Comment by dataviz1000

I don't think it is 2D vs 3D, but rather 2D and 3D vs 4D which probably has these guys a little worried. The human interface is still the same but now instead of something that responds instantly or deterministically like the doorbell interface which has tactile input and auditory output, the output is automated actions in time. This is what the thinking output in LLM interface is. I don't know if the language and semantics allow me to say this but the LLM output is more than a 2D screen, it is also in time. That is the fundamental change.

This might not be different from say ... clicking a button causing a massive Rube Goldberg machine to take money out of my bank account, send emails about my booking to all interested parties with signals moving across continents before in some time in the future sending me a confirmation message when everything is finished. People are beginning to think in terms of the output of the action of pressing a button to book as automation instead of just output on a screen that says successful or failed attempt to book.

(If anyone from the Airbnb design team reads this, please, please, please, work on making sure that the absolute important information about a booking in text messages are visible in the text message. And that the page before booking does not hide information about the booking on an iPhone 13. You do not have control of the thousand different combinations of phones, operating systems, ect so when the information has be presented in only words it has to be done in a way that people understand them.)