Again: stop calling the malware “rogue.”
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick Malware by design - IMHO the designers are responsible for the damage it causes and politics plus courts/attorneys are responsible to shut down the malware & designers and bring justice to their victims.
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick it's the same thing with car hitting pedestrians. It was the car that did it. The driver is not responsible.
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick I think that you are both making the same point. And the 66 and 99 (as we called them as kids) suggests to me it wasn't "nixcraft" using the term themselves but quoting the article?
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick They put rogue in quotation marks. Perhaps rethink this post?
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick They probably see wikis as a competitor now: why would people wait a minute for an LLM to reply to a question when they could just search for the answer on a wiki...
-
@dotsie well i thought they engagement bait about anything tech ._
@EvelynOctavia @dotsie
They bait about everything and anything that might spew up some sticky engagement.
And it works. They are master baiters.Mostly about tech. Good practical info on Linux in the mix, and boosted toots from others, but the account always seeks maximum engagement on all posts.
I blocked them a long time ago.
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick "It went rogue and it hacked that system" has the same energy as "it went rogue and that's how it got stuck up there, doctor"
-
@gleick They put rogue in quotation marks. Perhaps rethink this post?
@thejessiekirk @gleick
Naaah. They are engagement baiters. -
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick And the people responsible for runing them should be jailed like any other offense caused by carelessness.
-
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick reminds me of this old Simpsons scene
-
@thejessiekirk @gleick
Naaah. They are engagement baiters. -
RE: https://mastodon.social/@nixCraft/117390294462413805
Again: stop calling the malware “rogue.” These bots were designed, trained, prompted, and activated by humans at OpenAI. The fact that some of their behavior was unpredicted does not diminish OpenAI’s responsibility.
@gleick and the prompts were, according to their own account, not very different to what the agents actually did; what the agents did was 'overachieve'.
On top of that, these recent examples were hardly the first time we heard about agents doing unexpected and unwanted things ("it deleted my database"). To say that they didn't expect this is to say that they are beyond incompetent.
-
G gambolputte@expressional.social shared this topic
