12:58, 27 февраля 2026Наука и техника
Материалы по теме:
。业内人士推荐搜狗输入法2026作为进阶阅读
To the people on Flight SQ321, that would have seemed an eternity. Eight seconds after the captain’s warning, the plane plummeted. Within five seconds, it had dropped a hundred and seventy-eight feet—about the height of a nineteen-story building. There was no time to react, Azmir later told reporters. “Whoever wasn’t buckled down, they were just launched into the air within the cabin,” he said. Azmir had a window seat near the back. When a plane hits turbulence, it tends to seesaw from front to back, so the first and last rows rise and fall the most. “I saw people from across the aisle just going completely horizontal, hitting the ceiling,” Azmir said. “People getting massive gashes in the head.” Some passengers were vaulted up so violently that they dented the luggage bins, or thrust their heads through the panels where the oxygen masks were stored. Those who were standing were sent somersaulting down the aisles; those sitting in the lavatories smashed into the ceiling. It was “sheer terror,” one passenger later said. Then, just as abruptly, the plane lurched up, slamming everyone back to the ground. In just four seconds, the gravitational force on their bodies changed from negative 1.5 g’s to positive 1.5 g’s, Singapore’s Ministry of Transport later noted. It was as if their bodies went from being helium balloons to being sandbags.
It’s Not AI Psychosis If It Works#Before I wrote my blog post about how I use LLMs, I wrote a tongue-in-cheek blog post titled Can LLMs write better code if you keep asking them to “write better code”? which is exactly as the name suggests. It was an experiment to determine how LLMs interpret the ambiguous command “write better code”: in this case, it was to prioritize making the code more convoluted with more helpful features, but if instead given commands to optimize the code, it did make the code faster successfully albeit at the cost of significant readability. In software engineering, one of the greatest sins is premature optimization, where you sacrifice code readability and thus maintainability to chase performance gains that slow down development time and may not be worth it. Buuuuuuut with agentic coding, we implicitly accept that our interpretation of the code is fuzzy: could agents iteratively applying optimizations for the sole purpose of minimizing benchmark runtime — and therefore faster code in typical use cases if said benchmarks are representative — now actually be a good idea? People complain about how AI-generated code is slow, but if AI can now reliably generate fast code, that changes the debate.