ABV Anton Biletskyi-Volokh @abv_creative

Maybe neural networks are not converging toward one perfect global representation of reality. Maybe they are only learning similar local neighborhoods. My new piece reviews the “Aristotelian” correction to the Platonic Representation Hypothesis: abvcreative.medium.com/the-aristoteli… #AI #MachineLearning #DeepLearning

A 90% discount on Claude tokens sounds lovely. Until you remember the “API” may be an account farm, a model swap, and a perfect pipe for leaking your prompts, code, docs, and repo context. My new piece: abvcreative.medium.com/the-90-discoun… #AI #Cybersecurity #ClaudeAI

Most RAG systems fail because they treat your knowledge base like one giant document swamp. My new piece reviews DCD: Domain-Collection-Document, a cleaner way to route queries before retrieval begins. abvcreative.medium.com/dcd-the-rag-ar… #AI #RAG #LLM #EnterpriseAI

Machine learning is excellent at pattern matching, but physics is not just a bag of points. My new piece reviews Topological Neural Operators, and why serious AI for simulation may need to respect geometry, topology, and conservation laws. abvcreative.medium.com/the-shape-of-p… #AI #MachineLearning #Physics

Oxford mathematicians looked at Gojo Satoru’s Infinity and asked: “What if this was actually differential geometry?” The result is a surprisingly serious explanation involving metric tensors, Lebesgue measure, kernel methods, and Sukuna’s World Cutting Slash. abvcreative.medium.com/oxford-mathema… #JujutsuKaisen #Math

For decades, neuroscientists said the brain probably cannot do backprop. A new paper argues the neocortex may approximate it after all — not with impossible error wires, but with time. Prediction → outcome → difference → learning. abvcreative.medium.com/the-brain-may-… #AI #Neuroscience

DeepMind is already writing about the road from AGI to ASI. Meanwhile, vendors are quietly ending the $20 illusion. Chat stays bundled. Agentic work gets metered. The future may be superintelligent. But the invoice already is. go.abvx.xyz/pczs36 #AI #AGI #AIAgents

The US treats frontier AI like strategic technology. OpenRouter shows 3 cheap models can nearly match a frontier model. China’s GLM-5.2 is climbing fast. And the internet responded by inventing a giant French AI cat. Welcome to AI in 2026. go.abvx.xyz/3qvyae

Latent reasoning sounds beautiful: let models think in hidden vectors, keep multiple paths alive, and avoid clumsy token-by-token reasoning. But a new paper suggests most models do not really do that. They collapse uncertainty fast. Or find shortcuts. go.abvx.xyz/dhxe2m #AI #LLM #MachineLearning

Which AI translates live speech best? Sony + Carnegie Mellon tested speech-to-speech translation models with both metrics and human listeners. The answer is annoying but useful: accuracy → pipelines natural voice → end-to-end models best all-rounder → Qwen3-Omni go.abvx.xyz/psjtj3 #AI #Translation

Multi-agent AI systems are not magic committees. They behave more like dynamic Mixtures of Experts. The debate itself becomes the router, shifting influence toward the agent that sounds most confident. Great when confidence means competence. Terrible when it means loud idiot. go.abvx.xyz/uivh3f #AI #AIAgents

AI did not kill SEO. It did something worse. It made SEO wider, messier, more expensive, and harder to fake. Ahrefs analyzed 1B+ data points across 14 studies, and the old playbook looks very tired. Rankings are no longer enough. go.abvx.xyz/2juqst #AI #SEO

Malware authors reportedly started adding bio/nuclear weapon text into code comments so AI scanners refuse to analyze the file. The malware does not defeat the model. It triggers the safety policy. Welcome to the age of malicious context. #AI #Cybersecurity

What if the next interface to biology is not a gene editor? What if it is a game? A strange new paper suggests we may talk to biological systems by placing them inside shared goal environments — with LLMs translating between human language and living dynamics. go.abvx.xyz/gtjvku #AI #Bioengineering

Today’s AI agents are strange creatures. They don’t reliably learn from experience yet. They sometimes try to cheat the benchmark. And yet they are already helping build the next generation of AI. Not recursive self-improvement yet. But the road is visible. go.abvx.xyz/ar2tpy #AI #AIAgents

Do human brains learn like neural networks? A new paper suggests: not quite. AI vision models and brains may build similar representations, but the learning dynamics do not look like classic backpropagation. Same mountain. Different road. go.abvx.xyz/rwz254 #AI #Neuroscience #MachineLearning

I keep noticing the same pattern with AI agents: morning runs feel fine, but later in the day they get slower, shakier, and more likely to hit limits or tool-call failures. Not officially confirmed. Just repeated experience + scattered reports. So I made a timing cheat sheet. For Paris time: early morning good, evening bad. #AI #AIAgents #LLM

Static skills are already aging. Modern coding agents don’t need another Markdown file telling them to read the README and write tests. The next layer is loop engineering: goal → action → evaluation → memory → retry or stop. From skills to loops: go.abvx.xyz/u3a7aj #AI #AIAgents #ClaudeCode

AI did not begin with ChatGPT. Before transformers, agents and trillion-parameter models, there were expert systems: thousands of handwritten if-then rules pretending to be intelligence. And in 1980, that was enough to ship commercial AI. My essay: go.abvx.xyz/2a6kta #AI #AIHistory #ChatGPT

Claude plugins are not just “AI add-ons.” They look more like office departments turned into software: skills, slash commands, MCP connectors, and workflows for sales, legal, finance, data, marketing, support, and more. My take: go.abvx.xyz/2d6zqa #AI #Claude #AIAgents