Re: Energy constraints was Re: Androids & Strong AI in the Future

Michael Olea <oleaj@xxxxxxxxxxxxx> wrote:
Curt Welch wrote:

Michael Olea <oleaj@xxxxxxxxxxxxx> wrote:

"You are traveling through another dimension, a dimension not only of
sight and sound, but of mind. A journey into a wondrous land whose
boundaries are that of imagination. That's the signpost up ahead -
your next stop, the Twilight Zone."

That didn't sound right, so I looked it up. It was. For season 2.
Seems they had 5 different openings over the years:

http://en.wikiquote.org/wiki/The_Twilight_Zone

Yeah, I know.

Yeah, I didn't. Now I've learned something new. :)

Is having on-board wifi another sensory modality, or more like
a person with a wearable cell phone?

Of course it's another modality but it's also like a person with a
wearable cell phone. It's both.

Thing is, using a cell phone is mediated entirely through other
modalities. Using a sense of smell is possile for deaf, dumb, and blind
kids, using a phone is not.

Vision, audition, olfaction, and
touch allow for correlations among stimuli with a common cause -
things and events. A transceiver serving only as a communication link
is not a "sense" in this same way.

Of course it is the exact same way. It could walk near a transmitter
and see hear it getting brighter just like you notice a light gets
brighter when you get near it. It's just another eye that works at a
different frequency and which has very poor focusing power (unless you
gave it very complex set of directional antennas).

That's one hypothesis. Another is that it is not like that at all. The
key word is "only". Will was talking about this radio "sense" as
something that could connect to the internet, download stuff, browse
stuff. There is nothing in that about "getting brighter". Maybe the
minimal sensory aspect of a link is UP/DOWN (or ON/OFF), which could be
correlated with other stimuli and with robot actions. But LINK UP/DOWN as
a "sense" is a very different matter, it seems to me, from downloadable
content uncorrelated with the present world, the here and now of a
robot's position in space and time. The stream of packets from a blog in
china has at most a marginal connection with a situated entity.

Well, I believe the brain creates all these concepts of "situated" from
nothing more than undefined streams of data. I believe that general
intelligence can be created. I believe that it is possible to create a
generic intelligence algorithm where you simply plug sensory data into the
generic inputs (no need to connect vision to the vision input and smell to
the smell input), and all the concepts of being situated in a 3D world etc
will correctly emerge from the data. I don't get the impression that many
other people here agree with this view.

Now, I do agree that our brains our heavily optimized for just the type of
data we deal with. But I believe underneath the optimizations, it's using
some powerful general intelligence techniques to learn and constantly
improve it's behavior.

To me, sensory data is _any_ data you feed to a general intelligence
algorithm. If you want to give it a wifi connection, it will be a sense
(in the full situated sense) no matter what form you feed the data to the
general intelligence algorithm. You simply need to convert it to whatever
data format the system uses (such a pulse signals for my type of network
for example).

You can send high frequency raw RF data, or you can build an interface that
acts like a PC running windows with a wifi internet connection. The first
form would be a sense more like our hearing, the second would be a sense
that is like us using a computer - except it would be permanent part of us
that was always with us (like our eyes, or ears, or arms).

Now, if the machine could correlate a flash of
light, a burst of microwave noise, and a clap of thunder, or sense a
microwave oven nearby, use its receiver to survey the "scene" at or
near 2.4 gigahertz (the frequency of 802.11b - I don't know the /g
part), correltate it with other stimuli and with its own actions, that
would be another matter.

If it had a wifi receiver wired up as a sense why wouldn't it be able
to do that?

That's just it - it would have to be "wired up as a sense" to do that;
and "wired up as a sense" is not the same as "wired up as a bidirectional
packet stream". In other words, if you have low level wifi modules that
handle, in good modular fashion, all packet level assembly protocols,
passing up only the assembled byte stream, then the machine no more has a
"radio sense" than you or I have a radio sense when we jack in at some
internet cafe.

