[ExI] Near Future AI 'smartphone' Assistants

[Mike Dougherty]

On Sat, May 10, 2025, 9:05 AM BillK via extropy-chat <
extropy-chat at lists.extropy.org> wrote:

> The development of personal AI is becoming increasingly clear.
>
> These future AIs will do much more than merely respond to prompts.
> They will operate autonomously in the background, acting on your
> behalf and pursuing your goals with independence and competence.
> Your primary interface with the world will still be a device – the
> smartphone successor – which will host your personal AI agent.
>

Do you really think the entire agent will be on the 'device'?

I suspect the device will store personal configuration to shape the
interactions - like a facemask or cosmetics for an assistant that is
otherwise identical model to millions of consumers.  Sure, wealthy
consumers will have smarter models but it's likely the world won't need
more than three or four.

Consider DVR: Tivo was local storage for as much content as you could
afford to keep.  Then the cable boxes copied this idea.  Can't get a new
device without losing content or needing to manage it.  Now the data is
stored in the 'cloud' so you can upgrade the device simply by putting the
account credentials on the new device. Desktop computers are difficult to
move out/in, phones are easy.  I think this dependence on ubiquitous
internet will grow for personal ai agents.

 Your AI will be 'private' in the
> sense that Google, Meta, etc. won't be tracking your every
> transaction. But it won't be private if your device is ever seized by
> the authorities for any reason. The seizure reason could be quite
> trivial, just an excuse really. Then your life could be examined in
> total detail for any and every action that the authorities might
> decide is an offense.
>

It will be as private as you were strolling around the mall before that
physical experience went defunct.

There will be no need to seize your device. The Nth Party (far beyond third
party) will access the cloud version of you.  In the time it takes to
understand "network error" and refresh, the digital tracks of your life
will have been judged.  You might not even notice that the consequences of
your infraction include fewer options in the future.  You're already in
jail, the warden controls the TV because it's your only view of the
outside.  Why though? Because controlling the memeplex of humanity is much
easier than running the human zoo. Ultimately the results are the same.

>
> Some people might just refuse these future AI assistant devices.
> That will make life much more difficult for them in a society where
> these devices become indispensable. It will also be considered very
> suspicious behaviour in this new form of society.
> What have you got to hide?
>

Those outliers will still have profiles, but the details may be slightly
less nuanced or just slightly lagging on updates

>
> The opposite choice is to allow the personal AI assistant to become
> part of your life and record 'privately' every detail. This should
> make populations very well-behaved and become extremely cautious
> in order to avoid any action that might cause offense.
>

Behaved and cautious?  That doesn't seem to be the direction.   Offense has
become normalized.   The algorithm has bolstered trolls and extended their
reach.  Angry react generates far more metadata than agreement.  If the
purpose of the machine is to manage the overwhelming deluge of information,
wouldn't that serve its purpose to generate an even more overwhelming
downpour?

>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20250510/4c72022e/attachment.htm>