If we zoom really far out there are only two problems in the world:
- Figure out what to do (problem formulation)
- Do it (problem solution)
Modern AI is always in the “Do it” category – because problem formulation requires a will to change – when AI gets there I will update this article.
Most new challenges (99%) we are put in front are in the “Figure our what to do” category. And once we figure it out we will not need AI for 99% of the solutions we come up with since the solutions are linear or straight forward to solve with known tools that have precision and logic on another level than what modern AI has.
If my math is correct we will use AI on 0.01% of challenges we face.
This 0.01% will be insanely helpful though – since it will helps us “Do” things we have never been able to learn computers do before. Almost all of these things boils down to classification of unclassified data (pattern recognition).
If we can get a computer to classify data then we can make it create wonderful music. Create an AI that can take any sound-stream and classify if it is wonderful music or not – then we could leave it with a noise source and come back and check it once in a while – it will have found stuff to classify as wonderful music for sure. With back-propagation it will be very much faster than just waiting for chance. A person could do this too – but not as fast and not as long and not as cheap – and back-propagation is what a person does when she improve her ability by reducing her weaknesses.
It is a revolution!
But only for the 0.01% of the things we do – so stay in school – you will be needed a while longer.
What MDriven does with models is actually pure “Figure out what to do”—stuff (the 99% percent of what problem solving is all about), then we take the models and “Do it” with well known established software strategies that has evolved the last 40 years – in seconds – so that you do not have to. If you use MDriven to reduce the tedious “Do it” phase you will have more time to think on how to apply AI to your product.
