J’ai refait ma page 404 avec #inkscape. J’aurais bien besoin de conseils pour que ça soit + joli. @noroanka@noroanka ;) https://si3t.ch/pub/err/dune.svg

Je ne sais plus si j’en avais parlé, mais j’ai écrit un petit script pour m’envoyer par mail les pages que j’aimerais lire plus tard. Pour éviter de les perdre, elles sont aussi enregistrée sur archive.org. https://si3t.ch/code/soulcoud/

J’ai mis à jour mes idées pour écrire du texte propre et lisible. Ça ressemble au gemtext, mais en pire 👼. https://si3t.ch/misc/notes/plaintext/

@New_scientist@feeds.twtxt.net Silicon Valley’s top AI models are terrible at almost everything. They only seem otherwise because people are easily fooled into believing they have capabilities they don’t have.

Silicon Valley’s top AI models are terrible at rebus wordplay puzzles
Rebus puzzles provide wordplay challenges involving both images and text, and they can confound Silicon Valley’s most powerful AI models ⌘ Read more

Google Chrome Gains AI Features Including a Writing Helper
Google is adding new AI features to Chrome, including tools to organize browser tabs, customize themes, and assist users with writing online content such as reviews and forum posts.

The writing helper is similar to an AI-powered feature already offered in Google’s experimental search experience, SGE, which helps users draft emails in various tones and lengths. W … ⌘ Read more

21:30, madame est enfin rentrée, ses supérieures lui demandent de compléter le projet d’école et préciser des indicateurs de réussite. Elle est épuisée. Je viens de finir les bulletins, les cours de demain tiennent à peu près la route. Il faudrait vérifier tous les papiers pour la sortie de la semaine prochaine. Et répondre aux parents inquiets pour leur enfant qui va mal (à juste titre). Mon aîné a de la fièvre. L’environnement brûle. Ma patronne (AOC) se fout de la gueule du monde. Alors ce soir, tant pis, ça sera One Piece, j’ai du retard à récupérer. Eichiro Oda, tu as la lourde tâche de me redonner foi en l’humanité. Seul ce qui nous différencie des animeux le peut encore. Il y avait les peintures rupestres, il y a la musique, le dessin et les récits. La création, l’art, voilà tout ce qui permettra à nos enfants de tenir.

Je n’ai plus d’accès à internet, si3t.ch est donc down. La #freebox est bloquée sur Etape 4. Un technicien a du débrancher le cable qui correspond à mon accès…

J’ai réalisé que #dokuwiki pourrait facilement être utilisé comme lecteur de flux oO https://www.dokuwiki.org/fr:rss

Bon, j’ai trouvé des recettes, mais pas de distributeur de malt près de chez moi :/

Voici un livre que j’aurais bien aimé avoir sous la main dans mon enfance, qui m’aurait évité bien des tracas, des questions, des gênes qui pour certains ne s’estompent que doucement avec l’âge (j’ai 34 ans…) : Le petit illustré de l’intimité. Il existe plutôt à propos des garçons, à propos des filles, mais aussi de la puberté, du consentement, de la sexualité en général. Je recommande! https://www.babelio.com/livres/Dieumegard-Le-petit-illustre-de-lintimite–De-la-vulve-du-v/1314908 https://www.babelio.com/livres/Baudy-Le-petit-illustre-de-lintimite–Du-penis-des-tes/1360904

Je viens de découvrir que dans #nethack, c’est une mauvaise idée d’enchanter une armure déjà à +3. Me voilà à poil dans le donjon, je n’ai plus de robe XD

AI made from living human brain cells performs speech recognition
A biocomputing system consisting of living brain cells learned to recognise the voice of one individual from hundreds of sound clips ⌘ Read more

Voilà, j’ai craqué. Je viens de montrer à mes élèves comment utiliser invidious à la place de youtube

Interesting thing happening over on Xitter. Apparently some of the women in tech accounts are being exposed as being run by men that hire women to pose for images/videos. They would be invited to tech conferences but would always drop out last minute.

Image

Makes me wonder if maybe there is need for a sort of verifiable web of trust is needed where influencers can be proven as authentic by others. This will only get worse as AI generative content gets pushed into our feeds.

Lorsque j’ai levé les yeux au ciel, La lune avait changé de visage. Pas tout à fait entièrement, comme si elle montrait une nouvelle partie de son visage. Notre satellite s’est-il tourné, ou bien la rotation de la Terre s’est elle accéléré? Quelqu’un a peut-être trouvé ainsi une solution au réchauffement climatique

@prologic@twtxt.net its not.. There are going to be 1000s of copy cat apps built on AI. And they will all die out when the companies that have the AI platforms copy them. It happened all the time with windows and mac os. And iphone.. Like flashlight and sound recorder apps.

