News article and paper showing
bigger AI chatbots more inclined to spew nonsense - and people don’t always realize.
and some solutions
various emerging techniques should help to create chatbots that bullshit less, or that can, at least, be prodded to disclose when they are not confident in their answers. But some hallucinatory behaviours might get worse before they get better.
I used to parse PDFs using the Allenai method and the layoutparser.
This worked in many instances but is no longer maintained.
I still have Nougat on my to do list while a new paper now points to AceParse
AceParse includes various types of structured text, such as formulas, tables, algorithms, lists, and sentences embedded with mathematical expressions, among others. We provide examples of several dataset samples to give you a better understanding of our dataset.
From the creator of wordfreq
Generative AI has polluted the data
I don’t think anyone has reliable information about post-2021 language usage by humans.
The open Web (via OSCAR) was one of wordfreq’s data sources. Now the Web at large is full of slop generated by large language models, written by no one to communicate nothing. Including this slop in the data skews the word frequencies.
I wanted to allow a PDF download on my blog. Unfortunately this turned out to be complicated for multiple problems including WP and browser caching, CSS failure and mPDF misdirection. Here are the solutions.
Issue 1: If you are editing the print.css but the browser shows the old version: Comment out or delete the reference to parent in child css
/*
Theme Name: Twenty Twenty Child 1
Theme URI: http://wordpress.org/themes/twentytwenty
@import url('../twentytwenty/style.css');
*/
and include versioning in header.php
function my_theme_enqueue_styles() {
wp_enqueue_style( 'twentyfourteen', get_template_directory_uri() . '/style.css' );
wp_enqueue_style( 'twentyfourteen-child1', get_stylesheet_directory_uri() . '/style.css', array(), filemtime(get_template_directory() . '/style.css'), false );
}
add_action( 'wp_enqueue_scripts', 'my_theme_enqueue_styles' );
Issue 2: css editing is applied but printing is still not correctly formatted, possible due to higher specificity of formatting instructions in the parent theme.
Solution: This is a bit tricky as it needs try and error to find the offending element. Chrome can show the print version in the browser as shown on this SO thread
Then add the important flag to the specific css rule..
@media print {
.myelement1 { display: none !important; }
}
Issue 3: mPDF pulls the screen not the print version. Solution: Create an own mpdf media tag.
@media mpdf {
#sidebar, #header {
display:none !important;
}
blockquote, table, pre {
page-break-inside:avoid !important;
font-size: 0.7em !important;
}
}
Wir brauchen offensichtlich auch so ein “Kill Switch” Gesetz wie in Kalifornien das einen Notausschalter vorschreibt , wenn die Filter nicht mehr mitkommen undunethische Entscheidungen getroffen werden.
As we’ve previously explored in depth, SB-1047 asks AI model creators to implement a “kill switch” that can be activated if that model starts introducing “novel threats to public safety and security,” especially if it’s acting “with limited human oversight, intervention, or supervision.”
Nur – wann wird der Kill Switch aktiviert? Bilder wie die von Elon Musk’s X-Grok könnten wahlentscheidend sein.
Oh my god. Grok has absolutely no filters for its image generation. This is one of the most reckless and irresponsible AI implementations I’ve ever seen. pic.twitter.com/oiyRhW5jpF - Alejandra Caraballo (@Esqueer_) August 14, 2024
After changing the battery it needs to be calibrated
So the question was how to drain it most effectively? Maybe playing video would work but running this short line in the terminal as may times as there are CPUs was much more efficient
yes > /dev/null &
It seems that the initial high loading is most important
Researchers at SLAC-Stanford Battery Center report in a study published today in Joule that giving batteries their first charge "at unusually high currents increased their average lifespan by 50% while decreasing the initial charging time from 10 hours to just 20 minutes."
A recent stackexcange response confirms this
Apple Mail can make its Core Spotlight index available to other apps which include an app specific Mail Plugin. This is controlled (from the Apple Mail perspective) in the general preferences under Manage Plug-ins.
Unfortunately however the Houdah plugin stopped now working. So the only working solution seems InfoClick that can build its own index but takes 17 hours for 340.00 emails… 
Wafik S El Deiry has now uploaded more than 40 times a rebuttal letter to PubPeer complaining about being bullied by “academic terrorism”. Thomas C Südhoff is even more aggressive with new ad hominem attacks as I just learned
while his explanation of the numerous duplications is clearly wrong
… the tiny allegedly cloned areas of similar background signals partly overlap and are randomly distributed in the image. Besides the fact that it would make no sense to duplicate such small areas of background - a fraudster could just run a gel with empty lanes - and that such duplications do not improve the data, overlapping duplications like this are nearly impossible to manufacture.
Of course also small areas can be copied with the clone tool. If the placement is random or intentional can only be judged from the original image while an educated guess is certainly allowed. Running Photoshop is at least far more time and cost effective than running a gel with an empty lane.
A general problem here is that digital reproductions of images - both of immunoblots and of tissue sections or cells - can create artifactual microduplications especially if the image resolution is changed during reproductions.
This is outright wrong. Artifacts by capturing or stitching software is possible in theory while in practice we have found it only a few times.
So here comes my assessment of the now famous Synaptotagmin-1/Synaptopyhsin-1 immunoblot
The assessment is based on the directly extracted (inline) image from the PDF.
Identical patches were confirmed using 3 software packages Forensically, ImageDup and ICMF.
Also the manual annotation below shows 100% identical areas where the KW+ lane pixel has been copied to KW- (the other direction is less likely). Not sure what had been there, dust, dirt, text marker or another dot?
