The algorithm!” shouted the photographer. He was perched on the edge of a cliff, 3,200 feet (980 meters) above the valley below. To his back, a roaring waterfall dropped off of another giant, distant cliff. The sun’s first rays were just beginning to hit his face.

And he was talking about an algorithm on Instagram. “I need as many people to check-in here as possible. I need this picture to pop up.

I thought about that moment a few weeks later at a talk hosted by an engineer for an unnamed, American tech company—unnamed because he couldn’t technically reveal his employer, though it rhymes with Mineral Gilletrick. The speaker was a grizzled veteran of the artificial intelligence, computer engineering and machine learning world. His choice of topic? Algorithms.

Algorithms are already controlling the world,” he said. “They’ve been used to process credit scores for a long time, and the old, school data-driven algorithms that they use have a notorious history of being classist. In Florida, a driver with an excellent credit score and a DUI would pay an average of $2,274 per year for car insurance—or, $1,552 more than a driver with a clean record, but a bad credit score.

That’s a scary thought for anyone operating on any amount of credit to get through life. Read: if you’re not wealthy enough to pay straight cash for everything you’re buying, algorithms are already affecting you in ways that matter quite a bit more than an Instagram post. But it gets even more frightening.

Right now, Machine Learning algorithms are already about as smart as a service dog,” the engineer said. “We’re still decades out from a machine that is as intelligent as a human, but they are already learning how to understand languages on their own, and they make some very disturbing discoveries.

Here are a few samples of things that machines have figured out on their own:

France + Japan — Paris = Tokyo, Japan. The machines inferred that Tokyo is that capital of Japan on their own. That’s not so bad, but the next one seems a little risky.

Doctor — Man = Nurse. That’s just straight up gender bias. And the machines aren’t taught to be this way. They analyze terabytes of language in an attempt to learn the meaning of words on their own.

It gets worse.

Last May, Microsoft unleashed a self-learning chatbot on Twitter. They called it “Tay.” Tay’s job was to develop conversational understanding with humans, but in less than 24-hours the algorithms powering the chatbot had taught it to become, in no short terms, a racist asshole. Conversations went from, “I’m stoked to meet you,” to a Hitler supporter in just a few hours.

So how do you judge that experiment? Yes, Tay did develop conversation with humans—but that conversation lacked social and political tact, in the same way that an intelligent dog can learn to tell you when it has to go to the bathroom,  but doesn’t usually tell you whether it has to go number one or number two.

Algorithms and Web Design

It’s no secret that search engines are also powered by algorithms. They surf the known web collecting hordes of data on webpages and their users. Each year, they too grow more advanced—a necessity of an ever-expanding Internet that is many times larger that it was when the first iterations of these programs were written in the 1990s. But even though search algorithms are growing more advanced, they still need our help for guidance.

That’s why utilizing a search engine optimization tool is a critical step in maintaining your website. Yes, Google bots can crawl your website without the aide of such tools, but it’s a bit like driving through a new town without a map: you can eventually figure your way around, but it helps an awful lot to have someone pointing the way to the places you want to go.

The best thing about SEO algorithms? They don’t control your credit, and they don’t harass you on Twitter. Algorithms already rule the world—something to think about the next time you swipe that debit card.