[ExI] Superhuman Poker

Dylan Distasio interzone at gmail.com
Thu Jul 18 16:46:56 UTC 2019


On Thu, Jul 18, 2019 at 12:23 PM John Clark <johnkclark at gmail.com> wrote:

> On Thu, Jul 18, 2019 at 11:02 AM Dylan Distasio <interzone at gmail.com>
> wrote:
>
> > *Once (big IF) we see Level 5 self driving cars that can perform as
>> well or better as humans in all conditions and circumstances, I will start
>> to worry about the rise of AI and the singularity.  *
>>
>
> I'm sure today's self driving cars are not as good as the best human
> drivers but they are already better than many human drivers and perhaps
> better than average.
>
>
The problem with self driving cars is that they are also brittle.  When
they fail, they fail spectacularly because they have no common sense or
actual conception of what they are doing outside of their brittle model.
That's why we get Teslas ramming into barriers at high speed without even
braking killing their passenger(s).  There is zero knowledge of self or
passenger.   There is also zero intelligence behind the image recognition
which is another reason they fail spectacularly when something doesn't make
sense to them.

That said, as a way to drive on uncomplicated clearly marked highways, they
do a pretty good job and I'm sure could save people (truck drivers in
particular) who fall asleep on the road.  I think they will change the
trucking industry shortly, but a driver will still need to be present and
ready to take over from the off ramp forward or in bad weather.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20190718/0c6c44c4/attachment.htm>


More information about the extropy-chat mailing list