Update on my travails with using Codex to speed up and optimize an ML algorithm from scratch. After two days of where things seemed like they were *finally* going in the right direction, at the end of the day I discovered that almost all the speedups were only relevant for one
MACHINE LEARNING
-
Elon Musk announces V8 small foundation model with 500B params
By
–
This is just our V8 small foundation model, so 500B params
-
Grok 4.2 an order of magnitude smarter and faster after beta
By
–
Grok 4.2 will be about an order of magnitude smarter and faster than Grok 4 when the public beta concludes next month. Still many bug fixes and improvements landing every day. The public beta gives us more critical feedback to address.
-
Google Genie 3 team says world models are path to AGI
By
–
Well, the Genie 3 team at Google believes that world models are the only way we will really get to AGI. And now that reinforcement learning has been added to LLMs (along with other technologies) you probably are right.
-
PolyAI trained on millions of messy real calls
By
–
Real calls are messy. Crying babies. Street noise. Heavy accents. PolyAI doesn't care. It’s trained on millions of real service calls. It stays locked in when others fail. pic.twitter.com/vmZ3UVueAt
— Linus ✦ Ekenstam (@LinusEkenstam) 17 février 2026Real calls are messy. Crying babies. Street noise. Heavy accents. PolyAI doesn't care. It’s trained on millions of real service calls. It stays locked in when others fail.
-
Foundations of 4.2 Enable Weekly Improvement and Strong Recursive Intelligence Growth
By
–
The foundations of 4.2 are such that it is able to improve every week, so the recursive intelligence growth will be strong
-
AI as a Stack: From Rules to Agentic Intelligence
By
–
AI isn’t one thing — it’s a stack. Rules → ML → Neural Nets → Deep Learning → GenAI → Agentic AI Agents don’t replace the layers below.
They orchestrate them. If GenAI answers,
Agentic AI executes. -
LLMs alone won’t lead to Artificial General Intelligence
By
–
For our free newsletter this week, we talk about why LLMs won’t led to Artificial General Intelligence (AGI).
@IrenaCronin and I write this newsletter every week.
LLMs alone are unlikely to lead to AGI. They primarily predict text and can imitate planning and reasoning -
Experiential Reinforcement Learning: Teaching LLMs Self-Reflection
By
–
Making LLMs truly learn from its experience "Experiential Reinforcement Learning (ERL)" ERL makes an agent attempt -> get sparse feedback -> write a self-reflection -> retry All by distilling the improved retry back into the base policy so the correction sticks without needing
-
Machine Learning and KPI Design: Tracking the Right Signals
By
–
How machine learning reframes KPI design: systems optimize relentlessly. So the real leadership question isn’t “Are we tracking performance?” It’s “Are we tracking the right signals?” https://
sloanreview.mit.edu/article/what-a
i-can-teach-us-about-designing-better-kpis/
… via @mitsmr #AI #MWC26