Reminds me of the guy with Korsakoff’s syndrome from the Man who Mistook his Wife for a Hat.
Iirc, the nuns wonder if he is a “lost soul”
Actually it does seem to “know” that. I was asking about the Lorentz equation and how fast you’d need to go to fit in DP’s garage and it warned me against trying to get close to c for a couple of good reasons. But then it went into a repetitive spiral of responses that were like a stuck record player. But yeah it didn’t say “totally impractical, bht let’s think of it as a thought experiment…”
I read an article once about a condition where patients believe they are dead. I found it disturbing in a way. If an unhealthy brain is able to believe something that seems so fundamentally untrue, then does this mean our healthy brains are making us believe similarly untrue things about our own sense of identity?
I meant the hypothetical linear model does not know that the billiard ball cannot go the speed of light, because it is not built into its assumptions.
What I was getting at is that ChatGPT is unable to get at the “real” things behind the word “billiard ball”. It knows how that word relates to other words, but is unable to penetrate this “shell” and see the “real” billiard ball that the sign “billiard ball” points to.
Tl;dr: create solar energy in outer space, beam it down to earth, ???, profit
Are these the Jewish space lasers people keep talking about?
1 Like
Another paper claiming to have found room temperature super conductor. Different group, different materials than the recent retraction, but some crack potty word salad in article. Still would be cool if true. They wrote 2 preprints, first had 6 authors, 2nd has 3 = limit on Nobel prize recipients.
Sucks that it contains lead. But mega awesome if it works.
Update on fusion research. Looks like actual progress. 
2 Likes
(from twitter)
What’s your favorite tech innovation?
-
Illegal cab company 16%
-
Illegal hotel chain 12.2%
-
Fake money for criminals 38%
-
Plagiarism machine 33.9%
(I’ll add: Illegal gun making machine)
4 Likes
Google told DeepMind to generate its own prompts!
DeepMind came up tons of different prompts and they tested them all.
One big improvement was on Middle School Math (gsm8k)
We thought the best prompt was:
“Let’s think step by step” (70%)
But now we have:
“Take a deep breath and work on this problem step-by-step” (80%)
Forget about plagiarism or faked data – just bribe the editors to take the papers your paper mills generate!
https://twitter.com/briandavidearp/status/1749496902490501165
PAPER TRAIL
In the latest twist of the publishing arms race, firms churning out fake papers have taken to bribing journal editors
- 18 JAN. 2024
- 2:00 P.M. ET
- BYFREDERIK JOELVING, RETRACTION WATCH
One evening in June 2023, Nicholas Wise, a fluid dynamics researcher at the University of Cambridge who moonlights as a scientific fraud buster, was digging around on shady Facebook groups when he came across something he had never seen before. Wise was all too familiar with offers to sell or buy author slots and reviews on scientific papers—the signs of a busy paper mill. Exploiting the growing pressure on scientists worldwide to amass publications even if they lack resources to undertake quality research, these furtive intermediaries by some accounts pump out tens or even hundreds of thousands of articles every year. Many contain made-up data; others are plagiarized or of low quality. Regardless, authors pay to have their names on them, and the mills can make tidy profits.
But what Wise was seeing this time was new. Rather than targeting potential authors and reviewers, someone who called himself Jack Ben, of a firm whose Chinese name translates to Olive Academic, was going for journal editors—offering large sums of cash to these gatekeepers in return for accepting papers for publication.
“Sure you will make money from us,” Ben promised prospective collaborators in a document linked from the Facebook posts, along with screenshots showing transfers of up to $20,000 or more. In several cases, the recipient’s name could be made out through sloppy blurring, as could the titles of two papers. More than 50 journal editors had already signed on, he wrote. There was even an online form for interested editors to fill out.
1 Like
Maybe I should make a science fraud thread separate from this one…
Recently 5 papers were retracted by Harvard faculty and 31 are facing corrections. This was discovered by David Sholto in this piece. Here is what no one has done. We compiled a frequency table of the authors of the disputed pieces. Shown here.
What strikes me is that the names which appear on the most papers are NOT THE JUNIOR PEOPLE , but the PRINCIPAL INVESTIGATORS of the laboratories!
How can the PIs claim they are ignorant to the fraud if they are the commonality among the bulk of the papers? How can you say a junior person did it, when there is no junior person on all the papers? You either have to postulate a series of junior people who photoshopped, or…
So… a lot of photoshopping in their exhibits? The hell?
1 Like
They did a piece on the guy (David Sholto) who flagged the issues in the Guardian.
Interesting guy for sure. He has flagged over 2000 studies now (and damn, thats a lot).
3 Likes
Just don’t humiliate the profession by needing to retract it. I don’t want to see the name Meep on an unfortunate bar chart.
I’m making a separate research fraud thread:
I’ve often wondered about this…and what would happen to our language if we colonized a different planet.
tl;dr: developing a new accent appears to happen relatively quickly.
1 Like
I assume it changed toward the person whose accent was the most annoying or easiest to make fun of.
I also suspect that the same thing happened when euro’s colonized the Americas.
From 60 Minutes tonight…
video:
transcript (also with video):
2 Likes
That was the point of the problem as presented, as I understand it – prove the pythagorean theorem using only trigonometry…or something like that.
Part of the difficulty is that much of trig depends on the Pythagorean theorem, so using trig to prove it is, if you’ll pardon the pun, circular. But the law of sines, which is what they used, does not depend on the Pythagorean theorem.
1 Like