So is IBM planning to develop “SkyNet”? As much as many think that major corporations will be the downfall of society, it wont be because the Big Blue Machine suddenly felt that world domination was a good business venture (Microsoft has that market covered anyway).
Scientific American put up a story concerning autonomous systems, and what the next step for servers will be. IBM is in the second phase of their <a href=â€http://www-1.ibm.com/servers/eserver/introducing/eliza/” target=”_blank”>eLiza project (also a popular conversation engine, not related) which will manage workloads across numerous systems. The software, currently dubbed Enterprise Workload Manager, acts as an autonomous agent that will foresee and correct problems without expensive administration.
This would appear to be the beginning of a <a href=â€http://news.com.com/2100-1001-897080.html?tag=fd_top” target=”_blank”>trend in the IT industry. A self-governing system has many merits, especially that you will no longer have to hire that guy to live in the basement and watch the servers. Currently, when a server goes down the consequences can cost millions. And with the relative “popularity†of hacking, businesses spend a lot of money on security and administration. A system that can detect these attacks and make the appropriate actions does indeed sound like a viable (and cheaper) alternative. However the far-reaching goal might be in question.
IBM sees an autonomous system as a simulation of the human nervous system. It’s a system that has instinctive responses to interactive stimulus, and self-managing control functionality. Or more simply, it’s a system that makes life easy and thinks for itself. But is this AI? Is this software going to take over the company intranet and make us do stupid tricks for our e-mail? Not quite.
According to IBM’s Autonomic Computing Manifesto, they make many references to a computer, or a system, having human like qualities. Before we get philisophical about “can a computer be human?”, the human system is only a model for what IBM plans to accomplish. Among these are a ‘sense’ of self (hardware config, network neighbors, etc), to be able to “change it’s mind” (reconfigure, recompile, reload, etc), and be able to protect itself (security preferences). And apparently:
3. An autonomic computing system NEVER settles for the status quo — it always looks for ways to optimize it’s workings.
I got a chuckle from that one. I suppose for a computer this is great but not so much true for people. It could be true if by ‘optimize’ they mean ‘do less work’. Maybe it’s just me and my friends, but whatever pays the bills is enough for us.
In phase one of project eLiza, software was configured for a single server to maintain it’s own functions. Phase two will allow a network of computers to weed out the malfunctioning computer, take it off line, fix it, and bring it back up. So like humans, breaking a finger wont “crash your systemâ€. And once the finger is healed, you can use it again in whatever dark and dirty ways you see fit. While the system still operates on cause and effect, eventually the plan is to have it see problems ahead of time. Much like we move a hand if a door is about to shut on it. If the Workload Manager sees an increase in hacking, it can beef up security.
Instead of HAL9000, IBM hopes to create the ship’s computer on the Enterprise, making information very accessible, and handling all the tasks on an invisible level. And I don’t think there was ever an episode where the computer took over the ship (not without intervention anyway). They plan to release a version of the Enterprise Workload Manager sometime later this year. It’s obvious this is written for higher workload systems. But, with a certain adaptation, I’m sure you could start asking your computer for cake recipes… and have it know to preheat the oven.
Comments on this entry are closed.