In the corporate world, the failure mode is a crappy job that still pays as well as the AI gig; it's just not as sexy. Then you can slack, learn the skills for your next gig (it's not like you need to be in academe to read papers) and get something better. In academia, the failure mode is being dumped with no money and no reputation in some godforsaken No Country for Old Men reach, being 35 and trying to get an entry-level gig in something.
I wholeheartedly support the concept of slackademia...using idle cycles at work to learn about new, exciting things.
It's entirely possible to leave a paper up in a tab, read it over lunch or just when taking a break from your work, and continue.
This isn't rocket science--nobody is >80% all day erry day in their time usage, regardless of what they claim.
I suggest that both are true, though one or the other may be a bit more efficient for a given situation.
What I dislike is this semi-unspoken assertion by academics that, unless you have the long hours of "freedom" provided only within the ivory tower, people in industry cannot be creative or do useful research. It's insulting to everyone who has made continual progress for mankind while working normal jobs.
my impression of research at facebook/google/microsoft is that they generally hire phds and they generally take phd students as interns, so if you drop out of a phd program, you're not working there (in a research role at least). so you have to finish your phd, then compete with everyone else for those jobs, just like postdocs and professor positions. once there, you have to justify your existence / funding the same way that you do as a professor (go ask people at MSR how this is going for them lately).
what wondrous land do you propose that will support researchers and their desire for a few months out of the year to think about stuff? at most companies, you won't get a few minutes out of the day to do anything other than pack numbers tightly into excel or go to your next standup.