I Can’t Recommend This Highly Enough…
I referred to this in yesterday’s Infobore 12 but it deserves additional publicity. Long-term readers of Ubiwar and – possibly – my previous blog-venture KuiperCliff will know that I’m a big fan, and member, of the Long Now Foundation. This San Francisco-based collective is a purveyor of quality thinking on long-term issues that have consummate relevance to us all. Not panicky, not political, just sane.
Last August, Daniel Suarez (aka Leinad Zaurus) contributed to the Long Now’s Seminars about Long-Term Thinking with a talk on Daemon: Bot-Mediated Reality, based in part on his book Daemon. Paul Saffo – a forecaster for whom I have great regard – summarises the talk on the Long Now blog:
Forget about HAL-like robots enslaving humankind a few decades from now, the takeover is already underway. The agents of this unwelcome revolution aren’t strong AIs, but “bots”– autonomous programs that have insinuated themselves into the internet and thus into every corner of our lives. Apply for a mortgage lately? A bot determined your FICA score and thus whether you got the loan. Call 411? A bot gave you the number and connected the call. Highway-bots collect your tolls, read your license plate and report you if you have an outstanding violation.
Bots are proliferating because they are so very useful. Businesses rely on them to automate essential processes, and of course bots running on zombie computers are responsible for the tsunami of spam and malware plaguing Internet users worldwide. At current growth rates, bots will be the majority users of the Net by 2010.
We are visible to bots even when we are not at our computers. Next time you are on a downtown street, contemplate the bot-controlled video cameras watching you, or the bots tracking your cellphone and sniffing at your Bluetooth-enabled gizmos. We walk through a gauntlet of bot-controlled sensors every time we step into a public space and the sensors are proliferating.
Bots are at best narrow AI, nothing that would make a cleric remotely nervous. But they would scare the hell out of epidemiologists who understand that parasites don’t need to be smart to be dangerous. Meanwhile, the Internet and the complex of processing, storage and sensors linked to it is growing exponentially, creating a vast new ecology for bots to roam in. Bots aren’t evolving on their own — yet.
Left unchecked, bots will trap the human race because the automation they enable will make it possible for a few people to run humanity while the rest of us are unable to make decisions of any consequence. Bots are thus vectors for despotism, with the potential to create a world where only a small group of people understand how society works. In the worst case, the controls over bots disappear — for example, the only person who knows the password to a corporate bot dies– and the bots become autonomous.
We are in a Darwinian struggle with narrow AI, and so far at least the bots are winning. But there is a solution: build a new Internet hard-coded with democratic values. Start with an encrypted Darknet into which only verifiably human users can enter. Create augmented reality tools to identify bots in the physical world. Enlist the aid of a few tame bots to help forge a symbiotic relationship with narrow AI. Stir in some luck, and perhaps we can avoid the fate of the Sorcerer’s Apprentice who rashly enchants a broom to do his tedious chores and ends up terrorized by his imperfect creations. We had better succeed, for unlike the fable, there is no Master Sorcerer ready to return to break the spell and save us from our folly.
This is a subject I mentioned early on in Ubiwar‘s history and one I continue to take an interest in (when I have time). Suarez’ talk is fascinating and I really can’t do better than to suggest you click through and listen to his talk. Better still, listen and support the Long Now’s work. Some of the comments on the blog ignore the salient ecological point about bots/programs and Suarez’ perspective is, at the very least, a provocative take on the denizens of information space. Essential.

fascinating. and these bots sound enough like amoebas, dare i suggest a new notion for what’s emerging: amoebots?
Amoebots? Great name for these self-replicating beasties!
Tim, doesnt this sound just a little OTT? The bots that have data on your music purchases through amazon might be able to share data with bots working for Last FM – presumably both came from Pattie Maes’ work at MIT in the first place. At the present, that’s probably a far cry from suggesting that they can exchange data in a useful way with bots (software agents) working on totally different areas. When this data impediment is resolved – perhaps semantic web could have an interesting impact here – a more fundamental block arises. Software agents dont have agency. The concern surely is not Saffo’s vision of agents with initiative, but of malicious bots directed by humans to mine many data repositories and then act.
I didn’t think it’s OTT but I do think the whole shebang needs qualifying. What Suarez is talking about is an ecology – things evolve. Many software agents do have agency but within very limited parameters. I’m sure that will change, as people write ever more cunning code. Code is law (of the jungle)?
Interesting, as I think about it, I agree with you. (there is a comical side to this, since much of the information is trivial: Imagine a world where a cyber rogue plagues you each morning with the precise music that has been calculated to be the most opposite to your tastes… We already have that with early morning radio!)
It sure gets you out of bed though! It’s a bit like TV advertising – guaranteed to make me not buy anything…