Re: New Scientist Note: Singularity by 2010?

From: Hal Finney (hal@finney.org)
Date: Tue Jun 11 2002 - 18:05:33 MDT


Charles Hixson writes:
> http://www.newscientist.com/hottopics/sciencedebates/article.jsp?debate=3
>
> contains the fragment:
> Ian Pearson: ...Looking in turn at those--well, I'm a prime culprit. I'm
> working in the field of artificial intelligence. I have a prototype
> design for something that might be 50,000 million times smarter than the
> human brain. Target date is 2010.

This is probably the "resident futurist" at British Telecom that we've
discussed a few times over the years. Actually I think the above may
be a transcription error in what was apparently a verbal discussion.
If you look at http://www.btexact.com/ideas/futurology?doc=21045 you find:

: By 2015, computers will catch up in overall intelligence with humans,
: and will be designing their offspring. These will arrive faster and
: be even smarter than would be possible with human designers, and these
: offspring will design the next generation faster and better still. The
: impact of this positive feedback loop is hard to overstated. By 2020,
: machine knowledge will greatly exceed human knowledge, and even the
: smartest people won't be capable of understanding some of it. By 2048
: computers will be approaching the limits imposed by physics. We don't
: know what these limits are yet, but at least 1000 million times the
: capability of the human brain is possible in a single device.

So I think he may have meant that the ultimate limit of the device was
50,000 million human capacity, but that would be decades away. Maybe
by 2010 he thought he could build an early version.

To get an idea of Pearson's prognosticative abilities, you can look at
http://www.btexact.com/ideas/whitepapers?doc=42270, where the link at
the bottom is his full timeline. We'll have AI doctors by 2001, viruses
aimed at toys by 2002, robot talk show hosts by 2003, when we'll also
have a smart Barbie that demands money for clothes and accessories.
I think most of his projections are pretty outrageous. Even if we
restrict ourselves to just the 2002-2003 time frame, I'd say only about
1/3 of them are coming true, and even then most of them are just demos
or small scale projects.

You can also see an earlier version of his timeline via the Wayback
Machine at http://web.archive.org/web/20010727104308/btlabs1.labs.bt.com/library/on-line/calendar/doc4.htm.
This was from 1997; if you look at the biotech/medical predictions,
the first ten were supposed to be ready by 2001 at the latest, and
I don't think any of them have come true, at least on any wide scale.
Artificial blood, artificial pancreas have been tested in clinical trials
but are still highly exxperimental. Likewise with artificial hearts.
But there are still no health history on smart cards, no thought
recognition, no home telemed links with blood test kits, etc.

So on the whole, while I think Pearson can tell an interesting story, I
am skeptical about his ability to make accurate predictions. Really it
seems that people are more interested in being entertained by futurists
than informed by them.

Hal



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:44 MST