Separate model for review is the move. Same blind spots on generation and verification is exactly why I keep pushing for external eval pipelines.
GENERATIVE AI
-
RAAIS Summit Announces Jeffrey Hawke as 10th Speaker
By
–
the time is here – we're launching our 10th @raais summit speakers! first up, @jeffrey_hawke
, co-founder/cto at world simulator pioneers @odysseyml jeff's work intersects generative modelling, embodied intelligence, and large-scale learning he was prev vp tech @wayve_ai -
AI Inference Era: The Next Architectural Shift Beyond Training
By
–
The AI training era is reaching its limit. While the world is obsessed with $100B "Gigafactories," we’re looking at what happens next. At MWC Barcelona, Axelera AI CEO Fabrizio Del Maffeo sat down with theCUBE to discuss the Inference Era. The architectural reset of the
-
LLM Providers in Growth Stage, Not Yet in Enshittification Phase
By
–
It's wild to me that everyone's so paranoid about LLM providers cutting corners or chiselling revenue. Friends, we're in the "desperate race for growth" stage. The enshittification doesn't start until much later. (This isn't really a bad example of what I'm complaining about, but the people with wild conspiracies about how Claude Code is engineered to churn through tokens are always small accounts I don't want to dunk on. Anyway the point is that the products flip settings on or off or behave in unideal ways because they're being built not extremely well extremely extremely quickly. It's not safe to assume these companies have your best interests at heart, but it is very safe to assume they want you to have a good time — for now.) Miles Brundage (@Miles_Brundage) Lately, Claude has been defaulting to Sonnet in a way that I don't think it ever did before. PLEASE STOP THIS, IT'S REALLY ANNOYING — https://nitter.net/Miles_Brundage/status/2031110468014924232#m
-
OpenAI Launches GPT-5.4 with Native Computer Usage Capabilities
By
–
OpenAI launched GPT-5.4 on March 5 — first model with native computer-use built in. One system that can reason, code, and operate your desktop. That's a meaningful consolidation. openai.com/index/introducing-gpt-5-4/ #AI [Translated from EN to English]
→ View original post on X — @svenphilipsen, 2026-03-10 08:00 UTC
-
Reality of Agentic AI: Beyond Labels and Simple Workflows
By
–
𝗧𝗵𝗲 𝗿𝗲𝗮𝗹𝗶𝘁𝘆 𝗼𝗳 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜 𝗿𝗶𝗴𝗵𝘁 𝗻𝗼𝘄. → Many people are talking about AI agents. Only a few are actually building systems that generate real value. → Most “AI agents” today are still assistants or simple workflows with a new label. That gap is
-
LLMs’ jagged frontier: memorization versus inference capabilities
By
–
LLMs' jagged frontier is not so jagged when you look at it in the right space. LLMs have memorized vastly more sentences than any human, but their ability to correctly combine them into new inferences is still limited. The rest follows.
-
Leveraging AI-driven feedback loops for project development
By
–
I've been showing friends my new AI on X project, coming soon. They all love it. But then they send me a page of changes they would make. I copy that into my AI. Ask it its opinion. Give mine. Then it goes to work. Getting this feedback loop going is very powerful. My
-
Microsoft Chooses Anthropic for Copilot Cowork in M365
By
–
Microsoft just announced Copilot Cowork — built with Anthropic's Claude and shipping in M365. OpenAI signed the Pentagon contract. Anthropic refused it. Now Microsoft is betting on Anthropic for enterprise agents. Corporate AI choices are geopolitical now. #AIGovernance [Translated from EN to English]
→ View original post on X — @svenphilipsen, 2026-03-10 06:00 UTC