Thank you for the good presentation.
Would you say that real-world use cases can be solved just by focusing on data-centric approach? Given that model selection can be automated, what role can modelling play?
1 Like
Hi there,
In my opinion still models need and will evolve, the purpose of the model is to emululate as close as possible natural inteligence. When it comes to AI the natural part (and the most important is missing) i.e. originality. If its a mechanical process with repetitive nature AI is pretty good, when not the difficulty comes in. We need models to extract more and more in depth knoweledge about the phenomena, so definitely there will be model development to come. When it comes to data you need to be selective and efficient in it, to use data which makes contribution in the learning process and discard data which is junk. Sometimes theoretical synthetic data may also be used to accomodate future scenarios.