Maybe when people are younger and haven't observed a bunch of business cycles, they are more frustrated with hype.
To me it's just another flavor of the signal/noise problem. You can discern the signal by spending time in university labs, selective reading, following enduring industry trends and companies that survive, and reflecting on history - and by that I mean back into the 20th century. He he.
As someone who explored and then left AI in the early 1990s (even around MIT and its ecosystem - it was nuclear winter and computing was simply not where it needed to be to go faster anyway), I am still excited by this field and what we can do now. That is not marketing fluff; it is real. (Although people who write marketing fluff can also be genuinely interested in the subjects and are just paying their bills and buying vacation homes far from the Madding crowd, just like other sensible forward-thinking people).
Here are two books from my library that map out the early days of AI and machine intelligence. The first goes way back and has delightful interviews:
Talking Nets: An Oral History of Neural Networks - The MIT Press
Strategic Computing DARPA and the Quest for Machine Intelligence, 1983-1993
The second focuses on the role of government funding in developing the field over that decade - so big money and then really big money for awhile and then deemed a failure and pulled back. But still the quest survived!
A question I think about sometime - "What if I had stayed the course after 1992?" One thing is for certain - for better or for worse, I would not be who I am now! : D
Introduction to Algorithms (Cormen, Leiserson, Rivest, and Stein) is a little bonus here - now in its third edition (this is the second edition) and was part of my second foray. Great reading for a Sunday morning, or midnight anytime... ; )
Introduction to Algorithms 3rd edition - The MIT Press noted for its rigor and comprehensiveness.
And now I am in my third foray. Will it be my last? I guess that depends on longevity!