#generativeAI

These are public posts tagged with #generativeAI. You can interact with them if you have an account anywhere in the fediverse.

Anthony
"AI" is Google's "pivot to video" moment:

Google AI Search Shift Leaves Website Makers Feeling ‘Betrayed’

The now-ubiquitous AI-generated answers — and the way Google has changed its search algorithm to support them — have caused traffic to independent websites to plummet, according to Bloomberg interviews with 25 publishers and people who work with them.

From https://www.bloomberg.com/news/articles/2025-04-07/google-ai-search-shift-leaves-website-makers-feeling-betrayed

Remember when Facebook told everyone they should change all their content to video, because it got more traffic? And then that turned out to be such a blatant falsehood that companies went bankrupt trying to do this?

#AI #GenAI #GenerativeAI #LLM #Google #Gemini #AISlop
Google AI Search Shift Leaves Website Makers Feeling ‘Betrayed’

In March 2024, website owner Morgan McBride was posing…

Bloomberg
Apr 17, 2025, 12:49 · · · 2 · 0
Harald Sack

Interesting (short) paper of game-based training and evaluation of agentic behaviour in LLMs: Leon Guertler, Bobby Cheng, Simon Yu, Bo Liu, Leshem Choshen, Cheston Tan.: "Textarena"

arxiv.org/html/2504.11442v1

#llms #AI #generativeai #agents #agenticAI #evaluation

TextArena Soft-skill comparison. Frontier models and Humanity are compared across ten key skills. Each skill is normalised separately for presentation. The image shows a radar chart comparing the capabilities of different AI models in various areas. The image is an octagon radar chart that compares the performance of four different AI models (Humanity, claude-3.5-sonnet, qwen-plus, and gpt-40) in eight different categories: logical reasoning, memory recall, bluff detection, theory of mind, pattern recognition, spatial reasoning, strategic planning, and adaptability. Each category is represented as an axis, and the performance of each model is represented by a point on the axis, which is connected by a line to the points in adjacent categories. The axes are labeled with the categories. The models are represented by different colors (orange, purple, turquoise, and dark red), with the legend at the bottom assigning each model to the corresponding color. The areas between the lines create a polygon whose shape and size visualize the respective performance of the model in the individual areas. The overall atmosphere of the image is factual and informative, with a clear and easy-to-understand presentation of the data. The color scheme is neutral and contributes to the clarity of the image.
Apr 17, 2025, 09:56 · · 3 · 0
Simon D. ⏚

#Apple imite les données des utilisateurs pour entrainer ses #IA
next.ink/180977/apple-imite-le

#Meta va de nouveau entrainer ses IA sur les données des utilisateurs européens
next.ink/180850/meta-va-de-nou

Après avoir activé l'accès à Meta AI en Europe, l'entreprise de Marc Zuckerberg remet en route l'entrainement de ses IA avec des données des utilisateurs européens de ses réseaux sociaux #Facebook, #Instagram et #Threads. L'opposition est possible via un formulaire. #WhatsApp n'est pas concerné.

#AI #generativeAI

Apple imite les données des utilisateurs pour entrainer ses IA - Next

L’utilité et les performances des services dopés à…

Next
Apr 17, 2025, 08:14 · · 1 · 0
Miguel Afonso Caetano

"Let's say out the numbers precisely:

- Incoming monthly revenue: roughly $425 million, give or take.

- Theoretical revenue from Softbank: $250 million a month. However, I can find no proof that SoftBank has begun to make these payments or, indeed, that it intends to make them.

- Liquidity:

- $10 billion that it is yet to receive from SoftBank and a syndicate of investors including Microsoft, potentially.

- An indeterminate amount of remaining capital on the $4 billion credit facility provided by multiple banks back in October 2024, raised alongside a funding round that valued the company at $157 billion.

- As a note, this announcement stated that OpenAI had "access to over $10 billion in liquidity."

- Based on reports, OpenAI will not have access to the rest of its $40bn funding until "the end of the year," and it's unclear what part of the end of the year.

We can assume, in this case, that OpenAI likely has, in the best case scenario, access to roughly $16 billion in liquidity at any given time. It's reasonable to believe that OpenAI will raise more debt this year, and I'd estimate it does so to the tune of around $5 billion or $6 billion. Without it, I am not sure what it’s going to do.

