AI for the Rest of Us: What It Is, How It Works, and Why It Matters (in 5mins)
I started working at Anthropic recently, and naturally, I dove into learning about Artificial Intelligence (AI).
One of my biggest takeaways? AI experts are brilliant people who sometimes make simple things sound unnecessarily complex.
Here's the thing:
AI is reshaping our world–it's already making decisions about your loan applications, suggesting your next Netflix show, and even writing your kids' homework (whether we like it or not).
You'll either use AI tools, work alongside it, or compete with people who do, so you shouldn't need an advanced degree to understand it.
So this article is a deliberately straightforward overview of AI – not for engineers and scientists, but for the rest of us who just want to know what's actually going on.
What is AI, Really?
In its simplest form, artificial intelligence involves teaching computers to solve problems and generate ideas in ways similar to human thinking. With this definition in mind, let’s take a look at its evolution.
A Quick History
Remember playing chess against a computer in the '90s? That was actually one of the first widely available examples of AI.
The computer learned to respond to your moves based on pre-programmed rules – pretty basic by today's standards, but revolutionary at the time.
In the 1980s, we leveled up to something called Machine Learning.
Think about your email's spam filter: it learns from past examples of spam to catch new ones. That's machine learning in action – the computer learns from experience, just like we do.
Then came Deep Learning, which is like Machine Learning enhanced 10X. It's what powers your phone's facial recognition – the ability to look at a face and instantly know it's you, even in different lighting or with a new hairstyle.
Now we're in the era of Generative AI (GenAI), where things get really interesting. GenAI systems can create new things: write stories, make art, and even think ahead.
When your the map app on your phone suggests a new route because traffic is building up on your usual path home, that's Generative AI in action. It looks at real-time conditions and past patterns to come up with smart solutions on the spot.
How Does AI Actually Work?
Imagine you're teaching a child to identify shoes.
You show them lots of different shoes: sneakers, boots, sandals, dress shoes, and you explain what makes each one a shoe.
Eventually, they can spot any type of shoe on their own, even styles they've never seen before. They learn that shoes have soles, usually cover your feet, and come in pairs. Soon they can even tell which shoes are for running versus hiking or special occasions.
AI works similarly, just with more math and computers instead of pointing and explaining. Here's the basic recipe:
Start with an AI Algorithm (think of it as the brain, specifically one that's ready to learn)
Have the algorithm process massive amounts of data (like showing all those different shoes to a child) - this process is called “training”
Give the algorithm clear rules about how to process the data it’s being trained on (for example, Anthropic trains Claude based on an AI Constitution)
The result is an AI model (like Claude, ChatGPT, or Gemini) that can understand and respond to the world in sophisticated ways, based on what it has been taught.
Why AI Safety Matters
Here's why we’re seeing a wave of AI apocalypse movies:
AI models are only as good as what they're taught. When trained on biased or harmful data, they'll reflect those same biases and harms back into the world.
And if their underlying algorithms contain unclear or unethical rules, that's what they'll reflect back into the world when interacting with people. This is why there's so much discussion about AI safety and ethics.
The AI apocalypse movies, while intentionally extreme, are raising valid questions about how these systems learn, what data they should have access to, and how they’ll ultimately impact our world.
The Bottom Line
AI isn't magic – it's a powerful tool that learns from the information we give it. Understanding its basics helps us make better decisions about how to use it and what to expect from it.
You don't need to know the complex math or programming behind it, just like you don't need to understand engine mechanics to drive a car safely.
P.S.
Somewhere right now, an AI researcher is having a minor meltdown about all the technical nuances I left out. But hey, if you can now explain AI to your friends without using the phrase "neural network," I call that a win.
----
Thanks for reading. Want more insights like this? Check out more helpful articles and my weekly newsletter.