What are AI Startups?

[This article was written by a great friend of mine, Sean Tao]

When someone says artificial intelligence, the first thing that pops into my mind is a hologram person that tells you what you’re doing is going to fail with a 96.346% probability in a somewhat sarcastic voice (think Jarvis or Cortana). So, when companies say that they’re in the AI field, it’s natural to feel overwhelmed. Let’s debunk fact from fiction.

Artificial intelligence, at least in this context, refers to anything that does something clever but isn’t immediately obvious. Let’s break that down—say we’re making cars. A normal car just drives in the direction in which the driver points the wheel. There’s no difference between making poor decisions and good decisions while driving, besides potentially feeling of gut-wrenching fear, angering law enforcement, and failing to arrive at the desired destination in one piece. Now, let’s say we hired a really talented intern, and the intern added a camera to somehow help keep the car in its lane. How this camera does this is quite complicated. It could use advanced machine learning algorithms, such as convolutional neural networks, for image recognition. It could live stream the video feed to other interns, and when the car crosses the line, they press a giant red button that blares an alarm at the driver. Either way, the details are not interesting to the customer, and besides, they’re too complicated to explain in a commercial or sales pitch. Here comes “artificial intelligence” to save the day—now we can explain how we do all this cool stuff without giving away company secrets or confusing customers. As icing on the cake, we also used a buzzword! Marketing is going to love this…

Artificial intelligence is not a formally defined field or particular set of algorithms, like, say, differential calculus or variations of sorting a list. However, most people that don’t study computer science don’t care about the particular algorithm that is used to solve a problem. They just care that the problem is solved, and if the problem seems like it’s something a human would care about more so than a machine, well, that’s what the term AI was invented for.

On a side note, you may have heard people use the words “machine learning” along with artificial intelligence. Machine learning is more specific than artificial intelligence. Informally, it refers to the study of algorithms that perform better with more data. Therefore, when AI companies use ML to solve some problem, what they’re saying is that they’ve identified a problem and will collect data to make better predictions to solve the problem better.

I hope that makes everything clearer!

---

P.S. I would like to give a shout out to Max! He’s a great kid, and he is an entrepreneur with his heart in the right place.

---

Sean currently studies computer science at Carnegie Mellon University and will work at an AI startup in the Bay Area next year. Besides taking classes in machine learning, he is in the process of publishing a paper in the field and is a published writer in TowardsDataScience. Previously, he founded his own company, was awarded a gold medal in the US Physics Olympiad, and has been recognized in many other national competitions, ranging from math to debate.