Beep boop - this is a robot. A new show has been posted to TWiT…
What are your thoughts about today’s show? We’d love to hear from you!
Audio stops at 36:45
Edit: deleted and redownloaded and now it seems to be working again.
@leo was talking about the good and the bad of GLP-1 agonists. I recently found an excellent discussion of this topic: I’m a Pathologist: How To Boost GLP-1 Naturally (No Injections) . For anyone in a hurry, I uploaded a TXT transcript of the 20M video here.
I thought this would be a good time to kick the tires of Claude Sonnet 4.6. Claude also noted the unfortunate details with GLP-1 receptor agonists:
Natural GLP-1 is a peptide hormone secreted by L-cells in your gut in response to eating. It’s incredibly short-lived — it has a half-life of roughly 1–2 minutes in the bloodstream […]
Prescribed GLP-1 receptor agonists (semaglutide, liraglutide, tirzepatide, etc.) are engineered analogs that resist DPP-4 degradation. Semaglutide, for example, has a half-life of roughly 7 days, meaning it provides near-continuous, sustained receptor stimulation — something that never occurs naturally. This is intentional for therapeutic effect, but it’s a fundamentally different biological signal.
Beta Blockers have done a similar “jam the receptor” job with heart health since the 1970s. See the discussion here (same as the link above). Looking at the difference is an excellent way to understand how both the two different classes of drugs work. I then asked Claude to model these drug interactions with an impedance model. Here is Claude’s response. That is really good! Claude doesn’t completely agree with the premise, but Claude gives it the old college try anyhow. Ham Radio operators may like the connecting the dots between electronics and chemistry. Oliver Heaviside would be proud of us!
The GLP-1 mechanism is a signal. IMHO, the best analogy for GLP-1 agonists are jamming that signal. Our eventual resistance to these drugs is a meta-signal. As Claude notes in that first conversation, manufacturers of these drugs will “Not reliably [steer you to the best outcome], and structurally they cannot be expected to.” That’s a correct and well-tempered answer to the question. I like it, and I love Claude’s final conclusion.
Claude lacks one feature I really like in AIs; I asked it why it lacked that feature. Claude gave a bit of an incorrect answer. I dressed Claude down for its delusional answer, and I was satisfied with the correction.
I really like this AI. It gives thoughtful responses, and it responds appropriately when you catch it in an error. This is completely different from how AIs would give deluded responses a very short time ago.
There is no easy answer for our long-term health. After a very short couple of years, we’re already bumping up against the limits of these miracle drugs. We spend an awfully large amount of money on them, and the only real thing they provide is long-term reliance on the drug – just like the Beta Blockers. The GLP-1 agonists are excellent for the drug vendors and a giant “Meh” for the actual users. As Claude notes, Hijacking the GLP receptors to have them work with a half-life of days is fundamentally different than the way nature turns the signal on and off in minutes. There ain’t no cheese at the end of this maze. Thanks, Claude!
Regarding the XBOX discussion and the new head of XBOX coming from AI: it felt to me like there was a double standard being applied. One the one hand - no one believes her comment about not bringing AI “slop” into XBOX, and on the other hand - AI in game development for NPCs and maps and other things was seen as generally a good idea. So, which is it - is she going to make the service worse or better by using AI?
I’m also wondering why her claims of not bringing slop were not seen as credible by the panel. Am I missing something - does she have a history of lying or not doing things she said she would do? And let’s not forget that Matt Booty (unfortunate last name) is now the number 2 person there and I’ve heard positive things about him.
I don’t know - it just seemed like an unnecessary dog pile on Microsoft when it could have bene a substantive discussion on what she brings to the table.
It was a casual offhand conversation. The PC Gamer Article New Xbox boss promises no ‘soulless AI slop’ after moving over from Microsoft’s CoreAI products division was not even mentioned in the roundup for the show.
Agreed. Do you think the panelists were somewhat like NPCs in that discussion?
Fr. Robert Ballecer is so brilliant.
1 Like
Fr. Robert is one of the best. I feel like he was largely silent during that portion, but it could also be that the voices were similar.
2 Likes
“A 20-year-old woman says her life was ruined, ruined, I tell you, ruined” - and so began a discussion that deserved more compassion and nuance than it received by this panel. The most disappointing take was this dismissive flip-flop:
”I’ve got compassion for this woman and anyone like them because yes, we, we know that social media is addicting. Yes, we know that social media does change the brain, especially developing brains. However… this is adulting. You have to be an adult now.”
She was 6 when she and her parents put their trust in these platforms. The platforms took advantage of her.
A neighborhood family with 5 kids put their trust in books. AFAICT, it’s very rare these days for kids to read > 100 books a year. All of them do. The oldest is 16 now; they seem to have turned out fairly well.
I can’t quite imagine parents giving a 6yo an iPhone and unfettered access to those platforms. Did they think of also suing Apple? How many hours of advertisements on YouTube would the kid encounter in a week?