actual paper:

* https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0...

not sure if that sciverse website is "AI" but it does a really good roundup of interesting study news?

* https://thesciverse.org/feed/

sociopaths sometimes study people to learn how to emulate emotion

that's exactly the state of "AI" right now, it's cold, mathematical emulation

btw there are some fascinating papers on the concept that consciousness in humans is actually a quantum effect

brilliant Roger Penrose proposed it (and they thought he was nuts) but recent discoveries about microtubules make it plausible

so who knows, maybe a dozen exponential improvements in quantum computers could make "AI" really conscious next century

* https://www.youtube.com/watch?v=xa2Kpkksf3k

when I saw "NMDA receptor" I was immediately fascinated

I have long-covid and purposely take low-dose Dextromethorphan (just 15gm)

because it acts as a NMDA Receptor Antagonist

by blocking NMDA Receptors, it helps mitigate overactivation of chronic pain and fatigue pathways

it sounds like Andrew was experiencing the exact opposite effects by aggravating the pathways

* https://pmc.ncbi.nlm.nih.gov/articles/PMC7851375/

* https://images2.imgbox.com/0b/d7/AKg9AJg6_o.png

Two things that I would recommend trying out if you're interested in exploring this further:

1. If you're not paying for a model, the results will be worse. That sucks but the free access models are just not very good for anything where you need to trust the output, even for basic queries.

2. More important than #1 is access to tool use. If the LLM is just producing a nutritional breakdown from its weights, it's almost always going to be wrong. If the LLM is allowed to break the problem down into deterministic steps, it will do a lot better. In the nutritional breakdown case, an LLM with search + tool access can pretty easily break the problem down:

- Searching the web for a recipe or ingredient breakdown for the food

- Searching the web for nutritional qualities of each ingredient per some volume of the ingredient

- Writing and running a script with e.g. Python that takes in the recipe's projected serving output, the desired serving size, the amount of each ingredient etc, and scales the ingredients to match the desired serving size, and sums the nutritional qualities of the scaled ingredients.

I've tried this specific case with Claude + Gemini for my own purposes and they both handle it very well. The challenge currently is that the models will not always arrive at this approach when provided with an ambiguous prompt; sometimes they will, but sometimes they'll just vomit up a fully autocompleted response from their weights. Being more specific in the prompt or defining a skill that details the intended approach lets you get more useful + deterministic results while still taking advantage of the fuzzy glue that LLMs can provide here between steps.

Same with the classic strawberry r-counting case. IIUC LLMs have trouble with this because of how training data is tokenized, but any LLM will have no trouble farming out to e.g.

> echo -n "strawberry" | grep -o "r" | wc -l

> 3

There are basically two kinds of applications. One is where you want to correctly solve the problem at least 99 out of 100 times. LLMs generally don't (and not everybody realizes that) so there are a lot of debates and research around how useful and reliable they are or how to make them so.

The other kind of application is where you can try 100 times and you only need to be right once. Solving a mathematical research problem is like that.

I read somewhere this morning there is now more spending on datacenter infrastructure for "AI" in the US than all other infrastructure combined, roads, bridges, ship ports, etc.

Sounds plausible but I doubt it outmatches ICE warehouse concentration camp spending

Which is now the future of this country unless we force a course correction, by 2029 you'll drive down highways and it will just be one datacenter and ICE prison warehouse after another

I do not understand why you need as many GPUs powered up than people in the country or even a 1:10 ratio, it's all going to sit idle until they find something practical to do with "AI" other than entertainment purposes because it's not profitable, how are they going to monetize it, they cannot

yes but thorium reactors, not plutonium

once you learn about how much safer and controllable thorium reactors are you'll never vote for plutonium again

here's a great PBS SpaceTime on it

https://www.youtube.com/watch?v=ElulEJruhRQ

apparently they were also doing this going back a decade?

https://news.ycombinator.com/item?id=14772926

wild to see a 2017 post on the same subject

one thing I distinctly remember is that when simultaneous two-way Zmodem transfers came out replacing Ymodem, it absolutely blew my mind

(previously all transfers, Xmodem/Ymodem, were one-way with CRC checks on each block slowing things down)

My 8th-grade science project was doing a "statistical" analysis of X/Y/Zmodem transfers (and Kermit, I think?). It did well enough to get me to the county science fair here in Dallas, at least.

And yeah, Zmodem was mind-blowing for us.

zmodem/xmodem can still be used over ssh :) kermit is actually still in development.

I used it in like 2021 to transfer some data over serial from some really old laptop (386sx). Was faster than txp/ip over plip/slip

Ymodem-g (I think I remember that correctly) was faster than Zmodem - but if the CRC failed it aborted the whole send. Often I was willing to take the risk. (at 300 baud that adds up)

via ground telescopes no less, amazing

actual paper (paywalled)

https://www.nature.com/articles/s41550-026-02870-1