Data-Centrism VS Model-Centrism

I remember in 2016 how a conversation ended before it even started! It was with that proverbial "its the dumbest thing" look that stopped the interaction. It was my old Germany friend (name withheld!) on hearing that I had started an IoT Makerspace in Mbare, Harare. German friends can be brutal! Explaining that I was motivated to start the Makerspace by a search for quality data was like extinguishing fire with petrol! He didnt see the connection between data, IoT, Makerspace and impoverished youths, I crawled back into my wet shell!

Now that there are now lots of authoritative voices like Andre Ng and Mike West that are putting data is at the centre of intelligent products, I can come out of my shell and revisit my IoT Makerspace experiment with less shame. As data systems mature and scale the fact that "data is the new oil" is now a foregone conclusion and slowly becoming monotonous. What is now more pertinent in data initiatives is what to put at the centre of AI initiatives: data or models?

Andrew Ng: "AI Models a Solved Problem"

Andrew Ng, the co-founder of Coursera became an AI superstar in 2012 because of his famous CS229A course. The free course on Machine Learning, originally offered at Stanford, had a staggering enrollment (then) of 100 000 and even reached yours truly in remote Africa!

A lot has happened since 2012 and Ng's current focus appears to be to challenge enterprises to go beyond talking about AI but to create real AI products. Ng believes developing AI models or software is now a "solved problem".