ChatGPT is a big morale boost for the AI industry

Trust you have all tried ChatGPT, the latest release from OpenAI that allows (pretty impressive) dialogues and even drafting of arguments.

While a lot of the focus has been how fun it is (or how stupid it is), as well as what jobs it will eventually replace, I am personally excited about ChatGPT because:

It is really a big morale boost for the whole AI industry.

While image generation has been around for a while, with decent results since the days of GAN, ChatGPT is the first time conversational AI has become usable.

ChatGPT is a very powerful prototype model with an evolution path similar to that of AlphaGo and Alpha Zero. Actually since the release of GPT3 in 2020, researchers have been working very hard to train the model – before they are now comfortable to release it to the public.

A key consideration about releasing such models to the public is about sensitivities – it can’t tell things that are pro-Nazi, pornographic, discriminatory or just illegal. This means a lot of time needed to clean it up and correct the model – even though you still see ways of getting around that now.

Moving forwards, ChatGPT could be charging its users by API usage – pretty much the same as what Google Maps has been doing. In anyways it makes much more sense for others to pay for usage than rebuild the model themselves.

It’s a game changer.