Tag: AI point-missing
-
Duane does Bud
Duane Jones dealt with the Bud Light problem. I cited it before, in a discussion of AI misunderstanding human motives: = = = = = START REPRINT: The Jones agency had produced a successful magazine ad for Heinz ketchup. A painting showed an elegant waitress holding a silver tray with a single bottle of Heinz,…
-
Not the best argument
Via MindMatters as usual: New AI systems have carved inroads into many industries, not least of all those involving voice and audio. Now the audiobook business is in trouble; since AI has the increasingly good ability to mimic the human voice and generate words, many voice actors and readers have watched the demand for their…
-
Who’s the computer?
The Turing test is hopelessly outdated. Eliza passed the test in 1964 for all practical purposes. Dembski is now applying the test backwards. He presents a weird Godel-like chain of convoluted self-referential sentences, and finds that ChatGPT can’t “solve” the chain. Normal humans can’t “solve” this totally contrived and totally trivial “problem” which doesn’t NEED…
-
Who’s the bedbug 2
Latest pithy point from Kirn: It takes only moments for AI to “write” a book, but it will still take hours, even days, for a person to read one. It used to be the other way around; the writer labored far longer than the reader. It made reading books feel worthwhile, to know that. Now?…
-
Turing test?
Old joke: Joe was a simple-minded guy who lived in a small town. He was the town mascot, mowing yards and shoveling snow for a dollar here and there. Everyone liked him, so they made sure he had enough to eat and a place to sleep. One day the city decided to make the arrangement…
-
Cofounderus interruptus
AI leader Hinton is getting credit for quitting Google and warning us about the problem, now that AI is fully unleashed and destroying all creative skills. JESUS H CHRIST. If you want to prevent conception you need to pull out before you squirt. Pulling out after the baby is born won’t help. I’m terminally tired…
-
Wrong end of the variable
From Above The Line: = = = = = START QUOTE: For weeks, multiple studios and streamers have been planning to use AI to generate scripts based on books and other IP that is in the public domain, with lists of titles making the rounds among development executives, multiple insiders have told Above the Line.…
-
Inexcusable
Several articles are noting ChatGPT’s LIBELOUS habit of mixing and matching biographies. For instance, Brian Moore, an Australian mayor, is suing OpenAI for slander. In 2008 he worked in banking, and reported a money-laundering scheme to the authorities. ChatGPT insists with the full weight of authority that he was the CRIMINAL, not the WHISTLEBLOWER. This…
-
So close…
Leah Libresco is worrying about AI cheating. There’s an important and valid worry in this area, but the cheating itself is not the serious problem. These companies use machine-learning models to count how many times per minute a student blinks or shifts his gaze (while taking an online exam, for example); they then compare this…
-
What would Steinbeck think?
Peter Biles at MindMatters uses an old literary interview with Steinbeck to speculate on Steinbeck’s views of AI. Biles concentrates on the way an author writes, and concludes that Steinbeck would dislike the way AI writes. There’s a much bigger and simpler answer. I was reading Steinbeck constantly for most of my early life. I’m…
-
Lot of silliness
Reading some comments on AI. Some were hopelessly behind the times. “What if AI passes the Turing test?” “What will happen when we give AI the nuclear codes?” Already passed the Turing Test, already has the nuclear codes. The famous story of Blessed Hero Stanislav Petrov proves it. In the early ’80s both sides were…
-
Just brand competition
Headline: 1000 leading demons want to “pause” development of AI. The big-name demons are Elon, Woz, Andrew Yang, and WEFoid Harari. The rest of the names aren’t familiar, but MANY of them are in charge of IPOs using AI. Others are professors of “ethics”, which means eugenics. I don’t recognize any humans on the list.…
-
Maybe you shouldn’t
MindMatters has been focusing on the problems and failings of AI for a long time, sometimes missing the point and sometimes hitting it. I’ve been ‘reacting’ to their hits and misses for a long time, as a way of organizing and rationalizing my own thoughts. For some reason their host Discovery.org has a VERY SLOW…
-
Well, here’s the “use case”
MindMatters notes that a sci-fi magazine stopped taking submissions from new writers because most were obviously AI. Automatic checking isn’t reliable, and there’s no way of firmly verifying human action without getting expensively personal. You could require manuscripts in pen and ink, and require a video showing the author actually writing a few of the…
-
Not worried
Headline at Bari Weiss’s neocon opinion page: The rise of deepfake porn: AI can turn anyone into a porn star without their knowledge. That reality should frighten us all. Well, that’s one AI threat I definitely don’t need to worry about. Guarangoddamnteed.
Polistra's Mill 