1
$\begingroup$

One Czech sci-fi author presented idea on his feed on how people might deal with contacts in not so far future:

Imagine you can have your own "personal assistant AI" loaded to your smart phone. It knows what you want and what you do not. It can mimic "you" to the outside world:

  • It can pick up your call and decide if it transfers call to you (the phone starts normally ringing) or settle the deal with its counterpart
  • It can also answer texts
  • It can read your emails

For all above, it can give you summary. Either to text or it will just create calendar entry in your calendar. So if your mom calls, you will just end up with "Buy milk" agenda item in your calendar.

For more examples, call with your Significant Other can be handled just by your AI, ending with calendar entry "romantic time tonight"

This AI can deal with annoying marketers. It can represent you when dealing with government or your boss. It will just give you a summary of what is needed to do.

You can decide on "priority people" upfront or just train your AI to make it know that you want to have chat with your mom, but not with your boss, until absolutely needed.

While this personalized AI sounds like really distant future, the fact is:

There is already Jet Fighter AI better than human running on Raspberry Pi so having your personal assistant running on your phone does sound plausible in next 5 years.

But, Will this technology attract enough people to make it plausible to upgrade it to level where this AI cannot be distinguished from real you?

Also, please make sure that you know what your personal AI can do right now. Here is one example what people already do: https://www.youtube.com/watch?v=VFPCLy8Ej4g

Improve this question
$\endgroup$
7
  • $\begingroup$ For those who may be curious: The name of author is Patrick Zandl, However, I was unable to find any English info about this author to provide proper link $\endgroup$
    – Pavel Janicek
    Commented Jul 13, 2016 at 9:29
  • $\begingroup$ SMBC also covered this, so we know quite how bad an idea it is $\endgroup$
    – Separatrix
    Commented Jul 13, 2016 at 9:46
  • $\begingroup$ Read The Atopia Chronicles, by Matthew Mather, for inspiration. In this sci-fi world there is a country where everybody has at least one such assistant AI (built from fragments of their own minds), and some people have multiple ones. And they are developing the personal AI technology to export to other countries. The personal AI's cna do everything you mentioned above and much, much more. $\endgroup$
    – The Square-Cube Law
    Commented Jul 13, 2016 at 12:11
  • 2
    $\begingroup$ People already let someone else handle these tasks for them. Computers have already taken over some of a personal assistant's duties, such as maintaining a schedule, producing reports, or sending office notifications. With a little better natural language processing they could easily take on the additional task of interpreting requests. $\endgroup$
    – Kys
    Commented Jul 13, 2016 at 14:24
  • 2
    $\begingroup$ I don't think people would call an AI though. Want to set up a date night? Tell your AI to contact your significant other's. The two AIs can coordinate on scheduling and reserve your table at a fancy restaurant that suits both your tastes. The AIs process this much faster than you could. Ah, romance... $\endgroup$
    – Kys
    Commented Jul 13, 2016 at 14:29

3 Answers 3

Reset to default
1
$\begingroup$

Will this technology attract enough people to make it plausible to upgrade it to level where this AI cannot be distinguished from real you?

I guess that an auto-responder does not have to be that advanced. It might be developed to be indistinguishable from you in 95% of cases - sure. But it's unlikely that someone would bother programming another you, with all your knowledge and personality, just for the phone calls. Auto-responders just don't worth the efforts.
If you're, for example, a ballerina, and one calls you and asks "what is urea-methanal chemical formula?", it's perfectly ok if your auto-responder answers something like "dunno, google it" or "google says, NCH2N", even if you actually know the formula.

If you still want AIs that cannot be distinguished from their owners, you'd probably need a better reason. For example, they might be chatbots instead of auto-responders. Even if an AI of a great scientist cannot create new theories, it still can answer difficult questions. This way, people could learn from the greatest minds and get expert's opinion on anything they'd like. They could chat with their favorite celebrities, or with their own chatbots, or say anything to the friends' bots just to see how'd they react.

Improve this answer
$\endgroup$
1
$\begingroup$

I think it would be the other way round. Not: will this technology attract enough people to make it plausible to upgrade it to level where this AI cannot be distinguished from real you? But: the technology would only work, people would only be interested, when it got to the point where your mum couldnt tell whether you or your AI receives the call. It could be done in a really bad way by now: You set up an answering machine and if the message the caller puts on it contains the word "milk", milk will be put on your groceries list. Now we upgrade this technology until we get a version that is basically a digital copy of the user. Somewhere in between this and the "bad way" example there is a point in which the technogy is advanced enough to have a practical use.

Improve this answer
$\endgroup$
5
  • $\begingroup$ Are you aware of fact that "Siri, remind me to buy milk" already works and is used today? $\endgroup$
    – Pavel Janicek
    Commented Jul 13, 2016 at 10:47
  • $\begingroup$ @PavelJanicek yes, but I think this is about tricking other people into thinking they are talking to you while they are actually talking to a digital copy. $\endgroup$
    – user22893
    Commented Jul 13, 2016 at 10:50
  • 3
    $\begingroup$ then comes the natural extension: you die or are murdered alone, in your house, and nobody realizes it because your AI can impersonate you. $\endgroup$
    – Marshall Tigerus
    Commented Jul 13, 2016 at 13:46
  • 2
    $\begingroup$ @MarshallTigerus plot twist: your AI murders you and assumes your identity $\endgroup$
    – Kys
    Commented Jul 13, 2016 at 17:54
  • $\begingroup$ @Kys A la Marionettes Inc.? $\endgroup$
    – Xandar The Zenon
    Commented Jul 16, 2016 at 22:36
1
$\begingroup$

I cannot argue about personal matters, but, I believe there would be a lot of companies who would be interested in a virtual secretary. After all, after the initial cost, there will be no upkeep.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ Press 1 for HR, 2 for Marketing, 3 for the Research Department, etc... $\endgroup$
    – Peregrine Rook
    Commented Jul 14, 2016 at 4:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .