1. The DOGE website is garbage. The claimed savings has no relationship to the data provided. And the data that is provided grossly inflates the savings from each canceled contract. So we just launched our own site — the Musk Watch DOGE Tracker — that breaks everything down

Jon Stewart invited Elon Musk onto the Daily Show. Musk said he'd only do it if the show would appear unedited. Stewart and the Daily Show agreed -- calling Musk's bluff. Musk has now chickened out and refuses to go on the show.

Full system prompt omitted by Grok This is from 07:40am PST today Sunday 23, 2025 So much for free speech, no censorship etc. Link to full chat below

RT @Hareiamonline: @FlyingDutchPall @ludy1970 You can't make this shit up 😂

He got the community note removed

I used to live in Sweden. This kind of thing still amazes me.

The world's richest man has joined a growing chorus of right-wing voices attacking Wikipedia as part of an intensifying campaign against free and open access information. Why do they hate it so much? citationneeded.news/elon-mus…

Modal keeps impressing. Their GPU glossary documentation is a joy to read modal.com/gpu-glossary/readm…

🧵 A thread addressed to my fellow conservatives who are planning to vote, or considering voting, for Trump. 1/13

Medium has really convinced me to never publish content on a managed content platform. It went from being fun & open to paywalled & junky really quick. Any time I see a Medium link on a good topic my heart sinks.

God Who Took Form Of Swan Finding It Much Harder To Seduce Women Than Expected bit.ly/3yPpiva

What is everyone using these days to run DL/LLM experiments? Stuff like provision GPUs, farm out jobs that span hyper parameters, dataset splits, etc. I have used @DeterminedAI in the past. It required a bit of hacking so wanted to see what else is out there.

# On the "hallucination problem" I always struggle a bit with I'm asked about the "hallucination problem" in LLMs. Because, in some sense, hallucination is all LLMs do. They are dream machines. We direct their dreams with prompts. The prompts start the dream, and based on the LLM's hazy recollection of its training documents, most of the time the result goes someplace useful. It's only when the dreams go into deemed factually incorrect territory that we label it a "hallucination". It looks like a bug, but it's just the LLM doing what it always does. At the other end of the extreme consider a search engine. It takes the prompt and just returns one of the most similar "training documents" it has in its database, verbatim. You could say that this search engine has a "creativity problem" - it will never respond with something new. An LLM is 100% dreaming and has the hallucination problem. A search engine is 0% dreaming and has the creativity problem. All that said, I realize that what people *actually* mean is they don't want an LLM Assistant (a product like ChatGPT etc.) to hallucinate. An LLM Assistant is a lot more complex system than just the LLM itself, even if one is at the heart of it. There are many ways to mitigate hallcuinations in these systems - using Retrieval Augmented Generation (RAG) to more strongly anchor the dreams in real data through in-context learning is maybe the most common one. Disagreements between multiple samples, reflection, verification chains. Decoding uncertainty from activations. Tool use. All an active and very interesting areas of research. TLDR I know I'm being super pedantic but the LLM has no "hallucination problem". Hallucination is not a bug, it is LLM's greatest feature. The LLM Assistant has a hallucination problem, and we should fix it. </rant> Okay I feel much better now :)

finally got around to making my sci-fi book list: bbabenko.github.io/sci-fi/ (h/t @karpathy & @Noahpinion for the inspiration)

H&M:s vd Helena Helmersson vägrar ställa upp på en intervju med Aftonbladet. I stället går hon till DI där hon i luddiga ordalag försöker spela ner granskningen inför en rimligtvis inte lika påläst reporter som @staffanlindberg . Två ord: Fegt. Och uselt. di.se/nyheter/h-m-s-vd-klade…

The new Generative Fill feature in Photoshop is a game changer

First decide what problem you need to solve with AI, and then pick the right tool. As impressive as LLMs are, they are not the right tool for every job.