[TheOtherHobbes]

You're assuming the intelligence will be independent and not owned by the advertisers.

That may happen eventually, and I'd be surprised if commodified IQ enhancement with reasoning skills wasn't available within - say - 10-15 years.

But there has to be a mainframe-like stage before then, because the first versions of the technology will be industrial (Watson++), not domestic.

Google are well placed to be big players at that stage. More speculatively, so are Amazon and Facebook.

MS are maybe half-way to thinking that far ahead. Apple don't seem to be, because they're too fixated on shiny trinkets now and not on long-term strategy. (Siri was a creditable v0.1 attempt but doesn't seem to have improved at all since release. Watch is - to be polite - not a player in this market.)

But even after that stage, how can you be sure that a commodified agent won't be acting in the interests of its creators/sellers/sponsors instead of its owner?

[hyperion2010]

I would like to argue that if companies manipulate agents that people are supposed to trust deeply then they will rapidly loose customers. At an individual level I'm worried this wouldn't happen, but the massive losses the US tech sector has suffered following Snowden give me hope that we won't just roll over. At the end of the day my hope is that something akin to HIPPA (bad example I know) will be put in place to protect data that is shared/collected by a personal digital agent.