@orangy I think there will be a number of trends. I think folks will start to realize a neural network isn't optimal for every type of problem, and that "data science" means something radically different to each person and industry. I don't care about natural language processing in my line of work. But statistics and optimization... we can't operate without it.
I also think there will be a realization that data processing and mathematical modeling increasingly needs to be systematic and engineered for production, and notebooks have severe limitations and are not codebases. This will be primetime for Kotlin.
I find it interesting that the good "data scientists" I've come across often don't want to be called that. Theyd rather be called by their specialization. This is how things used to be, and I think it will snap back ultimately.
I ranted a little about this earlier this year.
http://tomstechnicalblog.blogspot.com/2018/01/is-it-time-to-stop-using-term-data.html