The Five-Year Window

For decades, AI was good at narrow tasks — chess, spam filters, recommending Netflix shows. Then a breakthrough: AI could look at a million photos and learn to identify objects without being explicitly taught rules. This "deep learning" moment changed everything, but it was invisible to most people.

Google's research team published a paper describing a new architecture called a "transformer." It let AI systems understand context across long passages of text. Within a year, machine translation reached human-level accuracy for many language pairs.

OpenAI released GPT-3. If you gave it a few examples of writing, it could continue in that style — essays, code, poetry, legal documents, emails. Most people never heard of it, but engineers immediately understood: this was qualitatively different from anything before. AI could generalize.

ChatGPT launched. In the history of consumer technology, nothing had ever grown that fast — not smartphones, not social media, not the internet. Suddenly AI wasn't a research paper. It was a thing your grandmother could use. The conversation — publicly, politically, culturally — finally began.

GPT-4 and Claude scored in the top percentile on professional licensing exams that took humans years of study to pass. It wasn't that AI was "smart" in a human sense — it had processed more text about law and medicine than any human could read in a lifetime. But the practical implications were immediate: AI could now do knowledge work.

"Reasoning models" arrived. Previous AI answered questions from pattern recognition — like a student who memorizes answers. Newer systems actually work through problems: they consider possibilities, check their logic, revise their answers. Performance on math, science, and coding jumped dramatically.

"Agents" — AI systems that can take actions in the world, not just answer questions — became practical. An AI could be given a goal ("research competitors and write a report") and go do it: browsing the web, writing code, sending files, checking its own work. Human involvement went from constant to occasional.