@prologic@twtxt.net the new product was GPTs. A way to create tailored bots for specific use cases. https://openai.com/blog/introducing-gpts (fun fact: I did an internal hackathon where we made something like this for $work onboarding. And I won a prize!)

The competed project is poe https://quorablog.quora.com/Introducing-creator-monetization-for-Poe which is basically the same idea. Make a AI bot tailored to a specific domain of knowledge. And monitize it.

The timing fits very well as openAI announced it just a few weeks ago.

wtf is going on with Microsoft and OpenAI of late?! LIke Microsoft bought into OpenAI for some shocking $10bn USD, then Sam Altman gor fired, now he’s been hired by Microsoft to run up a new “AI” division. wtf/! seriously?! 🤔 #Microsoft #OpenAI #Scandal

Game-playing DeepMind AI can beat top humans at chess, Go and poker
An artificial intelligence capable of beating humans at a variety of games is an important step towards a more general intelligence, says Google DeepMind ⌘ Read more

@xuu@txt.sour.is Right now they’re laying the groundwork for uncritical belief in the power of #AI, so the next step will be accepting the magical incantations as if they were real.

@abucci@anthony.buc.ci predicting weather is literally a step up from the 3 body problem into n-body chaos. AI is just statistics pushed up into chaos. The future of computing is indistinguishable from magical incantations

The AI bubble is now upon us!

Image

How AI avatars of the deceased could transform the way we grieve
Companies are now offering chatbots that appear to come from beyond the veil. But psychologists say this “grief tech” may interfere with the patterns of brain activity through which we adapt to loss ⌘ Read more

DeepMind AI can beat the best weather forecasts - but there is a catch
By using artificial intelligence to spot patterns in weather data, Google DeepMind says it can beat existing weather forecasts up to 99.7 per cent of the time, but data issues mean the approach is limited for now ⌘ Read more

Hop, j’ai modifié mon script qui me permet de publier sur mon #twtxt et sur mastodon pour qu’il ajoute un lien vers une pièce jointe au besoin. Merci arpinux pour l’image listant les biais cognitifs https://0x0.st/HtaP.jpg

Avec la tempête, j’ai éteint mon serveur lorsqu’il y a eu une coupure de courant. L’électricité n’était pas de retour ce matin lorsque je suis parti aux #utopiales2023. Je l’ai désormais rallumé ;)

Je cherche un site qui me permette de calculer la distance parcourue en cliquant juste sur un plan. Pour l’instant, je n’ai rien de très pratique, ou alors j’ai mal cherché. #help

Et voilà, je me suis décidé à publier mes marques pages après un peu de tri. Il y en aura d’autres à l’avenir, petit à petit, car il y a certainement encore des trucs à découvrir (sous gopher par exemple) et d’autres que j’ai oublié : https://si3t.ch/pub/public_bookmarks.txt

Elle: Chéri, j’ai un problème, je ne sais plus où j’ai mis un fichier ni comment il s’appelle. Par contre, je sais que dedans j’ai écrit ‘cahier du soir’. Comment je peux faire? Moi > T’inquiète poulette, un petit script find+odt2txt+grep et c’est la fête. Elle > Ooooh merci tu es si fort et si beau ! (non)

Trop content: j’ai réussi à changer la lampe du clignotant arrière de ma voiture tout seul. Fallait démonter le phare. C’est tout bête, mais j’en suis fier :). (je déteste les voitures, ça coûte cher et c’est tout le temps en panne/vidange/CT

GPT-4 wins chatbot lawyer contest – but is still not as good as humans
Several AI chatbots were tested to see how well they could perform legal reasoning and tasks used by human lawyers in everyday practice – GPT-4 performed the best, but still wasn’t great ⌘ Read more

@prologic@twtxt.net Horseshit hype:

• AI that we have today cannot think–there is no cognitive capacity
  
• AI that we have today cannot be interviewed–“inter” “viewing” is two minds interacting, but AI of today has no mind, which means this is a puppet show
  
• AI today is not free–it’s a tool, a machine, hardly different from a hammer. It does what a human directs it to do and has no drives, desires, or autonomy. What you’re seeing here is a fancy Mechnical Turk
  

This shit is probably paid for by AI companies who desperately want us to think of the AI as far more capable than it actually is, because that juices sales and gives them a way to argue they aren’t responsible for any harms it causes.

@New_scientist@feeds.twtxt.net No, Google does not predict this. “Google AI” has been self-promoting like this for decades. Remember when they used to brag that they could predict the onset of flu season weeks before it started? That silently went away because they got it badly wrong many times and people caught on to how bad their “predictions” actually were.

They can’t stop themselves. Anything about AI coming out of big tech companies these days is marketing, not real, and certainly not science.

@New_scientist@feeds.twtxt.net because of course they have.

