A few weeks ago, a colleague told me something that sounded almost too good to be true: “You can feed ChatGPT a PowerPoint deck and it’ll make you a podcast video.” The implication was magical—you could turn a dry lecture into an engaging two-person podcast, all with the help of AI.
AI does lie/hallucinate. But, to your task, I think your friend may have confused ChatGPT for Google’s NotebookLM. NotebookLM can create an audio — not video (as yet) — podcast (choose their “Audio Overview” feature) from the materials you feed it. It’s really quite cool. https://notebooklm.google
This was perfectly captured by the great Stanislaw Lem (60 years ago!): "A smart machine will first consider which is more worth its while: to perform the given task or, instead, to figure some way out of it."
I have very limited experience with AIs, but all have been poor. When I've asked it questions, it's always given me a reply. When the reply sounded wrong, I confronted it. It then apologized and thanked me for correcting it. Did this dance a few times. I finally asked it if it had any idea. how to get the answer I was looking for and got a vague wave off.
I've also read about them faking precedent cases in legal briefs, which citations either don't pertain to the subject matter they're supposed to support; or they simply don't exist at all. Anywhere. And if you don't look for or at them, you accept that they're real and talk to the legalistic points being put forth.*
I suspect that someday--and probably sooner than later--AIs will be a lot more capable than they are now.
* This isn't limited to AIs. Living, breathing humans do it all the time. It's been said that 87.6% of all statistics are made up on the fly. And then there's
Blessed relief for the pain and discomfort of percentorrhea. J Irreproducible Results, 21 (3) (1975), pp. 3-4. Which really exists.
lol - funny, and illustrative of the fact that AI is not Superman. At the same time, LLMs can do pretty impressive stuff. When I ask Grok about the correlation of the 10-year Treasury rate to the Fed funds rate over time, it gives a pretty sophisticated answer.
What's almost endearing is the LLMs' willingness to confess when confronted with mistakes. I've noticed that myself when I catch them in a non sequitur or contradiction. Much better than stonewalling, although that has happened in some famous cases. The way it stalls though is interesting.
AI does lie/hallucinate. But, to your task, I think your friend may have confused ChatGPT for Google’s NotebookLM. NotebookLM can create an audio — not video (as yet) — podcast (choose their “Audio Overview” feature) from the materials you feed it. It’s really quite cool. https://notebooklm.google
ChatGPT may not be intelligent but it could certainly run for Congress.
This was perfectly captured by the great Stanislaw Lem (60 years ago!): "A smart machine will first consider which is more worth its while: to perform the given task or, instead, to figure some way out of it."
If necessity is the mother of invention, then laziness is definitely the father.
Your colleague lied to you about ChatGPT's abilities, eh? ;-)
I have very limited experience with AIs, but all have been poor. When I've asked it questions, it's always given me a reply. When the reply sounded wrong, I confronted it. It then apologized and thanked me for correcting it. Did this dance a few times. I finally asked it if it had any idea. how to get the answer I was looking for and got a vague wave off.
I've also read about them faking precedent cases in legal briefs, which citations either don't pertain to the subject matter they're supposed to support; or they simply don't exist at all. Anywhere. And if you don't look for or at them, you accept that they're real and talk to the legalistic points being put forth.*
I suspect that someday--and probably sooner than later--AIs will be a lot more capable than they are now.
* This isn't limited to AIs. Living, breathing humans do it all the time. It's been said that 87.6% of all statistics are made up on the fly. And then there's
Blessed relief for the pain and discomfort of percentorrhea. J Irreproducible Results, 21 (3) (1975), pp. 3-4. Which really exists.
lol - funny, and illustrative of the fact that AI is not Superman. At the same time, LLMs can do pretty impressive stuff. When I ask Grok about the correlation of the 10-year Treasury rate to the Fed funds rate over time, it gives a pretty sophisticated answer.
What's almost endearing is the LLMs' willingness to confess when confronted with mistakes. I've noticed that myself when I catch them in a non sequitur or contradiction. Much better than stonewalling, although that has happened in some famous cases. The way it stalls though is interesting.
Interesting world we live in.