C'mon, ChatGPT, do better.
-
Mine is similar but not identical.
And yes, Ax, I’m a paying customer.
There have been no reports indicating that JD Vance was injured at a rally in Pennsylvania. The confusion may stem from a recent incident where former President Donald Trump was shot at a rally in Pennsylvania. Trump was grazed by a bullet on his ear, but JD Vance was not harmed during this event. Security has been heightened at subsequent rallies, including those involving Vance.
-
@George-K said in C'mon, ChatGPT, do better.:
That's not really how ChatGPT works. At least not yet. It's not Google.
The bigger the data pool it can reference, the better (albeit more milquetoast) the output is.
It's bad with current events because (1) it has to be updated and (2) the "Julius Caesar" dataset dwarfs the "JD Vance's connection with the Donald Trump assassination attempt that happened less than a month ago" dataset.
-
@jon-nyc said in C'mon, ChatGPT, do better.:
Google seems to be fine now.
"Now."
Google/Alphabet testified that it was an "algorithmic error."
Alphabet Inc.’s counsel informed the House Judiciary Committee that bugs in Google’s autocomplete tool prevented it from predicting searches about the attempt on Trump’s life. The built-in protections that Google installed for searches related to political violence were “out of date,” the attorney said, and prevented the search autocomplete feature from generating results on the assassination attempt against Trump three weeks ago.
Google’s autocomplete feature experienced similar issues when users searched for “President Donald” and related search terms. The attorney said the bugs were fixed after they were brought to Google’s attention.
Google also claimed that an algorithmic error was responsible for broadcasting news stories about Vice President Kamala Harris, Trump’s 2024 rival, when users searched Trump’s name.
"Out of date?" How does that work?
-
Google says they are aware of the problem and are fixing it.
-
Pretty much.
-
@George-K said in C'mon, ChatGPT, do better.:
Tried it on my Mac - confirmed.
That doesn't necessarily mean anything.
It could be that for every "bias" you can detect towards one side you got three examples of bias for the other side that you never hear about.
If people just look for examples of one kind of bias, they'll find those examples.