AI, Algorithms And Agreement

Have you ever known someone who was so eager to be your friend that they agreed with everything you said? Ask them to pick a restaurant, and they tell you they’re good with whatever you choose. Tell them about your favorite movie, and they agree, replying “oh, if you liked that, you’ll LOVE these similar movies.” Share your beliefs about a controversial topic, and they echo back your opinions.

On the surface, this might sound like an ideal relationship. After all, who wouldn’t want to have a friend who aligns perfectly with your own values? In reality, friendships like that rarely work out because nobody in the relationship gets what they need to grow. Think of two hockey players practicing together in the hopes of making the NHL, one a defender and one a forward. If the defender seeks to gain approval by only making plays that cause the forward to look good, the forward will never improve as a player. When it comes time for real competition, that forward will get schooled by defenders who are intent on stealing the puck.

As we enter the new age of AI and advanced algorithms, this phenomenon is permeating the very fabric of our society, from our businesses to our families. You see, the problem with artificial intelligence (AI) and algorithms is that they are horribly, terribly, horrifically, nightmarishly helpful.

As the decision making software behind the websites, search engines and apps we love, they relentlessly seek our approval. They help us find the answers we are looking for, and if they sense that we approve (“click”), they make note of it and show us more answers that are related. Pretty soon, they only show us information that they know that we’ll like.

Imagine if you only ate your favorite foods. Steak, lobster, pasta, cookies – for me it would be ice cream shakes – whatever your number one favorite is, that’s all you would be served. It wouldn’t take long for you to develop some serious health problems. A healthy diet includes variety, including some dishes we learn to like because they are good for us (I’m looking at you, brussel sprouts).

In short, unless humans get a chance to experience new ideas, we can’t grow as people. For example, we’d still believe that the Earth is flat because we’d never hear anybody arguing that it wasn’t.

This is why so many people in the world today are shocked when they learn that somebody disagrees with them. If their social media accounts and favorite search engine are to be believed, everybody who is anybody thinks exactly like they do. Is this the truth? Nope, but thanks to automated decision makers like ChatGPT and Google, we only see and hear what satisfies us.

Once we tell AI and algorithms that our favorite color is red, they never tell us that it makes our complexion look pasty. They just suggest more and more red shirts because that’s what we buy.

It’s ironic that something so complex has such a simple idea at its core: if one thing is good, then fifty of that same thing must be fifty times better; if someone likes jazz, then they would never want to hear any kind of music that isn’t jazz.

Maybe someday these programs will stop being so helpful and start being more valuable, but it’s probably a ways down the road. I, for one, am not frightened of being subjugated by technological overlords because our current system seems pretty broken as it is. Perhaps having an AI refuse to replicate a butterscotch shake for me because it has too many carbs would be a good thing, even if it did temporarily irritate me. It would be worth it just to see more new ideas in my social media feeds, something that would be healthy for us all.