The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
It's like if we gave Skynet control of the B2 bombers and it crashed every single one of them and we all had a good chuckle and decided it should be put in charge of the nukes, too.
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion yeah the few things it is good at are things we could already do pretty well
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion not to mention that the first time I've heard text-to-speech it was over 35 years ago on an 80286
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion omg this!
-
@gwynnion not to mention that the first time I've heard text-to-speech it was over 35 years ago on an 80286
@matus_chochlik I heard it on the Commodore 64.
️ -
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion We also had search engines, chatbots and cheat tricks for lazy students which were perfectly fine
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion I forget who made this observation but:
The AIs that just did their jobs fairly competently are female coded (Alexa, Siri) and the response was "Meh. Kinda neat."
The AIs that confidently spout bullshit are generally male coded (e.g. Claude) and the response was "We must give them billions of dollars and put them in charge of everything!"
-
The thing that gets me is we already had working speech transcription, we had useful digital assistants, and the insertion of LLM based "AI" made them worse and even less reliable.
@gwynnion LLMs truly are the boss's nephew.
-
@gwynnion We also had search engines, chatbots and cheat tricks for lazy students which were perfectly fine
-
J jwcph@helvede.net shared this topic
