This is a great point. I'd add audio. AV is trapped in the internet like oil before fracking. Finding the right videos or audio clips is being back on the 1997 internet. Try it: search for a specific sound-byte across podcasts, look for videos in which X person appears.
Internet users consume way more video than text, yet the web's core infrastructure is still built around text.
What does the web of video look like? At a minimum, we need better search engines.
This is one of the most interesting domains to apply AI breakthroughs today.