This person also “asked ChatGPT” if another writer on the panel went by her middle name or her first initial plus middle name,” instead of just… asking her
The other writer didn’t make it to the panel, unfortunately, but this person also mistook me for her (we are both Chinese American women)—even after I’d introduced myself—so it was *kind of* like she was there anyway
There is something here about the dogged insistence on clinging to “facts” from generative AI and trying to overwrite reality to match them, even when those “facts” are clearly untrue
it is just a glorified predictive text generator! Like if I start mashing whatever the first suggested word in my phone's keyboard, it might start out coherent but eventually the same Day service and the other day and I will be in the morning and then we can do it for you to the same Day service and
It drives me crazy when boosters say "But it'll get better!" That implies it's any good NOW. It's not. Technologies that "get better" at least start from a position of being good at something.
Plus, getting better dosn't mean it'll ever become good.
Automated driving systems are also "Getting Better" they still keep fucking up or actively trying to kill people.
Autocorrect is getting better but it still fucks up every 20th word and completely choaks on technical language.
I asked CoPilot to give me the Phillies lineup for that day's game. Never did get the right answer, but did get the interesting information that two players were playing first base, which CoPilot confirmed when I questioned it.
So you went through the entire panel with everyone under the assumption that you were somebody else? I don’t blame you for being too stunned to correct her but she should be so embarrassed.
Nonono, when I realized this person’s error (right before the panel) I corrected her, which was horribly embarrassing for both of us! I did have a fleeting thought that maybe I could just not say anything but then I realized that I couldn’t do the panel if she thought I was the other writer!
I do love that ChatGPT has to put a disclaimer basically saying "I'm pulling this from my digital ass, so if you believe me that's on you" to avoid the lawsuits.
Basic AI literacy is hampered by the media breathlessly reporting on what AI “might” be able to do one day instead of what it actually can do (write some extremely shitty prose, create some unsettling images) today
I like how the output from these things always feels like a fifth grader who’s trying to bullshit through a report on a book they absolutely did not read.
It's probably also in his will, along with the creation of the John Scalzi Memorial Golden Burrito Award For Excellence in Science Fiction or Burrito Making.
I love the idea of an award like this. "I'm sorry, you wrote a heck of a great SF novel and deserve all the possible awards but someone made a burrito with shrimp, hot dogs and Oreos. We HAVE to give them the Scalzi"
I was curious about the burrito bar thing, did some searching, and was shown this Rolling Stone article from 2016 about you looking for your quote in No Man's Sky. I hadn't known about it (the article). I've been playing NMS since launch so this was a fun discovery. It's a rather different game now.
I once asked GPT to list my academic publications. It returned a list of books and articles that don’t exist, many from journals and publishers that also don’t exist. The *actual* list is readily available on Google Scholar.