PSA posits that any quality DLE must be designed with some underlying understanding of media, and its affects on human communication. Learning often involves best communication practices. Adaptive learning involves supplying resources in real time using algorithms to find just the right “clip” or file.
This story from “The Atlantic” covers some of the most current “telemarketing” and “call center” technology, that uses pre-recorded phrases, that are interjected into conversation  with or without the human’s awareness. Clearly something similar could be used as part of adaptive learning. Curation and Distribution of appropriate resources for just in time learning could easily incorporate voice files too.
Perhaps this could even be used as an adjunct to “facilitation roles” in a SLC.
[gview file=”https://publicservicesalliance.org/wp-content/uploads/2015/04/Almost-Human-The-Surreal-Cyborg-Future-of-Telemarketing-—-The-Atlantic.pdf”]But the overall ambience of communicating with a computer using vocal  interactions that are controlled by but not produced by a human…is generally a weird brave new world situation. Like talking with “Siri” perhaps, but with “Siri” we supposedly know it’s a computer.
OTOH, after a while, there’s a human tendency to start acting “as if” a computer is human, if it’s able to provide a rich enough conversation, and avoid obvious “give aways” of its artificiality. And there’s the “Turing Test” or definition of “intelligent”…if we can’t determine when it’s a computer and not a human, then what’s the difference? (paraphrasing).
This is a current theme in much of sci fi media these days…”how are we going to live with robots”? Also addressed here in a NYTimes article on machines taking over more and more human jobs.
[gview file=”https://publicservicesalliance.org/wp-content/uploads/2015/04/The-Machines-Are-Coming-NYTimes.com_.pdf”]Another resource: David Chalmers bibliography of “The Turing Test”.