As a reminder: OpenAI loses money on every single user."

wheresyoured.at/openai-is-a-sy

#AI #GenerativeAI #OpenAI #ChatGPT #Microsoft $SoftBank #AIBubble

OpenAI Is A Systemic Risk To The Tech Industry

Before we go any further: I hate to ask you to do this,…

Ed Zitron's Where's Your Ed At
Apr 17, 2025, 02:55 · · 1 · 0
Mark Carrigan

Academic writing has always been in flux

It can feel when reading academics discussing LLMs that previously settled practices have been suddenly upturned by the introduction of this strange technology into higher education. The reality is that our practices of writing and communication have been through many such changes, often within the span of an individual’s own career. I was reaching the midpoint of a PhD when social media came to be a prominent feature of academic life, offering a potent forum through which to connect with others and discuss ideas alongside an ever present possibility of distraction. During the same PhD I remember talking to my supervisor about producing an 800 page book on a type writer. I simply couldn’t understand how such a thing was possible. Much as I struggled more recently when reading Lamott’s (1994) description of repairing a three hundred page manuscript by placing it on the floor of a cavernous living room in order to reorganise it page-by-page:

“I put a two-page scene here, a ten-page passage there. I put these pages down in a path, from beginning to end, like a horizontal line of dominoes, or like a garden path made of tiles. There were sections up front that clearly belonged in the middle, there were scenes in the last fifty pages that were wonderful near the beginning, there were scenes and moments scattered throughout that could be collected and written to make a great introduction to the two main characters. I walked up and down the path, moving batches of paper around paper-clipping self-contained sections and scribbling notes to myself on how to shape or tight or expand each section in whatever necessary way.” (Lamott’s 1994: 100).

It’s not that I couldn’t do this with my own text. While I’ve still not had reason to find out how to print at my university, in my fourth year since starting to work there, I’m sure I could quickly print out this text if I was motivated to do so. I remember the feeling of holding my PhD thesis in my hands the first time I printed out the draft, suddenly feeling a sense of mastery over this diffuse thing which had been the horizon of my experience for so long. I can recognise the appeal in the physical, the ways of relating to ideas opened up when we get our hands on their material expressions.

It’s just that I struggle to imagine relating in such a physical way, even allowing for the fact that I would undoubtedly be printing an electronic manuscript as opposed to Lamott’s manuscript produced through a typewriter. I was never a routine user of a printer to begin with but the separation from my office printer during the pandemic, combined with a diffuse dislike of the clutter of paper, inexorably led me towards working without print outs. It’s now been at least five years since I last printed something out and it wasn’t something I did much to begin with. The physical manifestations of my writing have slipped out of my immediate experience, no longer presenting as ready-to-hand, in a way that leaves them lodged as an intellectual possibility. In the same way that academics of my generation will often find it perplexing to be reminded that paper journals were once collected and consulted in physical form.

When our routines are disrupted we often feel compelled to account for that disruption. If things don’t work as planned, we are led to reflect on what we expected to happen. It’s easier to see routines when they don’t work because when they do they simply fade into the background. The same is true for the role of technology within these routines (Marres 2014: loc 1919). The introduction of LLMs into academic writing provides such a disruptive occasion because it unsettles many of the assumptions upon which our routines have previously depended. It’s no longer the case that a coherent piece of text we encounter must have been produced by a human author. It’s no longer the case that completing our own text requires only human effort.

This technological shift forces us to confront what writing means to us beyond its mechanical production. Just as word processors transformed academic writing by making revision less laborious, LLMs challenge us to articulate what remains essentially human in our scholarly production. Perhaps what matters most isn’t whether we occasionally use AI assistance, but how thoughtfully we integrate these tools into practices that preserve intellectual ownership and creative engagement with our ideas.

