rationalskepticism.org

Spearthrower wrote:

Spearthrower wrote:
Even then, it's on a one-to-one basis, whereas technology can predict the behaviors of a much larger number of people, and to a better degree of accuracy.

I should say that this isn't even close to summarizing what current computational science can do right now, let alone consider how quickly it's developing - month by month in some respects.

Data science, machine learning, algorithms - train an ape to press a few buttons while viewing something and programs have all they need to start calculating, amending their forecasting algorithms reactively, and retesting their own models iteratively. It's vastly more potent in so many respects.

It's not just a larger number of people that are reached (it's sort of not fair to compare a computer programs potential points of contact to just one person having encounters with real people in meat space, but human ToM doesn't network and thus is necessarily individual, unshared and independent - islands of mind theories) but the quality of prediction is also vastly more detailed.

If any given human can look at some other human and generally be right in mind-theorizing that person is upset, or determined, or fraught - it's not actually a predictor of what that other human is going to do next; there's no detailed behavioral prediction, more of an emotional state.

Meanwhile, a site like Facebook can - without human oversight in the slightest - correlate various data points drawn from your past behavior in tandem with your current use of a new Canon camera and make predictions on a suite of your future activities - namely purchases - that you are likely to make. That's a prediction that entails seeing much further into the future than any ToM, with far more complex behavior and emotions over a longer period of time accurately predicted.

And we've seen that this isn't just restricted to predicting purchasing behavior, but what this can mean when it comes to engineering opinion with, for example, Cambridge Analytica.

It's kind of quaint to believe that humans just know humans best; there's something earthy there. But I think the real differences between people just figure large to us rather than being particularly distinctive - zebras and their unique stripes. In reality, computers strip away a lot of the emotional baggage that can hamper our ToM predictions and just quietly store away data points until they become statistically meaningful, and at that point, it's no longer a contest.