arthwollipot
Observer of Phenomena, Pronouns: he/him
The best (?) one of those was dinsey.com.
ETA: that's now owned by, and redirects to, disney.com. Yes, I checked.
Merged Artificial Intelligence
- Thread starter Puppycow
- Start date
- Featured
ETA: that's now owned by, and redirects to, disney.com. Yes, I checked.
The good (I think): This can be useful in many ways: helping to improve text was just the start. It can create graphics, offer ideas for plans of many kinds, perform diagnostics, etc etc and that's just the beginning.
The bad: Oddly, after seeing several mistakes, I would never count on this for grammatical proofing. It also (despite claiming great ability in this respect) reports inaccurate financial information (like expense ratios for mutual funds).
The ugly: The increasing and long-term reliance we will have on this. "Hey AI, I don't want to think or do the work, do it for me." The wheels of idiocracy are in motion
The bad: Oddly, after seeing several mistakes, I would never count on this for grammatical proofing. It also (despite claiming great ability in this respect) reports inaccurate financial information (like expense ratios for mutual funds).
The ugly: The increasing and long-term reliance we will have on this. "Hey AI, I don't want to think or do the work, do it for me." The wheels of idiocracy are in motion
rjh01
Gentleman of leisure
One other problem it has is that if you use it to create some text and send it to someone they might be able to detect that used chat GPT to write it. This software creates text in a very strange style.
Yeah I find it's better if you start out with something and ask for suggested improvements.
It does have its quirks. When I try using it to create graphics, it seems to have an odd aversion to blue/gray tones....and the graphics almost always start out ridiculously "busy."
It does have its quirks. When I try using it to create graphics, it seems to have an odd aversion to blue/gray tones....and the graphics almost always start out ridiculously "busy."
arthwollipot
Observer of Phenomena, Pronouns: he/him
I like to use it (Copilot, which is built on top of ChatGPT) to quickly check facts or provide definitions. Which is ironic, since I occasionally need to fact check it. But I've been a skeptic for a long time and I've got a pretty decent sense for when it's hallucinating.
arthwollipot
Observer of Phenomena, Pronouns: he/him
I just had this exchange with Copilot.
Starts well. These are screenshots from my own browser, by the way.
To be fair, it's not wrong. There are indeed two Rs in "strawberry". There is also a third.
And it learns to count.
And corrects its original mistake. Okay, now let's start a new conversation.
Excellent. Well done.
Starts well. These are screenshots from my own browser, by the way.
To be fair, it's not wrong. There are indeed two Rs in "strawberry". There is also a third.
And it learns to count.
And corrects its original mistake. Okay, now let's start a new conversation.
Excellent. Well done.
lol
And there's the scary part, because unless it's an obvious mistake like that, pretty much everyone will take any AI's word as gospel. We are running headlong towards "Idiocracy" becoming a reality...
(PS: I had to try your strawberry test on regular ol chatgpt...sure enough, it failed again.)
And there's the scary part, because unless it's an obvious mistake like that, pretty much everyone will take any AI's word as gospel. We are running headlong towards "Idiocracy" becoming a reality...
(PS: I had to try your strawberry test on regular ol chatgpt...sure enough, it failed again.)
Ron Obvious
Graduate Poster
I was reminded of the movie Fletch:
"Is that one or two B's in your name, Mr. Babar?"
"One - B A B A R"
"That's two B's"
"Yes, but not not right together. I thought that's what you meant."
"Is that one or two B's in your name, Mr. Babar?"
"One - B A B A R"
"That's two B's"
"Yes, but not not right together. I thought that's what you meant."
Last edited:
Ron Obvious
Graduate Poster
Corrected. Brain fart.Don't you mean two Bs?
AI is that you?
Safe-Keeper
My avatar is not a Drumpf hat
You're making it sound as if human sources don't occassionally need fact checking
.Assumes facts not in evidence.
Last edited:
Don't think it does - humans overall do defer to those that speak or write fluently, we equate that with being "intelligent" whereas what the LLMs have shown is that ability alone does not require intelligence.You're making it sound as if human sources don't occassionally need fact checking
Strange my version of Copilot hasn't got that wrong in months which in LLM development terms i a long, long time. Same with the other old one "which number is larger 9.18 or 9.8?"
alfaniner
Penultimate Amazing
Not to mention, spell-checking.You're making it sound as if human sources don't occassionally need fact checking
Last edited:
Type of think I usually do!
Guybrush Threepwood
Trainee Pirate
I just tried chatgpt.com, it still thinks there are 2 Rs in Strawberry.
alfaniner
Penultimate Amazing
You should ask it "How many Rs are in 'a pirate'?"
The proper answer is of course, "As many as he can vocalize."
The proper answer is of course, "As many as he can vocalize."
arthwollipot
Observer of Phenomena, Pronouns: he/him
Exactly.You're making it sound as if human sources don't occassionally need fact checking
angrysoba
Philosophile
That's crazy. I thought they sorted that one out ages ago.
Hmmmm... I tried it and it came out correctly.
But it is still stupid. Look at this...