#academicWriting #books #creativeWriting #digitalScholarship #education #generativeAI #technology #writing #writingTips

Apr 16, 2025, 19:54 · · 1 · 0
isws

We are very happy that Heiko Paulheim from University of Mannheim, Germany, is joining the ISWS 2025 tutor team again. ...and as you can see by yourself, he seems to be proud, too, to be part of our summer school ;-)

2025.semanticwebschool.org/

#semweb #summerschool #isws2025 #knowledgegraph #AI #responsibleAI #reliableAI #generativeAI #llms #semanticweb #academiclife #PhDs @fizise @albertmeronyo @AxelPolleres @pgroth

Heiko, standing tall, on the balcony of Bertinoro castle during ISWS 2022. He is wearing shorts and a black t-shirt.
Apr 16, 2025, 13:50 · · · 2 · 0
Mx Verda

Also: can we have an AI Machine Learning LLM whatever thing trained ONLY on consentually offered materials, text and art?
The bare minimum effort of "you must send your materials to us" rather than trawling the internet to steal whatever isn't nailed down.
Sure, spam bots and nuisance or malicious submitters, but most won't be. Could always have a holding pen area of like 2 or 3 months before adding to training data.

Proactively request opt-outs as well, to make filtering a tiny bit easier?

#AI #OpenAI #ChatGPT #DeepSeek #MachineLearning #LLM #LargeLanguageModels #GenerativeAI #RegurgitativeAI

Apr 16, 2025, 12:16 · · 0 · 0
Dr. Anna Latour

Now starting at ICT.OPEN: a panel discussion on the Future of Education in the Era of Generative AI. As a freshly-minted assistant professor with increasingly many teaching responsibilities, I am keen!

ictopen.nl/programme/the-futur

#ICTOpen
#ICTOpen2025
#NWO
#Conference
#AcademicMastodon
#AcademicChatter
#ICT
#Research
#Networking
#Utrecht
#Jaarbeurs
#ComputerScience
#Education
#GenerativeAI
#AI

Five people sitting on a stage in front of a slide that says: Plenary session - The future of Education in the Era of Generative AI.
Apr 16, 2025, 11:06 · · 0 · 0
Christoph Becker

"AI models age like milk" is hilarious and quite precise. :)
The rest of this commentary is much less fun but worth reading. It's possible that business models can be tuned and efficiency gains improve margins, but the argument and the questions it raises remain relevant. By @tante
tante.cc/2025/04/15/these-are- #AI #generativeAI

Apr 15, 2025, 21:42 · · 0 · 0
Sage Sharp

Bad news: Google is rolling out generative AI tools for video creation (Google Vids). It's turned on by default for all Google Workspace users.

Good news: If you turn off Google Vids for your domain, your users won't be able to open Vids files shared with them. Hooray for avoiding deep fakes!

Go forth and disable the latest generative AI nonesense: support.google.com/a/answer/14

#enshittification
#generativeAI

Apr 15, 2025, 21:10 · · 10 · 0
MineEyesDazzle

This screenshot is just one of many examples showing why searchers should ignore the garbage AI Overview being force-fed at the top or the results list in Google search. Even better, avoid Google search.
It wrongly credits George Takei with having written "All Boys Aren't Blue: A Memoir-Manifesto" by George M. Johnson and overlooks George Takei's actual book: They Called Us Enemy!

#AI #generativeAI #garbage #SearchResults #books #Google #Scroogle

A screenshot of a Google search for the phrase "Texas author book banned" retrieving an AI Overview at top wrongly identifying George Takei as an author of All Boy's Aren't Blue written by George M. Johnson.
Apr 15, 2025, 17:40 · · 1 · 0
Simon D. ⏚

La CNIL irlandaise remet en cause la légalité de l’entrainement de Grok

next.ink/brief_article/la-cnil

La Data Protection Commission (DPC), équivalent de la CNIL en Irlande, a annoncé ce vendredi avoir lancé une enquête contre le réseau social X. Cette procédure vise le traitement des données à caractère personnel contenues dans les messages publiés par les utilisateurs de la plateforme pour entraîner des IA génératives, et « en particulier le modèles de langage Grok ».

#IA #AI #generativeAI #ia_generative #twitter #grok #cnil #dpc

La CNIL irlandaise remet en cause la légalité de l'entrainement de Grok - Next

La Data Protection Commission (DPC), équivalent de…

Next
Apr 15, 2025, 08:25 · · 1 · 0

Resources

Developers

What is Mastodon?

qoto.org

More…

v3.5.19-qoto · Privacy policy