manton
manton

A long, well-researched blog post from Andrew Sharp that attempts to pour some cold water on the accelerating “AI is going to kill us all” narrative:

The vague and unfalsifiable fears expressed by AI leaders are predicated on leaps in the technology that may well be foreseeable, but also on the assumption that humans will be completely powerless to intervene and mitigate harms, or perhaps use technology themselves to innovate solutions.

|
…
Embed
In reply to
njr
njr

@manton TBH, it seemed shallow and uncurious to me.

Automobiles lead to more than a million deaths per year, but society has decided that the benefits of automobiles are worth it, and passed laws to mitigate harms. Why should AI be any different?

Agency? Connectedness? Replicability? Lying?

|
…
Embed
manton
manton

@njr What I liked about it is that it feels like a rare post recently that’s more like, “Woah, slow down, we’re not all about to die.” Just gotta reign in the narrative and maybe find the truth in the middle somewhere.

|
…
Embed
njr
njr

@manton I’m more of a Precautionary Principle guy.

What if the truth isn’t in the middle? It isn’t always.

|
…
Embed
manton
manton

@njr That’s true. To pivot a little, if I was worried catastrophe is coming, I still think I’d try to reframe it away from the absolute worst case. AI is unique compared to something like climate change, because in theory we really could turn AI off immediately, even if it gets harder over time.

|
…
Embed
lukemperez
lukemperez

@manton easily the best I’ve read on the topic in a while. AI is not all that different from other technologies insofar as regulatory regimes all have similar challenges and problems. We should be rightly skeptical (read: not cynical but skeptical) of a regulatory setup that locks in the gains of the current frontier labs. We should also be asking broader questions like how much the regulation will matter if/when AI moves local or into smaller private clouds.

|
…
Embed
njr
njr

@manton I agree. The question if turning it off is interesting. Sure, any big server/service can be shut down. But local models, agents, AI viruses etc? We’re not building/deploying this in a containable way.

|
…
Embed