This reminds me of the opening quiz of Hans Rosling’s Factfulness. It’s good to be confronted by your own subjective biases. A poor understanding of the world as it really is can lead to sub-optimal decisions.

Here's the spiciest detail from the new o1 system card:

If you have a Macbook Pro M series with at least 64 GB's of RAM you can now run a GPT-4 level LLM locally! 1. Install ollama 2. Open your terminal and run ollama pull llama3.3 3. Then ollama run llama3.3 "your prompt" Your own personal AI is here!

This is wild. OpenAI just dropped Sora, their state-of-the-art AI video model. The video quality looks insane. 100% AI, these are not real. 1. 80s Japan Harajuku fashion street

This is going to be huge! I'm still amazed that I can run a GPT-4 quality LLM locally on my MacBook (ollama pull llama3.3) by I can already tell that 2025 is going to be a *wild* year in AI. 🔥

“DeepSeek outperforms other models, including Meta’s Llama 3.1 405B, OpenAI’s GPT-4o, and Alibaba’s Qwen 2.5 72B” - TechCrunch DeepSeek V3 has 671B params, has an open-weights release with permissive commercial licensing, and was developed by a Chinese firm trained with $6MM of

How cool is this?! The Government's Incubator for AI have released an open-source list of publicly available government datasets 🤓

Meet the OpenStreetMap Category Viewer! I made this to help people find and download data from OpenStreetMap ... it's free to use

"OpenAI may charge as much as $2,000 a month for full access to o3" o3 achieved an extremely impressive score of 91.5% on François Chollet's Abstraction & Reasoning Corpus, but at a cost of $3000 inference fees *per query*. economist.com/business/2025/…

If you're curious about current expert assessments of AI risk, this report is an excellent reference. It offers thorough content that is not overly editorialized.

Run the full Deepseek-R1 LLM locally for only $6000 hardware costs (post by Hugging Face engineer)

Are you an AI-assisted "vibe coder", or an artisanal coder using traditional tools and methods?

Pros: it’s on GitHub you can try it today github.com/microsoft/Omni… Cons: requires a Windows 11 VM to run

This is a rather convincing X scam: it’s possible for bad actors to spoof any application they want and have users authorise their X account, giving away control of your profile & timeline. Stay safe out there folks!

Six more until we hit #100!

What is MCP & why it's a big (huge) deal: (model context protocol) TLDR: MCP makes it possible for AI Tools to use external tools. E.g. Chatbot/IDE/AI-Agent can use Gmail/GoogleDrive/WeatherApp etc. Detailed explanation for both, tech & non tech people (+demos): 1) AI Tools

"Good forecasting isn't about predicting the car, it's about predicting the traffic jam." - official So given Llama are about to get 10x faster and 10x cheaper, what does this enable?

I will be talking about evaluating AI systems in my talk tomorrow for PyData London , "Death by RMSE: A Cautionary Tale of Metrics Gone Wild" "You’ve trained the model. The eval metrics look great. But somehow, it doesn't change anything — the KPIs are static, the business

it's happening 🥳🎉 I'm having a lot of fun writing this if you know of any papers or blog posts I should reference, please comment with the links!

I love that people are vibe coding and building stuff for the first time later in life, but how do we help them with security?