Comment by Forgeties79

10 days ago

I have a feeling the latest incremental release of ChatGPT is maybe less threatening to the survival of all living species than a nuclear weapon, but I’m no expert so could be wrong.

10 comments

Forgeties79

bumby 10 days ago

Sure, but it’s a bit of a strawman. Define what you think is the appropriate level of risk to society without resorting to the extremes and it becomes a more productive conversation.

Is it just short of a model that can be used for bioterrorism? Security vulnerabilities that can cripple banks? Or just undermining the current capitalist incumbents? There’s a broad spectrum

harimau777 10 days ago
I'd say something like: If you provide it to one company then you have to provide it to all companies. If you aren't comfortable with that, then either nationalize it or accept that no one gets it.
- bumby 10 days ago
  
  I can understand the sentiment, but that’s not how the govt has operated in the past. There are numerous examples of tech transfers to benefit a single company. It’s reasonably common in aerospace and health applications.
Forgeties79 10 days ago
A strawman? “Don’t resort to extremes”? It’s the example you proposed!
- bumby 9 days ago
  
  The strawman I was referring to was the binary option of "threatening to the survival of all living species...or not". Perhaps "false dichotomy" would have been a better description.
  The GGP point was that this level of control has been normal when dealing with other nations. The nuclear point was that it has also previously been applied internally, so the original framing of "other nations vs us" is not particularly good.
  If your counterpoint is "yeah but that's ok when it's based on existential risk" that doesn't support your first point, it just reorients the framework from "external vs internal" to "risk based". If you change the goalposts to the latter, you are actively undermining your original point. That's a fine point to make, just not coherent with your original one.
  
  5 replies →