Because of short-selling, index funds could own all the market or more and active funds could still exist. For example we could imagine that index funds grow until they're 105% of the market, and active funds own 5% of the market long and 10% of it short. That adds up: 105% + 5% - 10% = 100%. And the active funds would still be doing all the price discovery. Of course the index funds wouldn't have all the assets-under-management, those would be split in the ratio 105:15 (with 10 of that 15 being collateral for the shorts).
In other words: it doesn't matter if index funds own all the shares, active funds can just make side bets with each other.
With the progress AI is making, I predict a future where humans-run firms will mostly go out of business. What will emerge are companies that are great at training models that then act as active investors. These models bet against each other and some emerge as winners.
Ultimately, the models that make the most bank (sorry couldn't resist) will be the ones smart enough to make its decisions based on public data from companies, social media etc.
This is already happening, HFT, Two Sigma Investments, many hedge funds. And they don't just use public data, they work really hard to get as much proprietary information as they can. Just one example: https://www.cnbc.com/id/38722872