Zhemal Khamidun
Head of AI at Alpina Digital, CPO of AlpinaGPT (~30-person engineering & AI team). Founder of Vibe Academy & Hamidun Consulting, author of the AI channel «Готовим ИИшницу» (@jemal_hamidun).
OpenAI unveils ChatGPT Work — an agent for multi-hour projects
OpenAI has unveiled ChatGPT Work, an agent capable of acting across apps and files, staying on a project for hours, and turning a goal into a finished result.
Deutsche Telekom and OpenAI transform telecommunications with AI
Deutsche Telekom is integrating OpenAI technologies to transform customer service, employee workflows, and network operations.
AegisDx: AI framework for safe diagnostics outperformed GPT-5 in physicians’ evaluation
Researchers introduced AegisDx, an AI system with mandatory step verification and screening for dangerous diagnoses, which outperformed GPT-5 in a blinded clinical…
Graph neural network recognizes hand gestures in 48 ms: 99% accuracy for prosthetics and AR
Qwen3 language models easily make educational tasks more complex, but struggle to simplify them
NYIT developed an LLM assistant for an applied analysis course in MATLAB
Infinity-Parser2 surpassed DeepSeek-OCR-2 in document parsing and opened a dataset of 5 million examples
Hallucination Self-Play: a small language model surpassed large LLMs in hallucination detection
The new HSP framework trains a hallucination detector and generator in a closed loop — a small model outperformed large LLMs without external data labeling.
Gemini 2.5 Flash replaces human evaluators in voice agent testing
Researchers tested how reliably Gemini can evaluate voice AI agents instead of humans — and found that the model performs at 100 times lower cost.
TACO: a method removes credit “contamination” in RL training of language models
New MiniLM-based method more accurately detects queries outside chatbots' knowledge scope
Debiasing NLP models creates new stereotypes for other demographic groups
Scientists propose "alignment plausibility" as a safety standard for medical LLMs
LLMs prove theorems but do not discover mathematics: researchers outline the next step
A position paper on arXiv systematizes progress in AI4Math and explains why current LLM systems are incapable of discovering new theorems — and proposes a roadmap.
New AI tool combines GTAP and APSIM to assess shocks in agricultural supply chains
Researchers have developed an AI system based on GTAP and APSIM that analyzes shocks in agricultural supply chains through natural-language queries.