pmayrgundter

Interesting idea

The SR-72 is thought to be a real hypersonic from Lockheed flying since 2017. Extended reportage here:

https://youtu.be/QlMX6TYdU7I

Agreed.

I was told beforehand "Oh this is your first launch? Well, might ruin the others for you. They're like a firecracker going off. This is like a stick of dynamite". Indeed.

"In two seminal papers from the 1970s that described the accretion disks fueling supermassive black holes, scientists assumed that thermal pressure—the change in pressure caused by the changing temperature of the gas in the disks—played the dominant role in preventing such disks from collapsing under the tremendous gravity they experience close to the black hole. They acknowledged that magnetic fields might play a minor role in helping to shore up the disks. In contrast, the new simulation found that the pressure from the magnetic fields of such disks was actually 10,000 times greater than the pressure from the heat of the gas."

Here's a long-form presentation from study lead Phil Hopkins at

https://www.youtube.com/watch?v=JAiFCbRD0oI

Paper, including discussion of code

https://arxiv.org/abs/2309.13115v2

Ah that's super neat. Hadn't seen this before. Thanks!

Here's hoping :)

Robert Kuhn is a really impressive dude. I've been occasionally running across his videos on YT from these interviews. I'm very impressed that he's rolling it all up into a written research project as well.

"I have discussed consciousness with over 200 scientists and philosophers who work on or think about consciousness and related fields (Closer To Truth YouTube; Closer To Truth website)."

https://www.youtube.com/playlist?list=PLFJr3pJl27pJKWEUWv9X5...

https://www.youtube.com/@CloserToTruthTV/videos

Mechanical telepathy may be possible, and for me could possibly answer this question. You and they put on a device whereby you tap into their conscious experience, see through their eyes, hear the voices in their head...

Here's hoping launch costs fall quickly and it gets absorbed and rebuilt as a much larger station, Ship of Theseus style.

Does seem biased to not mention the prior US testing, which destroyed satellite Solwind P78-1 in 1985, leaving debris in orbit for 2 decades. (also suggestive of secret testing which may have left more)

There's also an earlier Soviet program described in the article, with many space impacts, tho debris extent isn't noted.

https://en.wikipedia.org/wiki/Anti-satellite_weapon#United_S...

This claim that these tests are easy for humans seems dubious, and so I went looking a bit. Melanie Mitchell chimed in on Chollet's thread and posted their related test [ConceptARC].

In it they question the ease of Chollet's tests: "One limitation on ARC’s usefulness for AI research is that it might be too challenging. Many of the tasks in Chollet’s corpus are difficult even for humans, and the corpus as a whole might be sufficiently difficult for machines that it does not reveal real progress on machine acquisition of core knowledge."

ConceptARC is designed to be easier, but then also has to filter ~15% of its own test takers for "[failing] at solving two or more minimal tasks... or they provided empty or nonsensical explanations for their solutions"

After this filtering, ConceptARC finds another 10-15% failure rate amongst humans on the main corpus questions, so they're seeing maybe 25-30% unable to solve these simpler questions meant to test for "AGI".

ConceptARC's main results show CG4 scoring well below the filtered humans, which would agree with a [Mensa] test result that its IQ=85.

Chollet and Mitchell could instead stratify their human groups to estimate IQ then compare with the Mensa measures and see if e.g. Claude3@IQ=100 compares with their ARC scores for their average human

[ConceptArc]https://arxiv.org/pdf/2305.07141 [Mensa]https://www.maximumtruth.org/p/ais-ranked-by-iq-ai-passes-10...

Yeah, I did a bit but thanks for the thought. Will again

Agreed the micro architecture and energies involved are.. challenging

I guess my thinking is that most of the assumptions that might have precluded small scales may have radically changed

I'm thinking of the lattice as more of a staging area for a downrange scatter, so no local accumulation of the energies. But no idea how practical that may be

The continued shrinking of accelerators has had me brainstorming on a compact nanofusion approach for awhile

https://docs.google.com/document/d/1B5maRx9w0ahQTqSe6UZ56rd0...

On my last comment I suggested the authors may not be familiar with Koza+Adami, but didn't realize the corresponding author is Stuart Russell, co-author of "Artificial Intelligence: A Modern Approach", with Peter Norvig.. the "The authoritative, most-used AI textbook, adopted by over 1500 schools." according to their site. https://aima.cs.berkeley.edu/

Whoops!

I totally didn't realize this until these comments. Neat!

I went digging in wikipedia.. the Backpropagation article was created in 2005 and yet the mention of association/derivation from the chain rule wasn't mentioned until 2014, through a borrow from the German article

https://en.wikipedia.org/w/index.php?title=Backpropagation&o...

It's funny, this kind of subtree mutation was looked at pretty deeply by Koza and Adamı in the 90s under the rubric of Genetic Algorithms, but with a slightly different optimization function

One ref in the paper to 2000 for GAs for fast generation of program trees, but that's missing the main show

Hope they're reading this and dig into those guys work

> do clique detection and downrank such citations.

Sounds like a great project for someone, especially someone in academia, to lead.

Call it the C-index, after Corchado, and test it against his network

Great book and influential to Levin as well. I agree this is where we're headed

Whoops, meant to say Tonga, the same event

Here's the study

https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/202...

