Show newer

<strong>Ancient Texts That Were FAKED</strong>

"_A close look at some of the more interesting (and infamous) ancient texts that were discovered and then found out to be forgeries._"

length: fifty eight minutes and twenty one seconds.

youtube.com/watch?v=7fLsbEWrRv

@archaeodons @antiquiodons

I’m more certain than ever that the first quarter of 2024 will see a grand reckoning for companies that have bet big on #AI. As I said before, I predict a lot of CFOs getting sticker shock at their compute bills, and we will see pivots and downscaling of expectations across the industry. By the middle of the year, there will likely be a general cool-down as users and investors come to grips with what “AI” can realistically do at reasonable cost (hint: not as much as the hype claims). Strategically deploying smaller models will become quite attractive, rather than monolithic solutions.

I also think 2024 is the year of the legal reckoning for the industry. Creators whose work have been used for training (i.e. plagiarized) will likely make big inroads into establishing legal frameworks for compensation, and some models will become poisoned because they were trained with unvetted data. Hopefully this also means that model-makers who have been meticulous about their training data’s providence will reap rewards.

How does electrical current move through water? Here's a little movie of it, made by Mark Petersen. A positively charged proton gets passed from one molecule to another!

This is called the 'Grotthuss mechanism' because Theodor Grotthuss proposed this theory in his paper “Theory of decomposition of liquids by electrical currents” back in 1806. It was quite revolutionary at the time, since ions were not well understood.

Something like this theory is true. But in fact all the pictures I've shown so far are oversimplified! A hydronium ion is too powerfully positive to remain a lone H₃O⁺. It usually attracts a bunch of other water molecules and creates a larger structure!

(2/n)

Show thread

If you could watch an individual water molecule, about once in 10 hours it would do this!

As it bounces around, every so often it hits another water molecule hard enough enough for one to steal a hydrogen nucleus - that is, a proton - from the other!

The water molecule with the missing proton is called a hydroxide ion, OH⁻. The one with an extra proton is called a hydronium ion, H₃O⁺.

This process is called the 'autoionization' of water. Thanks to this, roughly one in ten million molecules in a glass of water are actually OH⁻ or H₃O⁺, not the H₂O you expect.

And this explains why water conducts electricity so well. Let's watch.

(1/2)

"In this paper, we formalize the problem and show that it is impossible to eliminate hallucination in LLMs."
Hallucination is Inevitable: An Innate Limitation of Large Language Models
arxiv.org/pdf/2401.11817

<strong>The Genomic History of the Bronze Age Southern Levant</strong>

"_We find that Levant-related modern populations typically have substantial ancestry coming from populations related to the Chalcolithic Zagros and the Bronze Age Southern Levant. These groups also harbor ancestry from sources we cannot fully model with the available data, highlighting the critical role of post-Bronze-Age migrations into the region over the past 3,000 years._"

doi.org/10.1016/j.cell.2020.04

@archaeodons

<strong>The origin of every European country's name</strong>

"_attempt to explain the origin of every European country's name in English!_"

length: twenty three minutes and thirty nine seconds

youtube.com/watch?v=2Pcv2ySMi4

🇳🇴 <strong>Inside the Extreme Plan to Refreeze the Arctic | WSJ Future of Everything</strong>

"_A method normally used to create ice-skating rinks is now coming to the rescue of melting sea ice in the Arctic. Since satellite records began in 1979, summer Arctic sea ice has shrunk by around 13% per decade. Could making more ice be a potential solution to this issue?_"

length: eight minutes and eighteen seconds.

youtube.com/watch?v=7ebVUj2lh9

@science @climatechange @environment

@UlrikeHahn @ai Once again in agreement the sterile conditions of an academic setting do not always best represent the wide breadth of human responses.

@UlrikeHahn @ai In agreement with you regarding an empirical basis needed for assumptions.

@bibliolater @ai the other thing maybe worth note is that taking participant responses in the highly constrained pragmatic context of a psych experiment to reflect the full breadth of human responding is a bit misleading. It’s an interesting question how LLMs without pretraining or finetuning respond to experimental questions, but a better comparison might be with data from asking those questions of random shoppers at a mall if “breadth of answer” is of interest

@bibliolater @ai minor comment: the LLM data are not being compared to multiple responses by a single person on the same task as that is not a general feature of the primary human experimental literature involved. So, as far as I can make out, the levels of human self-consistency are simply imputed/assumed. Doesn’t mean the difference isn’t there, just that the empirical basis seems somewhat anecdotal.

<strong>(Ir)rationality and cognitive biases in large language models</strong>

"_First, the responses given by the LLMs often display incorrect reasoning that differs from cognitive biases observed in humans. This may mean errors in calculations, or violations to rules of logic and probability, or simple factual inaccuracies. Second, the inconsistency of responses reveals another form of irrationality—there is significant variation in the responses given by a single model for the same task._"

Macmillan-Scott Olivia and Musolesi Mirco. 2024 (Ir)rationality and cognitive biases in large language models R. Soc. Open Sci. 11: 240255. doi.org/10.1098/rsos.240255

@ai

U.S. #TreasurySecretary #JanetYellen will warn that the use of #AI in #finance could lower transaction costs, but carries "significant risks," according to excerpts from a speech to be delivered on Thursday.
"Specific vulnerabilities may arise from the complexity and opacity of AI models, inadequate risk management frameworks to account for AI risks and interconnections that emerge as many market participants rely on the same data and models," - #Yellen
reuters.com/business/finance/y

RNA has taken center stage, with remarkable versatility, such as >400 drugs approved and many more to come.
My conversation with Tom Cech, Nobel laureate, pioneer of RNA biology, with a new book called THE CATALYST, w/transcript
erictopol.substack.com/p/tom-c

Talks are now available at ias.edu/video/new-bounds-large ias.edu/video/new-bounds-large

The new progress can be summarized by the following chain of implications (see James' talk for some more detail):

Riemann Hypothesis
||
v
Lindelof Hypothesis
||
v
Density Hypothesis
||
v
Guth-Maynard zero-density estimate <- we are here
||
v
1940 Ingham zero density estimate
||
v
"Trivial" bound from Riemann-von Mangoldt formula

Show thread

@Smrki I agree that deindustrialization is not only problematic but myopic and counterproductive in the long run. It has not worked out well for many.

🇳🇴 <strong>Norwegian farmer finds 1,000-year-old Viking sword on family farm</strong>

"_X-ray scans of the sword revealed outlines of rare inscriptions with a cross pattern and the likely presence of letters on the blade._

_Based on these inscriptions, scientists said it could be a so-called Vlfberht sword, produced during the Viking Age or the early Middle Ages between 900 and 1050AD._"

independent.co.uk/news/science

@science @archaeodons

Show older
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.