Discussion
I find the polarised debate on AI's potential risks frustrating.
The ironic thing is if the AI companies were able to talk themselves into getting regulated, that could help with all the *other* harms people are worried they're just distracting from.
Instead it's siding with "it's a hoax" Trump, and diminishing the public pressure which is the best chance of getting regulation and international agreements; actually getting some control over a dangerous industry.
1/3
And shouldn't the burden of proof be on those saying there's NO risk, rather than the other way round?
Atomic reactors have to be spec'd before they're built w/ annual loss of control risk < 1 in a million.
AI companies are growing trillion-param models without knowing what they'll do til after. Racing while practically begging for the race to slow down. Because what if there's even a 1 in a 10,000 chance that it gets out of control? What level of risk is acceptable here?
You need to log in to see that page.
Install bonfire.cafe
Get the full app experience
And yes, they really DO believe there's a risk of losing control of AI, because that's the belief/philosophy of the communities they've come from, and they've been warning about it for decades. Just look into their history.
That doesn't mean they're right, but it does mean that we shouldn't just dismiss it on the basis that they've got upcoming IPOs. They were warning about this the whole time; most people just weren't listening.
2/3