2020: I started toying with GPT-3. Getting it to craft a poem or churn out a catchy paragraph felt incredible. I thought this was real AI (how naive).
Then ChatGPT burst onto the scene in November 2022, together with all the hype. I thought, this is real AI (and I'm sure I'm not the only one who felt amazed).
Since then, I've kept experimenting with what LLMs can help people do. They can make it easier to explore an idea or get past a blank page. They can also be confidently wrong. A plausible answer still needs checking against the task and the evidence. Useful, certainly. Real-ish, for sure.
Enter Claude Code and tools like Gemini CLI: helping write code, investigate bugs and work with GitHub or Cloudflare from a prompt. Impressive capabilities, but the code and the result still need review. Is this real AI?
But nothing prepared me for this week at WAIC, the World Artificial Intelligence Conference in Shanghai. What I've seen here is AI stepping out from behind screens into the physical world. AI embodiment.
Robots of different sizes and functions: playing musical instruments, giving massages, packing boxes, helping elderly people.
Out on Shanghai's streets (which, by the way, struck me as impressively clean), I caught sight of robotaxis and noticed plenty of electric vehicles. Those were impressions from a visit, not a measure of adoption across the city.
Even my hotel, which is comfortably average, greeted me with a robot casually rolling into the lift alongside humans as though it were the most normal thing in the world. A small encounter, but a memorable one.
At Tencent HQ I saw, among many things, a demonstration of remote truck operation. I also heard Ian Tong discuss enterprise AI, consumer AI and governance. That stayed with me: what controls do we need when an experiment becomes something people rely on?
And in the toilets… well, let's not go there. I can't help noting the control panel though: the lid opens and closes automatically. Somehow I felt a bit eerie and certainly prefer my non-smart toilet back in Melbourne.
What I took from the visit was the value of moving beyond discussion into practical experiments. Start with a clear task, put appropriate controls around it, observe what happens and learn. A demonstration is not proof of safe or reliable operation at scale, but it can help us ask better questions about the next step.
Massive thanks to Xue Huang and MLAI, who made this eye-opening trip possible.
First posted on LinkedIn in July 2025. Revised for this site in November 2026.