The pace of AI
Seven years. A different world.
Scroll the last few years of AI — from “too risky to release” to agents that do the work.
GPT-2 — “too risky to release”
A model that could write a coherent paragraph felt dangerous enough to hold back. Today it looks quaint.
GPT-3 learns from examples
175 billion parameters. Show it a couple of examples in the prompt and it just… does the task.
Text-to-image, and an AI pair programmer
DALL·E turned words into pictures; GitHub Copilot started finishing developers’ code as they typed.
Stable Diffusion, then ChatGPT
Open image generation for everyone — then a chatbot that reached 100M users faster than any app before it.
GPT-4 — now it can see
Text and images in, expert-level answers out. The year AI showed up at actual work, on real teams.
Voice you can interrupt, video from a prompt
Sub-second spoken conversation, and text-to-video that stopped looking like a gimmick.
AI that takes action
Agents operate software on their own; humanoid robots start real shifts on factory lines.
New capabilities ship every month
Keeping up has become a job of its own. The best way to do it isn’t alone.
See it for yourself
Watch it evolve
Pick a capability, then step through the years.
PromptExplain, in a sentence or two, why the sky is blue.
The sky is blue because the sun is blue and the light from the sun makes the sky blue, and when the sun goes down the sky is not blue anymore because there is no more blue light, and the ocean is blue too because the sky is blue and they are the same thing.
…which is why we get together.
Two days to catch up — hands-on, with the people building it. October 6–7, 2026 in Charlotte.