By Howard Rheingold
I commend Pattie Maes for not just raising but acting upon the social implications of intelligent agent technology through her support of the Open Profiling Standard -- before I raise a few questions about the wider implications of the technology. I believe that it is not just proper but imperative for the people who discover and apply knowledge to ask about what their applications might do to us. I have three questions about nomenclature, privacy, and potential systemic social effects of hypersegmentation of groups of people according to what they buy.
It is non-trivial to continue questioning the use of both words, "intelligent" and "agent" because of the anthropomorphic associations each word evokes. This question has come up before, but I am more interested in what kind of world -- and what kind of human -- those associations imply than in whether the terms are properly used. Words used to describe technology wield considerable power of political persuasion. Consider how the word "progress" is almost universally regarded as an unequivocally beneficial description of the way people or societies should move, but the question, "progress...toward what?" is almost never asked when this potent word is invoked (usually as the argument for designing, deploying, distributing new technology.)
I spent the last five years dealing with the implications of the term "virtual community," which I helped inject into popular culture. Some of the critical attacks on my use of this phrase to describe online discussions caused me to change my thinking about the assumptions and associations the word evokes. I know Jaron Lanier and I have both spent years answering questions about whether "virtual reality" is oxymoronic. It's too late to withdraw any of these phrases from popular parlance. But it is not too late to think and discuss and perhaps act upon the changes in our world and ourselves that these phrases imply.
I'd love to have the use of more sophisticated search software and interest-matching utilities. The advent of internet search engines totally transformed my daily info-hunting-and-gathering experience. I would pay for search engines that could react more precisely to my needs as it compiled and indexed a database of my past searches, and traded meta-info with the databases of others on this mailing list. But this is a sophisticated computer program for information finding and matching via the Internet, not a generalized intelligence amplifier, the way alphabets and GUIs are. Compare intelligent agents as intelligence augmentation with Doug Engelbart's ideas about "augmenting human intellect" -- still worth reading in this regard, thirty five years after it was published.
The privacy implications of intelligent agents are important now because the business model that could drive widespread diffusion of this technology derives primarily and overwhelmingly from the value of collecting detailed information about people's behavior and interests in order to sell them goods and services. The service this technology provides the people who use it -- precise and accurate matching of people with information, commodities, and other people -- is potentially a far smaller source of revenue than the service this technology provides to marketers. Intelligent agentry is more than a service for people looking for books or CDs or soul-mates or good conversations. It's a potential revolution in the way marketers can identify, understand, persuade, and manipulate customers. The longer-range implications of intelligent-agent hypersegmentation are worth thinking about now.
A market for privacy can be conducted with decent ethics, as the Open Profiling Standard attempts to enable. If people derive value from the use of information and communication technologies, and the information collected about the way they use these tools can be used to sell them goods and services, then the individual ought to be able to benefit from the value created by computer profiles of their transactions and interactions. If you want inexpensive high-speed Net access and state-of-the-art affordable intelligent agentry, perhaps you would consent to let consumer electronic companies track your web-surfing for a few months? But this is a question about the technical application of a tool for identifying people by their interests and purchases. There is another question to be asked, about where the collection and use of such information might be pushing society.
For example, consider this excerpt from an article in the November, 1997 issue of "American Demographics:"
Breaking Up America: The Dark Side of Target Marketing
by Joseph Turow
With the triumph of target marketing in the last decades of the 20th century, the United States is experiencing a major shift in balance between society-making media and segment-making media. Segment-making media are those that encourage small slices of society to talk to themselves, while society-making media have the potential to get all those segments to talk to each other. In the ideal society, segment-making media strengthen the identities of interest groups, while society-making media allow those groups to get out of themselves and talk with, argue against, and entertain one another. The balance can lead to a rich and diverse sense of overarching connectedness or understanding: what a vibrant society is about.
As with most ideals, this one has never existed. It has been far too easy for both segment-making and society-making media to lapse into stilted stereotyping of many groups rather than to act out the complex, fascinating texture that is America. But as marketers get better at targeting desirable customers in media environments designed for them, eventhe possibility of the ideal is fading.
The hypersegmentation of consumers into specialized media communities is transforming the way television is programmed, the way newspapers are "zoned," the way magazines are printed, and the way cultural events are produced and promoted. Advertisers' interest in exploiting differences among individuals is also woven into the basic assumptions about media models for the next century -- the so-called 500-channel environment. In the next century, it is likely that media formats and commercials will reflect a society so fragmented that the average person will find it impossible to know or care about more than a few of its parts.
Intelligent agents are what Seymour Papert or Sherry Turkle would call "objects to think with." We can think about what the tool might do if we use it, and how we might try to design the tool to minimize negative effects, but we also need to think about what kind of world tools like this will be used to create. Perhaps new technologies ought to have societal impact reports, not as an attempt at political regulation, but as a way of thinking systemically instead of just instrumentally. Do we know where we are going? Do we want to go there? Is there anything we can do about it?
©1998 howard rheingold, all rights reserved worldwide.