Habr AI
AI news source. Articles are auto-selected and adapted by Hamidun News editors.
Latest publications
From an Idea to 800 Million Users: The Complete Timeline of OpenAI’s Rise
We trace OpenAI’s path from its founding in 2015 to ChatGPT’s rapid growth. How a technological breakthrough changed the industry and became part of our everyday lives.
Transformer Architecture Without Library Magic: A Step-by-Step Implementation in NumPy
A fundamental breakdown of modern AI architecture. Learn how to write and train a mini-transformer from scratch without using ready-made frameworks like PyTorch.
From a simple chat to a powerful tool: 28 days with Spring AI
An AI Advent Challenge participant explained how tight deadlines helped them master the Spring AI framework and build a working product instead of endlessly studying documentation and reworking code.
The End of Romanticism in AI: Mass Layoffs and Ads in ChatGPT
February 2026 marked a point of no return for the generative AI industry. The departure of key developers and the introduction of ads in ChatGPT signal that research startups are becoming hard-nosed Big Tech.
Foreign vs. Russian: how to choose an LLM for real-world business
Choosing a language model for production is not just about benchmark numbers. We look at what matters more: generation quality or infrastructure stability.
OpenAI strengthens its position in the agent race: OpenClaw creator joins the company
OpenClaw founder’s move to OpenAI signals a shift in the industry’s priorities. The focus is now not just on smart models, but on reliable agents capable of doing real work.
The Art of AI Coding: 6 Strategies for Effective Development in 2025
The move to AI-assisted development requires a paradigm shift: from manual coding to delegation and automated checks. Learn how to lay the groundwork for agents in 2025.
Claude 4.6 Opus vs. GPT-5.3: Anthropic changes the rules of the game
For the first time, OpenAI and Anthropic have released flagship models simultaneously. Claude 4.6 Opus bets on flawless logic and code, challenging GPT-5.3's multimodality.
15 prompts for QA: how ChatGPT automates a tester’s routine
Using ChatGPT in a QA engineer’s work makes it possible to automate requirements analysis and bug report writing. Learn about 15 scenarios that speed up processes and free up time.
AI in Investing: Real Opportunities Versus Market Hype
We examine how machine learning and LLMs are transforming capital management and where the boundaries lie between marketing promises and the real effectiveness of algorithms.
The myth of AGI: why universal superintelligence remains an unattainable dream
As tech giants promise the imminent arrival of superintelligence, experts point to fundamental barriers on the path to AGI. We examine why the concept of universal artificial intelligence may be a utopian idea.
DGX Spark with Qwen3: the NVIDIA test that didn't tell the whole story
We analyze the capabilities of NVIDIA's new LLM solution: when 128 GB of memory is too little.
Russian AI models come to OpenClaw: integrating GigaChat and YandexGPT
Expanding the capabilities of the open-source platform OpenClaw: a step-by-step guide to integrating GigaChat and YandexGPT for building effective Russian-language AI agents.
The AI Ecosystem Through an Analyst's Eyes: From Simple Utilities to Powerful Solutions
A senior systems analyst's experience using neural networks for work and pet projects. An honest review of tools that genuinely help with analysis and programming.
50 neural networks built a "digital organism" and invented "cancer"
An experiment with 50 tiny LLMs showed how a complex structure and self-organization can emerge from autonomous agents.
Advertising in ChatGPT and Gemini’s breakthrough: the week’s top developments in AI
OpenAI is bringing advertising to ChatGPT, Google is expanding Gemini’s capabilities, and Z.AI has released an ultralight model for coding. We break down the key trends and releases of the third week of January 2026.
Microsoft and the AI trust crisis: why its earnings report caused market panic
Microsoft's earnings report triggered a record sell-off in its shares. Investors are beginning to doubt the return on billions invested in AI and OpenAI.
AI giants compared: who won the real-world stress test?
Forget the benchmarks. We tested ChatGPT 5.2, Gemini 3 Pro and Claude Opus 4.6 across five rounds of tough trials, from complex logic to building games in a single file.
Nine Robotics Trends: How the Industry Is Moving from Hype to Real Deployments
An in-depth analysis of the robotics market identified the key development vectors: why demo videos no longer work and how AI is already changing physical automation today.
AI security: why prompt injections are only a symptom of deeper problems
Companies are deploying AI agents and copilots but overlooking real threats. We explain why protection against jailbreaks will not save your corporate network from being hacked.
Yttri: An all-in-one productivity tool with local AI
A developer has introduced Yttri, a powerful Rust-based tool for project and knowledge management. No cloud, just local AI and full privacy.
Quantum breakthrough: will technology symbiosis become the key to AGI?
We explore the murky prospects of combining quantum computing and neural networks. Why experts call this union the “ghost” of strong AI and what to expect from the technologies of the future.
GEO vs. SEO: myths and reality of optimization for AI search
We examine the phenomenon of Generative Engine Optimization. Will content optimization for neural networks replace traditional search or remain only a temporary addition to marketing?
From Boring Lists to the “Wow Effect”: How to Automate Creative Brainstorming
Standard LLMs often produce banal ideas for naming and marketing. We explain how to overcome the limitations of ChatGPT and Gemini to get a truly original result.
Claude 4.6 vs. GPT-5.3: Global model showdown and Olympic scandal
Anthropic and OpenAI are locked in a head-on clash as Chinese AI models capture the video generation market. The week's main takeaways in our roundup.
Hands-on: how to build the Xiaozhi robot from readily available components
We break down the process of building the Xiaozhi robot yourself, from ordering parts to final assembly. Find out why this project is becoming a new favorite among DIY robotics enthusiasts.
The Age of Digital Illusion: How Neural Networks Erased the Boundaries of Reality in Three Years
From Midjourney’s early experiments to Kling’s hyperrealistic videos, neural networks have made a quantum leap. We examine why the boundaries between real and synthetic content have been erased for good.
The illusion of simplicity: why the hype around AI resembles infobusiness
The current trend around neural networks increasingly resembles the era of “performative success.” Instead of real technology, we are being sold a sense of ease, while errors and the high cost of trial and error are hidd
AI speeds up the search for materials for safe nuclear waste storage
Researchers from Skoltech and Sber AI used graph neural networks to model the properties of technetium carbides. The new method speeds up the search for materials for nuclear waste storage several-fold.
OpenAI Losses: $15 Billion and Gloomy Prospects
A recent article addressing OpenAI's scaling challenges contained an unfortunate error in financial calculations.