
Ron Clabo
@dotnetcore
Founder GiftOasis LLC. Tweets about software development: .NET, AI, ASPNET Core, C# & LuceneNET. Top 1% on StackOverflow. I Love learning & helping others.
Our new Gemini 2.5 Computer Use model is now available in the Gemini API, setting a new standard on multiple benchmarks with lower latency. These are early days, but the model’s ability to interact with the web – like scrolling, filling forms + navigating dropdowns – is an…

Really great commentary on Microsoft's recent In-Context Learning (ICL) Paper. Good insights on how many shots are needed for "few-shot" prompting.
Cool paper from Microsoft. And it's on the very important topic of in-context learning. So what's new? Let's find out:

Different AI models have different strengths, weaknesses, biases, and safety profiles. Blending them to achieve an optimal user experience is nontrivial.
How important is Context Engineering to leveraging AI? Even humans can’t be intelligent without good context. Full stop. Read that again.
“What techniques are you using to customize your AI systems?" Amplify, a VC firm, surveyed 500 software/AI engineers. Here are the results:

Google DeepMind’s research found vector search hits a scalability limit due to fixed embedding dimensions, underperforming BM25 in most cases. Time to explore hybrid approaches.
Google DeepMind Finds a Fundamental Bug in RAG: Embedding Limits Break Retrieval at Scale Google DeepMind's latest research uncovers a fundamental limitation in Retrieval-Augmented Generation (RAG): embedding-based retrieval cannot scale indefinitely due to fixed vector…

In the blink of a cosmic eye, we passed the Turing test. ... And yet the moment passed with little fanfare, or even recognition. - Mustafa Suleyman
"Learning is not supposed to be fun ... the primary feeling should be that of effort." 🎯 Andrej Karpathy on how to learn. I did a ChatGPT DeepResearch (link below), for studies justifying this. Basically it found👇 If learning feels easy, results are usually shallow. If…

🎯 Andrej Karpathy on how to learn.

OpenAI’s two new OSS models are gaining GitHub likes at a rate that would’ve been called impossible six months ago.
GPT-OSS: - 5M downloads in <1 week on @huggingface 🚀 - 400 new models - already outpacing DeepSeek R1’s launch numbers, and that’s without counting inference calls - also the most-liked release of any major LLM this summer

The growth of GitHub Copilot is just phenomenal. This slide comes from an excellent recent presentation by @mkristensen

Still skeptical about AI and its usefulness? Think about this: The ceiling of a model's capabilities is usually the operator themself.
United States الاتجاهات
- 1. Happy Birthday Charlie 55K posts
- 2. #tuesdayvibe 3,555 posts
- 3. Good Tuesday 32.3K posts
- 4. Shilo 1,669 posts
- 5. Pentagon 76K posts
- 6. Victory Tuesday N/A
- 7. Happy 32nd 9,587 posts
- 8. Dissidia 6,757 posts
- 9. #NationalDessertDay N/A
- 10. #PutThatInYourPipe N/A
- 11. Standard Time 3,042 posts
- 12. #14Oct 2,051 posts
- 13. #TacoTuesday N/A
- 14. Happy Birthday in Heaven 1,581 posts
- 15. Happy Heavenly 8,411 posts
- 16. Martin Sheen 6,587 posts
- 17. Janet Mills 1,671 posts
- 18. Larry Fink 5,721 posts
- 19. No American 64.8K posts
- 20. Time Magazine 18.7K posts
Something went wrong.
Something went wrong.