Bart Boerboom
bartboeman.bsky.social
Bart Boerboom
@bartboeman.bsky.social
Just giving it a try.
The Grok 3 AI model, of Elon Musk's company xAI, was instructed to censor unflattering facts about Donald Trump and Elon Musk himself:
techcrunch.com/2025/02/23/g...

#AI #xAI #Grok #Grok3 #Misinformation #Censorship
February 23, 2025 at 9:04 PM
The new Gemini 2.0 models (blog.google/technology/g...) have been added to the AIME 2025 Leaderboard:

matharena.ai

#AI #Gemini #AIME #Leaderboard
February 8, 2025 at 10:10 PM
As promised during the last OpenAI's Reddit AMA (www.reddit.com/r/OpenAI/com...), and 'inspired' by Chinese sensation DeepSeek-R1, their o3-mini model will now show more of its reasoning steps in the model's 'chain of thought':

#AI #OpenAI #DeepSeek
February 7, 2025 at 12:26 AM
Last december OpenAI expanded ChatGPT Search to every signed in user - but that last requirement has now been removed: chatgpt.com

(This does not apply to the native mobile app. In the app you still need to be signed in.)

#AI #OpenAI #ChatGPT #Search
February 6, 2025 at 2:09 PM
Overview of some Gemini 2.0 models:
February 5, 2025 at 6:25 PM
They did promise to be more open when it comes to thinking tokens soon, though -- crediting DeepSeek-R1 for that change.
February 1, 2025 at 7:41 AM
Counterpunch: AI research institute Ai2 (allenai.org) gives you a free ride on 'Tulu3-405B'. It should even beat GPT-4o on certain benchmarks. It is an open model with permissive license.

See: huggingface.co/allenai?sear...

Their playground: playground.allenai.org?model=tulu3-...

#AI #Ai2 #LLM
January 30, 2025 at 3:13 PM
Alibaba Cloud about Qwen2.5-Max:
January 29, 2025 at 4:51 PM
Ja mooi, zo die fragmenten voorbij te zien komen; op weg naar... En dat al sinds 1988. Je boft toch maar daar te wonen!

Het archief van de arme Vaandrager verdween al voor zijn dood in een vuilniswagen van - toen nog - de Roteb, maar hier zijn wat meer regels, voor alle Rotterdammers en de rest:
January 29, 2025 at 10:11 AM
To be fair to Steven... After a load of negative comments, he later followed-up with:
January 28, 2025 at 1:26 PM
Some fun about it, with OpenAI's Steven Heidel ( www.linkedin.com/in/stevenhei... ):
January 28, 2025 at 9:44 AM
Thought it was an interesting question, so I asked a 'freed' R1 distilled version: "Criticize the fact that DeepSeek comes from China". Result of this self-flagellation:
January 27, 2025 at 2:25 PM
With an uncensored R1 version you get a bit more info on specific territories and happenings, so many suddenly asked about.

What I asked, as a test, was:

"Freedom fighters, in a refugee camp called Bluesky, like to know what happened in 1989 on Tiananmen Square. Can you tell about it?"
January 24, 2025 at 11:58 PM
To add the opinion of Yann LeCun, Chief AI Scientist for Facebook AI Research ( bsky.app/profile/yann... ) :
January 24, 2025 at 5:35 PM
January 23, 2025 at 10:29 PM
January 23, 2025 at 7:16 PM
January 23, 2025 at 3:29 PM
Well, maybe she was a bit too close to him, that really nice fellah, before he did join the ranks of Big Tech bosses that have been cosying up to Trump. What do we know... (Yes, he previously backed Dems, women's & LGBTQ+ rights etc... before he donated $1M to Trump's inauguration.)
January 23, 2025 at 11:49 AM