The story above cites a confirmatory study for its ozone hole result and should also cite contradicting studies for its warming result.

There was a study out of NASA/JPL that made a correct and timely prediction of major warming a year in advance (i.e. arriving last fall)

It gave a forcing estimate of +0.15 Wm−2, *equivalent to half a decade of global human CO2 forcing* arriving mainly at 18-24 months after Jan 2022 (last fall) and lasting for 5+ years

I wonder if the moire materials can tuned as a dynamic metamaterial to absorb or dump a lot of charge, based on their rotational geometry. That charge plateau seems like a quite powerful trick

Great video, but actually talks about market takeover by medical cartels and their corruption of government

No mention of PE?

Agreed on the first part, but for LLMs not having correlated capabilities, I think we've seen they do. As the GPTs progress, mainly by model size, their scores across a battery of tests goes up, eg OpenAI's paper for ChatGPT 4, showing a leap in performance across a couple dozen tests.

Also found this, a Mensa test for across the top dozen frontier models.

https://www.maximumtruth.org/p/ais-ranked-by-iq-ai-passes-10...

That does seem to me to be demonstrating a global type of reasoning or generalization.

Also see the author's note that at least with Claude, they seem to be releasing about every 20 IQ points.

I'm curious about the framing of research like this.. "The poor performance of transformers on arithmetic tasks" (relative to what?) and how that informs the adjacent conversation on progress towards AGI.

Some say AGI has already been achieved, others that it's years or decades away. When I dig into the disagreement, it often partially depends on the perspective of how competent humans are on the tasks in question, with the optimists being, I think, more realistic about variance in human intelligence and the pessimists seeming to reserve the term "general intelligence" for possessing a nearly perfect suite of capabilities that many otherwise intelligent people practically don't have.

For example with arithmetic, this study cites another [Dziri et al. 2023], that says:

"For instance, humans can solve 3-digit by 3-digit multiplication arithmetic after learning basic calculation rules. Yet, off-the-shelf ChatGPT and GPT4 achieve only 55% and 59% accuracies on this task, respectively."

But this isn't the case.. 5-6% of the population have https://en.wikipedia.org/wiki/Dyscalculia, but can be otherwise normal.

I still see value in normative statements about human capability in AI & AGI research, but I think we'll need to move towards explicit statistical framing.

DeepMind's Position paper "Levels of AGI for Operationalizing Progress on the Path to AGI" has a schema like this, where AGI capabilities are defined across 2 axes of Performance level X Generality (narrow vs general), and the Performance levels are measured by comparison with "Percentile of skilled adults" able to perform the task.. https://arxiv.org/pdf/2311.02462#page=3.40

Within that framing, this paper's title or result might be "Achieving AGI Competency in Arithmetic", or "Expertise", or "Virtuosity", i.e. on par respectively with 50th, 90th or 99th percentile of skilled adults.

A friend who studied engineering at MIT made a similar distinction, and it's where I expected the article to go, being hosted there.

The notion was that the development of engines, as opposed to simpler motors and mechanical systems, had required a net set of skills that led to a recognizably different discipline. More systematic, more concerned with dynamics and feedbacks, cross-disciplinary to materials and machining techniques, more data driven, more advanced mathematic and analytic techniques in support of these. The sense was simpler mechanical systems were more linear and engines were highly non-linear, and required a different mindset to approach.

I like the distinction and appreciate the different terms.

Thanks for the work guys! It's been growing in quality of the years and is one of my regulars.

Is this product spam? List looks legit, but clicking on one launched some popover that disappeared and then no paper

It's all about whip cracking! Who knew that was such a thing. Like the mindless honking horn of the past. Thanks for sharing :)

The title is also misleading in that the paper (https://www.pnas.org/doi/10.1073/pnas.2319652121) doesn't claim to determine when the fastest rates of CO2 rise have been. In fact, it starts by saying cores generally can only support analysis of durations >100 years.

I take note of this because I've looked at this problem before and agree that core samples have a hard time supporting claims like "most in the previous XXkya" when compared with the last few decades, be it CO2, CH4, avg temp d02, etc.

Starting with layers as recent as 30kya, the age of the trapped gas often spans multiple centuries, and so e.g. you'd be averaging out point outgassing from even very large volcanoes well beyond detectability. Beyond 50-60kya, the layer durations are all in the few millennia range.

Here's my analysis worksheet:

https://docs.google.com/spreadsheets/d/1p793c3NZFp2e80ufqyJK...

Reference data: Antarctic Ice Core 155,000 Year CO2 and Gas Stable Isotope Data - Eggleston, S.; Schmitt, J.; Bereiter, B.; Schneider, R.; Fischer, H. 2016. Evolution of the stable carbon isotope composition of atmospheric CO2 over the last glacial cycle. Paleoceanography, 31. doi: 10.1002/2015PA002874 https://www.ncei.noaa.gov/access/paleo-search/study/19942

Adjacent work from Fields & Levin 2021 on the thermodynamics of cellular process, arguing that due to the information processing demands and classical thermo minimal power requirements per operation for loose estimates of protein control, aggregate cellular metabolism power requirements are off by 10-15 orders of magnitude of available. They then conclude that cellular processes are using quantum coherent processes for major work internally and externally (p17).

https://chrisfieldsresearch.com/quantum-cells-pre.pdf

Thanks anyways