A quick review of the AI narative at Davos 2026.

💬 Topics:
- AI hype versus reality;
- Coding agents limitations;
- AGI is not possible;
- Transformer architectural has limitations;
- AI economics don't work.

🥇 Join the community - https://www.youtube.com/channel/UCXzw-OdotBUcNA9yhuYQBwA/join

Awesome Coding's resource offer curated lists of resources, tools, and libraries for software developers. Readers can learn about programming languages, frameworks, and development tools across various domains. With carefully selected recommendations and community contributions, Awesome Coding helps developers discover new tools and resources to enhance their productivity and skills.

Awesome

AI predictions from tech CEOs at Davos claim developers will be replaced within 6-12 months, repeating similar claims from a year ago. Current transformer-based AI architectures have fundamental limitations—they rearrange existing knowledge probabilistically but cannot generate truly novel ideas. While AI will likely handle more boilerplate code generation, the technology has plateaued and requires unknown breakthroughs before achieving major industry disruption. Meanwhile, companies like OpenAI struggle with profitability, now introducing ads despite previously calling it a last resort.

AI will replace developers in 6 months. Again...

Promises. Promises. Hurry up, I want to retire…

Same energy as nuclear fusion’s clean energy that’s been “10 years away” for half a century

True. This is a mathematical limitation of the LLMs: the transformers are very expensive stochastic parrots, which only “rearrange existing knowledge probabilistically”, and RL doesn’t help here.
I personally don’t believe in AGI, because there is a lot of “unwritten” human knowledge which is the part of our everyday life, but LLM just can’t be aware of.

At a certain point, AI can’t do any more. That’s when we need to use our knowledge.

Software Engineers have been under constant threat of being replaced in the next 6-12 months for the past 3 years.
Software Engineering isn’t about coding, which AI can do, but about solving problems and making connections, something AI can’t do.