Peter Cochrane's Hard Drive 1996
Homepage > Publication & Opinion > Hard Drive

---

Agents for better or for worse

IN OUR real world we have become conditioned to all manner of agents, writes Peter Cochrane.
Examples include insurance and banking, police and state, sellers of property, company, security and, of course, the secret agent. There are also rafts of abstract and less visible agents at work in chemical processes, medicine, biology, cookery, farming and much more.

In almost every case these agents have beneficial properties and are used for good. Occasionally, the temptation to become more efficient at killing and maiming prompts us to create defoliants and other agents of damage and death. So, we cannot assume all agents are good: they are not, but mostly because we are the controlling agent.

Recent developments in our electronic world have seen much excitement among those working on software agents. In this virtual world we can look forward to new forms of agent that will search out information, manage vast networks of technology, and even organise our daily lives. Their intelligence can be expected to progress to the point when they will be able to negotiate information technology services, software application upgrades and modifications.

In a relatively short time we may see them scheduling journeys, buying and selling commodities, thereby relieving us of a considerable amount of mundane work.

Managing complex systems, resources and workloads should be the forte of agent technologies. They are born of an era of chaos, and are ideally suited to deal with it. To date we see modest levels of agent intelligence doing increasingly complex tasks well beyond the established routes created by conventional software. Experiments are also underway to create systems that allow agents to breed, taking advantage of the best attributes created through a continual "genetic" exchange and chance mutation. Such technology should, in theory, be capable of keeping up with us as we adapt and adopt new technologies and working practices.

Now the obvious question arises: will all electronic agents be good? This seems unlikely, looking at our past history of bending good technology to malicious purpose. Someone is bound to create bad agents - spies and wreckers of systems. In fact anything we have already done in the physical world is likely to be emulated and exceeded. But is there more?

For the first time the agents we will be dealing with will have the ability to mutate and develop beyond human imagination. They may assume the proportions of medieval diseases, to disable individuals, organisations and society. Their evolutionary rate will exceed that of biological systems, and their hosts will be every computer and network on the planet.

Should we be panicking or trying to stop these developments now? Probably not. For one thing, this technology is such a power for good, and for another, it is unstoppable. Someone, somewhere will do it anyway.

But the overwhelming balance has to be good, just as with society itself; it is a necessary condition, otherwise the system would collapse. A network immune system, complete with white agents (cells) could combat almost any attack provided it is big enough to surround and smother the offending entities.

Peter Cochrane holds the Collier Chair for the Public Understanding of Science & Technology at the University of Bristol. His home page is:
http://cochrane.org.uk