What is it?
The main purpose behind working on this project was to think differently about product management in our household. We were building an assistive learning companion that is very smart, has all the knowledge, and is extremely good at reasoning. Later work was how to enable it to manage the household.
It is not a chatbot.
I think after the boom of only famous products by OpenAI since November 2022. The view of what AI could do has been narrowed down to Chatbots. The turning point was, of course, Transformers. Now, the main limitations of whether AI will get smart or not depend on processes like scaling and scaffolding. Majorly, scaling. The tremendous pace of the landscape has enabled the pouring of vast resources such as h100s(which is more important than oxygen these days).
The vision of Alex-1 requires more than millions of h100 to make it reason like Einstein. Quantum Computing and many advancements are needed. My speculation is the barriers and thresholds of hardware will demand an astronomical amount of money.
Team Work & Application
Get the team of A players.
Whenever new tech emerges from an existing industry, being world-class at creating products in an existing industry is the prerequisite.
The loop that formed during ALEX-1 is:
Application⇋Execution⇋Results⇋Feedback⇋Learning⇋Re-application