subject: Dear AI Software Designer - How Would You Design a Nationwide ATC System? [print this page] Dear AI Software Designer - How Would You Design a Nationwide ATC System?
Most of the details are all worked out, but that does not prevent myself or other think tankers such as Troy Laclaire from considering the best ways to design such a system to handle the intensity which awaits, namely a 365 day 24 hours a day system of perfection. Besides that, not everyone nation can afford the $45 Billion price tag.
In fact, even our own Congress in the US is debating putting the NextGen ATC on the chopping block to shore up the bloated budget of the blob of bureaucracy. Troy and I discussed the need to have a mostly artificial intelligent software system running things, with supervisory roles for humans to keep it running smoothly just in case. Troy started with a few relevant suggestions, when I challenged him and asked; Dear Software Designer - How Would You Design a Nationwide ATC System? He replied with this statement;
"First, this is a multi level problem/solution. The AI computer would continuously be learning, you have to have data storage, both long term and short term, so you can limit the system from being able to learn beyond a certain point by limiting its storage capabilities." Yes, I thought, because this is where I see the big difference between organic thinking, and computer thinking. Humans screw up with too much data, and become paralyzed to decision making and usually make poor decisions with too much information, computers seem to do better with %'s and proximities, when they get more data.
You see, Troy believes that; the only problem with computer systems, and the reason they can handle such large amounts of date, is linear thinking. However, if you give a computer garbage, and even it will fail.
Yes, the old garbage in, garbage out motif which computer software designers must consider well in advance. Then Troy makes another interesting comment and that is; the software designer needs design the system so that it only learns from fields in which it is meant to be used for - for example, there is no reason for a ATC AI to learn about biology on its own. Next, the software designer needs to make sure that it's going to follow your orders and not come up with its own.
Troy believes that; this could be accomplished somewhat through something similar to the "3 laws of robotics" as Asimov describes, however I find it funny that in a majority of his robot series books deal with the robots getting around these laws. Something similar but more thought out would be necessary.
Then of course you want the system to be able to be shut down in case it ever does decide to go on its own, so this could be accomplished through multiple means, such as limiting power capabilities, having hardware separation systems, or even having the system require regular input by a human to keep running. Something as simple as a countdown clock that the AI itself cannot access would work here.
Further, Troy reminds us that the designer of the software would need to limit the system's communications between non necessary systems. there would be no reason for example for a ATC AI to need to communicate directly which an AI that runs a hospital system, except for emergency services. This way if you do have an AI become "self aware", you limit its capabilities to pass this on to other systems.
Well, Troy makes some great points and I ask you Mr. AI Software Designer, how would you do it differently? Because all this kind of goes against the whole "Road Ahead" digital nervous system of the corporation or the future of the eRepublic - AI Government. But I totally in agreement with Troy to proceed with caution along with proper checks and balances, otherwise your perfect AI system could turn into a digital Hitler dictator. So, this is all good wisdom and philosophy.
Finally, Troy admits; "Overall, it's going to be one hell of a balancing act, to allow the systems to learn and function as you want them to, while keeping them outside of areas you do not want them to learn." So, I hope you will please consider all this and think on it.
Lance Winslow is the Founder of the Online Think Tank, a diverse group of achievers, experts, innovators, entrepreneurs, thinkers, futurists, academics, dreamers, leaders, and general all around brilliant minds. Lance Winslow hopes you've enjoyed today's discussion and topic. http://www.WorldThinkTank.net - Have an important subject to discuss, contact Lance Winslow