Emily Bender, a computational linguistic and excellent critic of this generative AI nonsense, uses an analogy of an oil spill to characterize what is happening as a result of generative AI. It’s polluting the world with false information, false images, false “academic” articles, false books. The companies that create this stuff are not cleaning up their misinformation spill; they’re letting the mess spread all over. It’s being used to commit crimes, and that’ll only get worse. Just like an out of control oil spill will destroy entire ecosystems.

Tricks for making AI chatbots break rules are freely available online
Certain prompts can encourage chatbots such as ChatGPT to ignore the rules that prevent illicit use, and they have been widely shared on social platforms ⌘ Read more

Google AI predicts floods four days early in South America and Africa
An artificial intelligence from Google can predict floods even in regions with little data on water flow, and its predictions four days in advance are as accurate as conventional systems manage for the same day ⌘ Read more

So in the wave of all things AI and this roller coaster we’re all on, apparently actors, writers and so on are all on strike. I don’t recall seeing anything in my feeds about this, so I had to ask a few folk in real life wtf was going on there…

Turns out they’re all on strike because they fear that AI/ML models will take over their jobs. There are numerous cases where “tech” has already replaced an actor, now it will just get much easier to do.

why am I not surprised?… https://uk.pcmag.com/ai/147757/elon-musk-will-train-his-ai-project-using-your-tweets

So given’s Google™’s recent policy changes where they now outright and blatantly just admit they’ll crawl, index and feed your (yes your fuckind) writings, thoughts, conversations, etc into their AI models; Should we as a small niche community (still growing) think about perhaps finally building Yarn.social v2 where we have encrypted feeds? 😅

@marado@twtxt.net It’s very different. Language models are part if traditional search engines and translation engines. The new policy mentions Cloud AI abd Bard specifically. This is a weird change and probably a good preemptive move as I said previously. I’m not sure why you’re downplaying it

@marado@twtxt.net It can’t possibly be defensible, which to me always signals an attempt at a power grab. They never explicitly said “we will use anything we scrape from the web to train our AI” before–that’s new. There is growing pushback against that practice, with numerous legal cases winding through the legal system right now. Some day those cases will be heard and decided on by judges. So they’re trying to get out ahead of that, in my opinion, and cement their claims to this data before there’s a precedent set.

Google Says It’ll Scrape Everything You Post Online for AI

Google updated its privacy policy over the weekend, explicitly saying the company reserves the right to scrape just about everything you post online to build its AI tools.

Google can eat shit.

@prologic@twtxt.net The hackathon project that I did recently used openai and embedded the response info into the prompt. So basically i would search for the top 3 most relevant search results to feed into the prompt and the AI would summarize to answer their question.

Crypto collapse? Get in loser, we’re pivoting to AI – Attack of the 50 Foot Blockchain

Someone on here gave me a hard time when I suggested that the crypto grifters were pivoting to AI after crypto collapsed. But, they were and they still are.

Seems to me you could write a script that:

• Parses a StackOverflow question
  
• Runs it through an AI text generator
  
• Posts the output as a post on StackOverflow
  

and basically pollute the entire information ecosystem there in a matter of a few months? How long before some malicious actor does this? Maybe it’s being done already 🤷

What an asinine, short-sighted decision. An astonishing number of companies are actively reducing headcount because their executives believe they can use this newfangled AI stuff to replace people. But, like the dot com boom and subsequent bust, many of the companies going this direction are going to face serious problems when the hypefest dies down and the reality of what this tech can and can’t do sinks in.

We really, really need to stop trusting important stuff to corporations. They are not tooled to last.

Dear Stack Overflow, Inc.

Stack Overflow is being inundated with AI-generated garbage. A group of 480+ human moderators is going on strike, because:

Specifically, moderators are no longer allowed to remove AI-generated answers on the basis of being AI-generated, outside of exceedingly narrow circumstances. This results in effectively permitting nearly all AI-generated answers to be freely posted, regardless of established community consensus on such content.

In turn, this allows incorrect information (colloquially referred to as “hallucinations”) and plagiarism to proliferate unchecked on the platform. This destroys trust in the platform, as Stack Overflow, Inc. has previously noted.

It looks like StackOverflow Inc. is saying one thing to the public, and a very different thing to its moderators.

“Sam Altman’s AI Hype Roadshow”

“The project of Altman and his merry band of doomsayers appears to be to capture power and create obfuscation by making new myths and legends”

“It assumes that no one will pull back the curtain and expose it as a market-expansion strategy”

Yes.

On Understanding Power and Technology

Debt Collectors Want To Use AI Chatbots To Hustle People For Money

Starting to get ugly already.

@obsidian-roundup@feeds.twtxt.net how many damn AI plugins does obsidian need? This shit is so annoying; it’s sucking the oxygen out of every other development effort.