Brutkey

Dare Obasanjo
@carnage4life@mas.to

@farlukar@mas.to @kjhealy@mastodon.social Finding the constantly diminishing set of edge cases that generative AI handles poorly does not negate my statement.

I remember when people use to joke about six fingered AI generated images and now Google Veo can generate video nearly indistinguishable from the real thing.

Pointing out what AI models do poorly today doesn’t negate the value of what they do well today.

Weyoun 6
@weyoun6@kolektiva.social

@carnage4life@mas.to @farlukar@mas.to @kjhealy@mastodon.social It demonstrates that ANN's do not "understand" anything. That doesn't mean they can't be useful tools, but they are not intelligent. Between the hype and the pushback to the hype, there is a lot of noise drowning out evaluations of the technology. There are some tasks it can do well, and on most it does poorly.

Van Sice
@vansice@infosec.exchange

@weyoun6@kolektiva.social @carnage4life@mas.to @farlukar@mas.to @kjhealy@mastodon.social This thread is missing what AI _does._ It is a probability engine. That is how it works: it puts together the next most probable thing. It's not actually 'learning' anything. Books written with AI lack continuity for this very reason. If 6 fingered hands go away, it's only because this special case was addressed not to happen. Engineers learned. AI did not.