Right. It has a sense of the data you send it. If you send it direct
radio data, it has a radio sense, if you send wifi packets, it has "wifi
packet sense", if you send it a 2D view of a browser window, it has a
"internet browser sense" (which it sounds more like what Will might have
been talking about - I've not followed the whole thread - I just jump in to
try and cause confusion at times).

Now, if the sense was just a stream of IP packets coming from a wifi
receiver instead of raw radio data, then it would be a different sort
of sense, but it would still be a very real sense. It would see
someone typing on a laptop using a character protocol and "hear" the
PING PING PING
of the IP packets. Even without decoding them it would no doubt learn
to recognize many different standard patterns (web access, email
downloads, etc).

Just as you or I might do, without claiming we have RF senses.

I don't imagine you want the machine to have to learn protocols it can
simply download any more than a person needs to learn
telecommunications to order pizza.

Yeah, if all you wanted was an internet link then you could just build
in all the wifi protocols. But if you wanted an RF-aware learning
machine then it should be able to learn the protocols as well as any
other new protocols that came along if you had enough learning
hardware. It would probably take a lot more hardware for it to learn
the protocols than if you just built custom protocol code however.

Again, learning the protocols would be like learning telecommunication
theory so you could place a call to order pizza - just not
cost-effective.

Right. Unless you want an AI that had a keen sense of RF so it could hear
and learn all the chatter going on in the air. I'm sure the NSA would love
to have a lot of super AI brains optimized to that type of problem for
example.

The protocols are documented, they do not have to be
discovered via correlation studies. To do so would be like turning the
task of writing a bubble sort into inventing a language and writing a
compiler for that language, then writing bubble sort in that language. It
would be silly. A machine could be RF-aware without having to learn IP by
trial and error. But if it were RF-aware it could be one heck of a
troubleshooting technician.

Right. If all you wanted is to have a link to the internet that worked
like our computers work for us, you don't want to flood it with all the RF
or IP data. But if you wanted a technician to help monitor your networks
and report problems before they became problems, having it optimized for a
direct Ethernet sense would be a huge win. Or other AIs that were
optimized and trained with an wide brand RF sense would be great radio
technicians.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.

Relevant Pages

  • Re: XP SP2 seems to cause unwanted traffic to printer
    ... There were 1220 packets between the printer and this PC in 60sec. ... It just seems odd that this wasnt a problem till SP2 was loaded. ... even when it's supposed to be disabled it really isn't, especially at startup, and to get around some of its limitations we've had to write ..reg files to modify the firewall behaviour in order to pass some Netware protocols fluidly from SP2 boxes by specifying ports, packet types and service names, and even then it's sometimes hit and miss. ... What I suspect is the activity you're seeing could be the result of timeouts and retries of the printer protocols trying to communicate with between the printer and PCs through the kludgey XP firewall. ...
    (microsoft.public.windowsxp.general)
  • Re: Rate Monotonic Scheduling (RMS) vs. OS Scheduling
    ... anothe in house scheduler algorithm) or using the OS scheduling. ... "dynamic control" layer is responsible for handling protocols' control ... duration and deadlines in advance. ... Of course this only applies after packets have been received. ...
    (comp.arch)
  • RE: [fw-wiz] tests about latency
    ... > to other protocols? ... ICMP handling is generally a special case for most ... most devices with minor changes to the packets you're generating. ... different sized buffer, will it affect overall performance for the other ...
    (Firewall-Wizards)
  • Re: NDIS IM: queueing question
    ... This is why TCP protocol has the concept of ACK packets. ... upper level protocols should guard against this and not NDIS. ... > Maxim Shatskih, Windows DDK MVP ...
    (microsoft.public.development.device.drivers)
  • Re: Energy constraints was Re: Androids & Strong AI in the Future
    ... That didn't sound right, ... If it had a wifi receiver wired up as a sense why wouldn't it be able to do ... typing on a laptop using a character protocol and "hear" the PING PING PING ... then it should be able to learn the protocols as well as any other new ...
    (comp.ai.philosophy)