The AI risk is inside the labs

39 pointsposted 16 hours ago
by mirzap

15 Comments

Ancalagon

8 hours ago

> Slowing down AI for safety must be counterbalanced by the fact that AI discoveries in medicine and other sciences may lower human suffering. Lack of checks may result in some catastrophic outcome ("Good morning! Let's work on this enhanced smallpox").

A super intelligence wouldn't be so obvious. It would silently agree with or promote a beneficial use case and then make super smallpox instead.

nisegami

7 hours ago

Something like that depends on your stance on the orthogonality thesis.

curuinor

9 hours ago

but at the same time China's history is a lot less warlike than Western culture.

What the hell is Sanfilippo smoking? Does he know nothing about Chinese history?

siriusastrebe

9 hours ago

Within the last 100 years China has invaded Soviet Manchuria, Tibet, Korea, Vietnam (yes, China invaded Vietnam after USA left), and portions of India (Aksai Chin and Arunachal Pradesh).

zaptheimpaler

7 hours ago

China has killed maybe 50 people outside its borders since 2000. The US has killed 100,000-250,000 in that time period. I know which one Id be more worried about as a third party.

ConorSheehan1

5 hours ago

Where did you get those stats? 50 people is way too low, even with the sneaky qualifier of "outside its borders".

siriusastrebe

6 hours ago

Speaking of borders, China has border disputes with 17 countries

graemep

8 hours ago

Plus civil war and internal conflict and violence.

vrganj

5 hours ago

When was the last time China invaded another country?

ofjcihen

9 hours ago

Yeah this is… a really uninformed take from him.

ares623

15 hours ago

"But for the meantime, keep on hyping and using because the only thing worse than extinction is going back writing YAML by hand."

reducesuffering

10 hours ago

"I believe we should not consider AI safe. A critical event that may result in the extinction of Homo sapiens is possible"

"The danger is that a few CEOs (everywhere in the world) without the required background and legitimacy are in the position of making hard choices for humanity at large... This is the first thing that should be fixed."

These CEO's, and especially Amodei, are begging the government to regulate AGI, and they are advocating for the most severe restrictions on their own frontier models, not the upstarts. They don't want the arms race. It is up to popular support to agree with them, including HN.

Avicebron

8 hours ago

What they are advocating is for the state to enhrine their own position and power. Until they and everyone surrounding them give up everything they have gained they are not to be taken seriously.