The ones I hate are the AI captioning on YouTube videoes where the spelling is appalling and words that are somewhat homophonic will be spelt differently when they are captioned. Sometimes correctly.
From what I've seen you'll spend as much time checking the AI's work as just doing it. Useful for search, not otherwise.
I've had some really amusing experiences with CoPilot when I've asked it to identify buildings and works of art that I've taken pictures of on recent trips. Usually, it correctly identifies them , but there are times when it gets the identity and location completely wrong. For example, it will tell me that a building I took a photo of in Brazil was actually in Germany. Recently, it identified a tower in a German fortification as part of the Great Wall of China. When I tried to correct its error, it pushed back, giving me lists of reasons why I was mistaken, even after I input the information that I myself took the picture in the country I said it was in. I can get it to eventually admit that it had made such an error, but it is surreal to have an argument with a simulated person. AI engineers can tweak these programs to improve their accuracy, but fixing a problem with some subject matter might introduce anomalous behavior with regard to other subject matter. It is like trying to weigh frogs. They won't all sit still on the scale. It is laughable to think that chatbots can be scaled up to real consciousness or intelligence. Their only connection to reality is through patterns of text and pixels that real intelligent agents actually create.
www.bbc.com
Just more PR bullshit. Same as "our models are soooo dangerous we can't control them".Anthropic thinks its AI might be conscious. It has profound consequences for your life
Some of tech's most powerful leaders are treating AI like it's alive. It could have profound consequences for our world and your own personal safety.
That's as far as I got before the paywall slammed on me. Sorry.
Their chatbots are definitely not conscious, but I think one could say that Anthropic is very conscious of how precarious their future profitability is. I think that the Chinese are pulling ahead of the US precisely because their investments in technology have had to depend on less expensive means of achieving the same results with low-cost open weight models.Just more PR bullshit. Same as "our models are soooo dangerous we can't control them".Anthropic thinks its AI might be conscious. It has profound consequences for your life
Some of tech's most powerful leaders are treating AI like it's alive. It could have profound consequences for our world and your own personal safety.
That's as far as I got before the paywall slammed on me. Sorry.
At this point I'm just waiting for the bubble to pop and for the mania to end.
We'll find out if the marginal productivity gains are worth the expense, but more importantly, the C-suite will shut the fuck up about wedging LLMs into every aspect of our jobs.
I can tell you now that they are not, and it's not even close. The gains fall many orders of magnitude short of the expense already incurred, and expenditure is still accelerating, as the gains stagnate, and the techbros become ever more frantic in their absurd claims of (our new AI overlords will turn us all into) jam tomorrow.