There’s a lot wrong in this statement from Mark Zuckerberg on machine telepathy:
While some of these ideas might seem more like the stuff of sci-fi, Zuckerberg says there is scientific research going on in these fields right now.
Telepathy is one such area. “You’re going to be able to capture a thought in its ideal and perfect form in your head,” he says. “You’ll be able to share that with the world, in a format where they can get it.” [source]
One of the more obvious problems is that the “thought in your head” is really a network of ideas and experiences. More specifically of your ideas and experiences.
For this reason, ideas can’t be transferred, only recreated in a different but related form in someone else’s mind (i.e. networked with _their_ experiences and ideas in an analogous way). Language is a means by which we try to trigger that recreation process in a directed way.
This is to say that the thought, as it exists in your head, is _not the ideal form for transfer_, not by a long shot, even if the science catches up in the ways Zuckerberg predicts. The analogy is loose, but the thought in your head is as useless to another person as is machine code from a machine which exists nowhere else in the world. Given his background, Zuckerberg of all people should get this.
This is not to say we can’t do better. Interfaces that can track subvocalization of language may appear sometime in the distant or not-to-distant future, and could revolutionize the way with interact with machines. But ultimately it’s the language stream you’d want to tap into, not the thought stream, because it’s the language stream that is the recipe, and recreation is about the recipe not the cake.