Is it possible to use Machine Learning and AI to trade profitably in stock market
Hello, I am new here. I am a financial student in a University and I don't know much about machine learning or AI until I see how OpenAI beats human players in a computer game called DOTA2. (Yes, I am a dota2 player). It's fascinating how AI can practice 247 all year along to adapt human player's style and strategy, using that information to evolve. In the game, human most of time don't stand a chance against Machine learning AI.
It sounds very strange to me, the developers of OpenAI for Dota2 are not professional dota2 player, that means their program doesn't know anything about the game but they learn their own!
So my question is, is it possible to use AI to learn behavior in stock market and profit from it consistently? Stock market behaves very similarly to DOTA2, where all information is generated by human behavior; thus, AI should be able to use machine learning to absorb all the information in the market and then produce a best strategy itself to beat all other stock market participants. If it's not possible, then what are the limitations or difficulties for AI doing so in stock market?