AI: Slowly, Then Suddenly All at Once

There has been a flood of interesting AI hitting my Twitter feed this last month or so. And once I started collecting some interesting AI folks involved in, or experimenting with all these new advancements in a Twitter list, it started to become clear – things are getting very interesting in the AI space.

Let's look at the timeline of releases:

2017

 August 11th - OpenAI's bot beats professional Dota 2 game players in 1v1 games - https://openai.com/blog/dota-2/

2019

April 15th - OpenAI's "OpenAI Five" AI bot beats an esports team in a livestream for the championship in the game Dota 2 - https://openai.com/blog/openai-five-defeats-dota-2-world-champions/

2020

June 11th - OpenAI announces public API to it's language model GPT-3 - https://openai.com/blog/openai-api/

2021

January 5th - OpenAI announces a neural network called DALL-E that creates images from text - https://openai.com/blog/dall-e/

August 10th - OpenAI announces improved "OpenAI Codex", their AI system that translates natural language to computer code - https://openai.com/blog/openai-codex/

November 18th - OpenAI's GPT-3 API widely available - https://openai.com/blog/api-no-waitlist/

2022

April 6th - OpenAI releases version 2 of DALL-E - https://openai.com/dall-e-2/

July 12th - An independent research lab named Midjourney Has Public Beta Release of their tools of the same name - https://www.midjourney.com/home/

July 20th - OpenAI release beta of DALL-E - https://openai.com/dall-e-2/

August 22nd - Stability Announces Public Release of Stable Diffusion - https://stability.ai/blog/stable-diffusion-public-release

August 31st - OpenAI releases "outpainting" feature to DALL-E - https://openai.com/blog/dall-e-introducing-outpainting/

September 21st - OpenAI releases "Whisper", a neural net that does automatic speech recognition - https://openai.com/blog/whisper/

Looking at the timeline, it's accelerating. 2021-2022 were real banger years for AI doing very impressive things. Before that, also interesting things, but no one was really paying attention. But since the paid APIs, and open sourcing of these AI models, things are moving quickly, and this year is not over with yet.

Why does it all seem so sudden?

I think like all technologies, it goes in starts and fits. Remember folks talking about the AI winter? And I even found this article from twenty fifteen about a "resurgence" - https://www.wired.com/insights/2015/03/ai-resurgence-now/. This article talks about an injection of money and related data tech that was open-sourced such as Spark.

Models. Models everywhere. But specifically around text and images. First it was GPT-3 that was a public API that allowed folks not only use the training already there, but train their own models on top of this one.

It’s all only accelerating.

Tweet of the Week