Car sales in Indonesia rose 28.1% year-on-year in September.

IMF Asia-Pacific director said by 2050 countries including South Korea and Japan could lose about a quarter of their labor force to population aging.

According to renowned X blogger Andrej Karpathy, roughly 6 billion people (~75% of the population) have almost never encountered LLM. About 1-2 billion people (~20%) are occasional users of free ChatGPT-like products. This group sees it as a clunky chatbot, using it as a better Google search or writing assistant, perhaps asking the agent to book a flight. Many of my non-technical friends (which seems reasonable to me) say it's useless to them. Many professionals outside of math and coding are also in this category. For example, executives and many other roles spend a lot of time dealing with people, so while they intellectually understand what's happening, it's ultimately secondhand and somewhat abstract. Further down the hierarchy are those who professionally use cutting-edge LLM in the fields of math and coding. Approximately 20 million people (0.2%) have witnessed firsthand how large, complex projects that used to take weeks or months can now be completed with a single prompt from an agent. Developing apps, copying apps, translating apps, decompiling apps from binary code… all of this happened so quickly and recently—less than a year ago, I was still manually writing code, typing memorized commands character by character into the editor, occasionally using Tab for auto-completion. Finally, there are the approximately 5,000 people (~0.00006%) within the industry who have access to cutting-edge systems. The outside world only sees the trailer: thousands of agents forming swarms, collaborating for weeks to complete massive software projects—massively mining zero-day vulnerabilities, conducting network attacks and defenses at machine speed, discovering new science, and advancing the frontiers of mathematics. This is work that even top experts in the field would take years to complete. Meanwhile, human verification and understanding have begun to lag behind. For example, the OpenAI-HF incident a few months ago is still being "archaeologically investigated." Mathematicians will likely have to pore over those 722 cutting-edge mathematical manuscripts for quite some time.