How could an AI surpass human abilities? Let us count the ways . . . Speed. Our axons carry signals at seventy-five meters per second or slower. A machine can pass signals along about four million times more quickly. Serial depth. The human brain can’t rapidly perform any computation that requires more than one hundred sequential steps; thus, it relies on massively parallel computation.2 More is possible when both parallel and deep serial computations can be performed. Computational resources. The brain’s size and neuron count are constrained by skull size, metabolism, and other factors. AIs could be built on the scale of buildings or cities or larger. When we can make circuits no smaller, we can just add more of them. Rationality. As we explored earlier, human brains do nothing like optimal belief formation or goal achievement. Machines can be built from the ground up using (computable approximations of) optimal Bayesian decision networks, and indeed this is already a leading paradigm in artificial agent design. Introspective access/editability. We humans have almost no introspective access to our cognitive algorithms, and cannot easily edit and improve them. Machines can already do this (see EURISKO and metaheuristics). A limited hack like the method of loci greatly improves human memory; machines can do this kind of thing in spades.
50:17
Extortion is the practice of obtaining benefit (e.g., money, goods, or regular payments) from an individual or group through coercion, usually by threatening them with future psychological or physical harm. In most jurisdictions it is likely to constitute a criminal offence. Unlike extortion, robbery is the obtaining of goods using immediate personal violence, or the immediate threat of violence, usually in a one-off situation.
To sum up the principle briefly: your brain builds you up a self-image. You are the kind of person who says, and does... whatever it is your brain remembers you saying and doing. So if you say you believe X... especially if no one’s holding a gun to your head, and it looks superficially as though you endorsed X “by choice”... you’re liable to “go on” believing X afterwards. Even if you said X because you were lying, or because a salesperson tricked you into it, or because your neurons and the wind just happened to push in that direction at that moment.
All familiar phenomena, right? You probably already discount other peoples’ views of their friends, and you probably already know that other people mostly stay stuck in their own bad initial ideas. But if you’re like me, you might not have looked carefully into the mechanisms behind these phenomena.
The unsettling part comes next; Freedman and Fraser wanted to know how apparently unrelated the consistency prompt could be. So, with a third group of homeowners, they had a “volunteer” for an ostensibly unrelated non-profit ask the homeowners to sign a petition to “keep America beautiful”. The petition was innocuous enough that nearly everyone signed it. And two weeks later, when the original guy came by with the big, ugly signs, nearly half of the homeowners said yes -- a significant boost above the 19% baseline rate. Notice that the “keep America beautiful” petition that prompted these effects was: (a) a tiny and un-memorable choice; (b) on an apparently unrelated issue (“keeping America beautiful” vs. “driving safely”); and (c) two weeks before the second “volunteer”’s sign request (so we are observing medium-term attitude change from a single, brief interaction).
1a. Avoid petitions, and other socially prompted or incentivized speech. Cialdini takes this route, in part. He writes: “[The Freedman and Fraser study] scares me enough that I am rarely willing to sign a petition anymore, even for a position I support. Such an action has the potential to influence not only my future behavior but also my self-image in ways I may not want.”