Here’s why artificial intelligence should probably terrify you

I was talking to a friend on Sunday, just about the state of the world — as one is wanton to do on a Sunday — and we started talking about the generally sci-fi idea that evolution won’t end with humans and, instead, sentient robots will be the next phase of that process. Eventually the robots will start to view us as spam, and bam, bye-bye humanity. Drastic? Yes. Incomprehensible 20 years ago? Yes. Incomprehensible now? Mostly, but not entirely. 

Artificial intelligence probably shouldn’t broadly scare you, but in some ways, it should positively terrify you. Here’s one such way. 

Start with this study: “Computer-based personality judgements are more accurate than those made by humans.”

Here’s a summary of that study, which is from Stanford University and the University of Cambridge.

Let’s talk about some of what happened here, shall we?

The Methodology and Process

Here’s what they did:

In the new study, researchers collected personality self-ratings of 86,220 volunteers using a standard, 100-item long personality questionnaire. Human judges, including Facebook friends and family members, expressed their judgment of a subject’s personality using a 10-item questionnaire. Computer-based personality judgments, based on their Facebook likes, were obtained for the participants.

Here’s What Was Learned

Essentially, the computers — operating from Facebook likes, basically — won. By analyzing just 10 likes, a computer could predict someone’s personality better than their real-life work colleague; it took about 300 likes to be better than a spouse, though.

Here’s a little more context on what “likes” means in this study:

The computer predictions were based on which articles, videos, artists and other items the person had liked on Facebook. The idea was to see how closely a computer prediction could match the subject’s own scores on the five most basic personality dimensions: openness, conscientiousness, extraversion, agreeableness and neuroticism.

Here’s The Broader Context Of Why And What All This Means

Logically, think about it this way:

  • A computer can store more information.
  • It can access more information (typically using algorithms).
  • (Remember IBM’s Watson?)
  • Humans tend to give a lot of weight to 1-2 examples when analyzing someone else’s personality or decision-making, which leads to non-rational thought.
  • Computers don’treally do that.

Here’s The Requisite “OMFG, THE MACHINES ARE COMING” Quote

“The ability to accurately assess psychological traits and states, using digital footprints of behavior, occupies an important milestone on the path toward more social human-computer interactions,” Wu said.

The Role of Subtle Cognition

Researchers all up and down this study are careful to mention that “some aspects” of personality trait-ID’ing, such as those that require “subtle cognition,” are better left to humans. Makes sense. This study analyzed Facebook likes, after all. It doesn’t analyze body language or tone of voice or past relationship between two people or anything else; at this point, it doesn’t appear machines/robots/future overlords are en route to doing that perfectly. (Although I’m sure there are people somewhere who are getting close on those fronts too, or will be in the coming years.)

Data-Driven Decisions

I’m all for data-driven decisions influencing people’s lives — for example, I feel like programs such as this one that can analyze Facebook likes should be turned directly towards health care — but when an artificial being could potentially know you better than your spouse (which is the click-bait headline version of this story, essentially), that’s a little bit terrifying, right? That’s a little bit like “as we run around being busy, the robots are gathering more data about us than we have the potential to gather about each other…” and then, bam, bye-bye humanity.

Again, a bit much? Yes. But in the last two decades, has this become a vague possibility as opposed to a total sci-fi wonderland? Yes as well.


Ted Bauer

Reply If You'd Like