselves, with no one else coming close, is the biggest fallacy of most Hollywood SciFi movies.
They really have no idea how the world works. -
AI Safety and Existential Risks Without Proper Safeguards
By
–
Are rocket engines capable of doom or not?
In some stupid scenarios, they are.
Build thousands of nuclear intercontinental ballistic missiles, and launch them. Is AI capable of doom or not? In some stupid scenarios, they are.
Build powerful AI systems without guardrails to make -
Recurring AI Gone Wrong Theme in Science Fiction
By
–
Same story in 2001: A Space Odyssey.
Same story in Frankenstein.
A recurring theme in SciFi. -
Jaan Talinn funds major AI existential risk research institutes
By
–
Max Tegmark is the President of the Board of the Future of Life Institute, which is bankrolled by billionaire and Skype and Kazaa co-founder Jaan Talinn.
Talinn also bankrolls the Cambridge Centre for the Study of Existential Risk, the Future of Humanity Institute, the Global