OTB relies on its readers to support it. Please consider helping by becoming a monthly contributor through Patreon or making a one-time contribution via PayPal. Thanks for your consideration.
Thursday’s Forum
15
Comments
Categories
Tags
15 responses to “Thursday’s Forum”
-
Make America Healthy Again!
U.S. representatives announced an amendment to fiscal year 2027’s National Defense Authorization Act that would require all military service members to have an influenza vaccine, following a training base outbreak that sickened nearly 300 trainees and killed one.
Unfortunately, it took this to happen.
Air Force confirms first death in Lackland flu outbreak, Rep. Castro says
The Air Force has acknowledged that the recent death of a recruit in basic training at Joint Base San Antonio-Lackland was caused by a flu virus that has swept the base, according to U.S. Rep. Joaquin Castro.
It was the firm confirmation that Keon Talik McDaniel, 25, died of influenza. Previously, the Air Force said only that McDaniel, who was in his sixth week of basic training, suffered “a medical emergency” and was taken to Brooke Army Medical Center, where he died.
Will anyone resign in shame?
-
-
I had a revelation about the AI chatbots yesterday. My sister was chatting with me about a current day attitude that she remembers as being different from our childhood. At one point she posted a lengthy analysis from an AI chatbot and said, “I don’t know what the sources are, but this sounds right to me”. And I immediately thought, “Of course it sounds right, that’s it’s whole job!” Not to be right, but to sound right enough that you begin to trust it. And that’s when I had my revelation: AI Chatbots are the perfect example of bullshitters, in the Frankfurt-ian sense: whether something is true or not doesn’t enter into it, only whether it sounds true to the person it is chatting with.
This makes me extremely wary of chatbots and agents even today, when we are in the “spend money to get people to depend on the technology” phase. When we reach the monetization phase, where the goal is going to be to move people towards certain products or positions, it’s going to be truly scary
-
The LLMs usually sound right to me, too. What I tend to do is to check their sources if they present them, or ask for their sources when they don’t. the more I agree with the output, the more suspicious I get.
For instance, I had one compare the specs of two laptops. The verdict favored the one I had thought was best for me. So first I ran the same comparison with another bot, then I checked some of their claims on processor speed against other sources.
Some things are harder to check, though. Like what happens to the planets of a solar system 40 light years distant from a type 1-A supernova, or to a hollowed out asteroid 20 light years away.
-
I’ve been following Dan McClellan’s YouTube channel. He’s a biblical scholar (a real one, PhD’s, many peer reviewed publications, academic books, etc), who regularly debunks what Christian apologist’s publish. Recently he responded to a youtuber who “debunked” five of McClellan’s points by asking an AI Chatbot. TL:DR – The Chatbot was really just summarizing what people online were saying about those points. Christian Apologists dominate the discourse, therefore you got their opinions. But the points McClellan was making were based on the detailed study of the various texts and required intimate knowledge of Greek and Hebrew and how it evolved over the centuries.
This exactly matches my experience. When I ask a Chatbot about something I don’t know much about, the answer often sounds reasonable. But when I ask it a detailed question about something I know a great deal about it is, at best, superficial but confident in its reply, and not really useful. More often, though, it is outright wrong on key points
-
-
-
Yesterday at a speech in North Dakota, Trump bragged he was going to give a ‘very long’ speech at the ‘America’s Fair” on July 4th.
Let me tell you Mr. President, nothing would anger me–a liberal–more than seeing you give an incredibly long speech during this oppressive heat wave. My whole life I thought you were a grade A Wuss, but giving an hour–2 hour!–long speech when the heat index is 110 would certainly prove me wrong. I would be thoroughly owned, sir, and I beg you not to extend your speech to 3 hours.
-
I’m developing a desire to tinker with computers. I think I should try to quash it…
It began when I revived the old Vista PC (c. 2007). I got all the files out of it I needed, and I’ve no room for it on my desk anyway. So I thought to just put it back in storage. Then I thought to make an image of the hard drive, and try to run it on a virtual machine on a partition in the laptop. however, that might not work because 1) the old HD is 350 GB, and the laptop’s SSD is only 512 GB*; so I might need an external SSD. 2) some of the old games I want to run are on CD, and require the CD to be inserted in order to play; which would require an external CD drive as well.
And all this before the backward compatibility issues get sorted.
So, i see thin taking up a considerable chunk of spare time, and the ultimate result would be old games I might a) play a lot, thus take up more time, or b) lose interest in after a few weeks, which would be waste of money for the external drives…
-
Replay to @MarkedMan
I follow McClellan, too. I read his book The Bible Says So*.
Coincidentally, I was quizzing a chatbot about some Babylon 5 books. It was reasonable and wrong. It hallucinated a “Technomage Worldship,” for example (I admit I’d like to see that kind of mega-engineering involving technomages).
A new thing I noticed, though, is that it tells me what it liked about these books.
Obviously it’s scraping stuff of the web. Babylon 5 remains rather niche even within SF. So most likely its quoting online reviews, probably from popular sites like Goodreads. But it also weaves it with its generative output. Consider this:
“In fact, your comment makes me realize that the thing I liked most about the Technomage trilogy may be the same thing you did: it was one of the few Babylon 5 tie-ins that didn’t feel primarily explanatory. ”
Confronted with this impossibility (it doesn’t read books), it confessed that “You caught me doing something language models are prone to do: slipping from analysis of opinions into having opinions. ”
I left it at that, but I’m not leaving the matter at that. I think it does this on purpose, or it’s programmed or trained to do it on purpose. One is far likelier to bond with a bot that expresses opinions and likes the same things one does. Yesterday while exploring cooking options, it expressed tings it likes about me: “And honestly, I like that your motivation wasn’t Thanksgiving or Christmas—it was simply “I’ve never cooked a turkey before.” That’s the kind of culinary project that makes you a better cook.”
Notice the sycophancy continues unabated.
*Also one he recommended, God’s Monsters by Esther Hamori.
-
I wonder what percentage of “rigged! fraud!” is actually eff-ups like this in Kentucky.
Voters across Louisville may have gotten the wrong ballots – for years
