COMPUTING

–

28 September 2025 23h46

4. With SCIGEN, AI proposed millions of candidates and directly enabled the synthesis of two exotic materials with Kagome and Archimedean lattices. This could open doors for quantum computers, superconductors, and other transformative technologies.

AI Models Generate Safe But Unremarkable Materials for Scientific Research

By

–

28 September 2025 23h46

2. Until now, AI models tended to generate safe, stable, but unremarkable materials. They rarely produced the exotic structures needed for breakthroughs in quantum physics, superconductivity, or advanced computing.

GPT-5 Breakthrough in Quantum Computing Error Probability Problem

By

–

28 September 2025 22h32

GPT-5 helped solve an old problem in quantum computing. Here's what it was about: The researchers investigated a central problem in quantum computing: In the complexity class QMA, the quantum version of NP, the question is whether the probability of errors in proofs can be

→ View original post on X — @antgrasso

Unplanned Cloud Migration Costs Demand Structured Preparation

By

@antgrasso

–

28 September 2025 20h30

Unplanned costs in cloud migration arise when choices on teams, timing, priorities, or technical foundations are misaligned, reminding that efficiency in such journeys depends on structured preparation and the ability to anticipate complexity Microblog @antgrasso #CloudComputing

OpenAI’s AI Arms Race Demands 17 Gigawatts Infrastructure

By

–

28 September 2025 19h13

https://
cnbc.com/2025/09/26/ope
nai-big-week-ai-arms-race.html#:~:text=Buildingout17gigawattsof,arenofirmanswersyet
….

Power Grid Challenges in an AI-Driven Energy Future

By

–

28 September 2025 19h13

The question that arises is not so much how the electricity should be produced (nuclear energy, fusion energy, even if this is a Herculean task), but rather how the power grids should cope with this.
AI will continue to drive GDP growth over the next 10 years, leading to

→ View original post on X — @saboo_shubham_

oLLM: Lightweight Library for Local LLM Inference

By

@saboo_shubham_

–

28 September 2025 19h11

oLLM is a lightweight Python library for local large-context LLM inference. Run gpt-oss-20B, Qwen3-next-80B, Llama-3.1-8B on ~$200 consumer GPU with just 8GB VRAM. And this is without any quantization – only fp16/bf16 precision. 100% Opensource.

→ View original post on X — @skathirmani

Teaching Logo Programming with Python to Improve Kids’ Skills

By

@skathirmani

–

28 September 2025 18h34

Started teaching logo programming (using #python) to kids. Aim is to improve their maths, logical and programming skills. Was teaching coordinates today.