Cognitive warmup. Microsoft seems to be living in a bubble, aren’t they? There seems little in terms of any realisation, or even attempts at introspection, concerning how badly they’ve dropped the ball on AI. From claiming as much as 30% of code in Microsoft is now written by AI, to having a series of botched Windows 11 updates in subsequent months, which broke critical functionalities on millions of PCs. The aspirations of making Windows 11 in an “agentic OS” received backlash, as soon as those dreams were published on X—something I classified as chaos.
The never-ending saga now adds Microsoft’s Copilot and LG’s webOS TVs.
The TV maker recently rolled out Copilot to users’ TVs, in a way that it was impossible to disable or uninstall the AI. First, LG quietly installed it hoping no one would notice. Secondly, why not give users a choice? After some backlash, LG says it’ll now be possible to delete the Copilot shortcut from your Smart TV home screen, and the next webOS update will allow Copilot to be uninstalled completely. I’m afraid AI debris will still be discovered long after the hype has subsided.
PAST PERSPECTIVE
2025 wasn’t the year AI became sentient, autonomous, or unstoppable. It was the year the excitable, and loud AI industry discovered where the real bottlenecks lie. Trust, relevance, supervision, economics, integration, geopolitics, and electricity. Models supposedly improved. Claims of intelligence and smartness grew louder. Yet, reality quietly caught up, and the supposedly smartest beings on the planet were left floundering more often than not.
OpenAI’s GPT-5 and an illusion of “thinking by default”
OpenAI, amidst much fanfare for its late summer release, positioned GPT-5 as the model where reasoning became intrinsic. Smarter than anything before it, and the PhD-level intelligence claims were being dropped regularly in the…