Thinking Machines Need To Be Dumb - InformationWeek

InformationWeek is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them.Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

IoT
IoT
Infrastructure // PC & Servers
Commentary
4/14/2009
07:40 AM
50%
50%

Thinking Machines Need To Be Dumb

Machines have started making independent scientific discoveries, according to two reports made public a few weeks ago, perhaps heralding the start of true artificial intelligence. I think machines need to be dumb before they can ever be smart.

Machines have started making independent scientific discoveries, according to two reports made public a few weeks ago, perhaps heralding the start of true artificial intelligence. I think machines need to be dumb before they can ever be smart.The reports are both pretty cool: in one lab, a robot can reason after conducting an experiment, and design the next one accordingly. Already, it has made a discovery based on its own hypothesis, without the guidance of a human interloper. In the second instance, a computer can not only crunch the numbers behind the movement of a pendulum, but has deciphered in them Newton's laws of motion.

This is amazing and scary stuff, bringing to mind images of HAL 9000, Nexus 6, and Cyberdyne Systems. But I still think the real breakthrough that differentiates intelligence (Turing's definition of self-awareness) with consciousness (my theory; see below) will be stupidity. Machines need to get dumb.

Insight and intuition are hallmarks of consciousness; synthesizing apparently random data with an endless array of possible criteria is a distinctly human trait, and it's the cauldron from which both brilliance and stupidity arise. So while it's cool that a machine might find the right answers to questions, it isn't enough, in my opinion. They need to be able to reach the wrong ones...for all the right reasons.

Machines need to figure out how to skip replicating Newton, and give us Ptolemy's armillary. Discover Aristotle's spontaneous generation instead of Mendel's genetics, or substitute Velikovsky's planetary near-misses for Kepler's orderly motion. It's no surprise that a computer program could discern fact from fiction, but show me one that comes up with something as complex -- and gloriously wrong -- as the hermetic medicine of Paracelsus, and I'll believe that machines are on the way toward making the insights that come only from consciousness.

Come to think of it, the AI devices I mentioned earlier were all dumb in their own ways...the plots of each movie depended on it. So maybe it isn't such a leap of faith to imagine stupidity emerging from ever-smarter intelligence?

After all, it's how we humans manage it.

Jonathan Salem Baskin writes the Dim Bulb blog and is the author of Branding Only Works On Cattle.

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Print  | 
More Insights
Slideshows
11 Things IT Professionals Wish They Knew Earlier in Their Careers
Lisa Morgan, Freelance Writer,  4/6/2021
News
Time to Shift Your Job Search Out of Neutral
Jessica Davis, Senior Editor, Enterprise Apps,  3/31/2021
Commentary
Does Identity Hinder Hybrid-Cloud and Multi-Cloud Adoption?
Joao-Pierre S. Ruth, Senior Writer,  4/1/2021
White Papers
Register for InformationWeek Newsletters
Video
Current Issue
Successful Strategies for Digital Transformation
Download this report to learn about the latest technologies and best practices or ensuring a successful transition from outdated business transformation tactics.
Slideshows
Flash Poll