Südhof comments on this image on his website
Mistake identified: Dr. E. Bik claims that the Suppl. Figure 6b immunoblot stripes (reproduced digitally at low resolution by the journal from a non-digital original blot) contains tiny areas of microduplications in the background pattern (not the actual signal). These areas are tiny, within a blot, randomly distributed, and only digitally identifiable. She implies that these blots are suspicious and could be manipulated.
Resolution: This is an unusually bizarre accusation since it refers to digital low resolution images in which tiny image areas would have been scrambled by a person if Dr. Bik's accusation were correct. Even though she maintains publicly that she won't speculate about motivations, her accusations imply a motivation that would be difficult to understand since any manipulation here would produce a partly altered background. The most likely explanation here is, like for many of the 'mistakes' identified by Dr. Bik's A.I.-powered software, that these random microduplications are simply a reproduction artifact of a digitized image.
Classification: unfounded
Great story: The journal Nature Structural & Molecular Biology received the original blots and digitized them? So this is their fault? These are neither tiny spots, nor are they randomly distributed and of course, they can be seen by naked eye.
German newspapers covered the Südhof stor already (SPIEGEL, FAZ but also Science Magazine). Ulrich Dirnagel/Tagesspiegel believes that any intentional manipulation or deception cannot be recognized. I am not sure when looking at the images above.
We are being spied on and the data are being sold.
https://x.com/RuthieClems/status/1813478033975623697Report by The Chronicle
The two reports note that Informa will explore how AI can make its internal operations more effective, specifically through Copilot, Microsoft's AI assistant. "Like many, we are exploring new applications that will improve research and make it easier to analyze data, generate hypotheses, automate tasks, work across disciplines, and research ideas," a Taylor & Francis spokesperson wrote in an email to The Chronicle.
Publishers neither analyze data, generate hypotheses and work on research ideas – it is just a money making scheme after the
Another publisher, Wiley, also recently agreed to sell academic content to a tech company for training AI models. The publisher completed a "GenAI content rights project" with an undisclosed "large tech company," according to a quarterly earnings report released at the end of June.
mit interessanten Details zu ihrem Rose Diagramm auf historyofinformation.com
In 1858 nurse, statistician, and reformer Florence Nightingale published “Notes on Matters Affecting the Health, Efficiency, and Hospital Administration of the British Army”. … This privately printed work contained a color statistical graphic entitled “Diagram of the Causes of Mortality in the Army of the EastOffsite Link” which showed that epidemic disease, which was responsible for more British deaths in the course of the Crimean War.
Found at futirism.com
Google researchers have come out with a new paper that warns that generative AI is ruining vast swaths of the internet with fake content - which is painfully ironic because Google has been hard at work pushing the same technology to its enormous user base.
The study, a yet-to-be-peer-reviewed paper spotted by 404 Media, found that the great majority of generative AI users are harnessing the tech to “blur the lines between authenticity and deception” by posting fake or doctored AI content, such as images or videos, on the internet. The researchers also pored over previously published research on generative AI and around 200 news articles reporting on generative AI misuse.
The authors painfully collected 200 observed incidents of misuse reported between January 2023 and March 2024 and find
– Manipulation of human likeness and falsification of evidence underlie the most prevalent tactics in real-world cases of misuse…
– The majority of reported cases of misuse do not consist of technologically sophisticated uses … requiring minimal technical expertise.
– The increased sophistication, availability and accessibility of GenAI tools seemingly introduces new and lower-level forms of misuse that are neither overtly malicious nor explicitly violate these tools' terms of services, but still have concerning ethical ramifications.
Twitter was the #1 information source during COVID times but it is now officially declared dead in a new preprint “The Failed Migration of Academic Twitter“.
Utilizing publicly available user account data, we track the posting activity of academics on Mastodon over a one year period. Our analyses reveal significant challenges sustaining user engagement on Mastodon due to its decentralized structure as well as competition from other platforms such as Bluesky and Threads. The movement lost momentum after an initial surge of en- thusiasm as most users did not maintain their activity levels, and those who did faced lower levels of engagement compared to Twitter.
While the paper isn’t particular well designed – I would expect some Sankey plots showing the activity movement like those found after elections – the main message is clear. The paper summary also reflects my own experience as I am writing now only occasionally at my blog while my Twitter, Mastodon and Bluesky accounts are left in an orphan state.
Leider funktionieren die Onleihe Screenshots über GoFullPage in Brave nicht so schön wie in Chrome da zusätzliche CSS Elemente von Onleihe stören . Mit drei Regeln im Inhaltsfilter
kann man sie aber ausblenden. Die Zoom Buttons sind zwar dann auch weg aber der Zoom funktioniert weiter mit CRTL+ bzw CTRL-.
In another website constructed more than a decade ago, I integrated WordPress pages just as iframe .This was quite common at that time but not very elegant as all the formatting had to be adjusted to the parent site for a coherent view. When revising the website last week I have chosen a radical different approach by completely silencing the WP output using a plugin
<?php
/** Plugin Name: shutup **/
add_action( 'template_redirect', function() {
wp_die( __( 'disabled' ) );
});
?>
WP pages and posts are retrieved in the main website using the WP API only
$api_url = 'https://www.DOMAIN.de/BLOGNAME/wp-json/wp/v2/pages/xxxx'; $response = file_get_contents($api_url); $posts = json_decode($response); echo $posts->content->rendered;
ephorie.de has a convincing re-analysis that Daniel Kahneman is wrong, who purportedly demonstrated that emotional wellbeing increases with income but plateaus at $75,000…
.