The Evolution of AI in 10 Fast Years

By Katherine McKean, Junior and President of my high school AI Club

A decade ago, artificial intelligence was mostly hanging out in the background—filtering spam, autocorrecting texts into weird poetry, and occasionally mistaking a dog for a bagel. Fast-forward to now, and it’s front and center. It writes, draws, animates, tutors, translates, and sometimes over-apologizes for things it didn’t do. The speed of change has been dizzying, and in true Bay Area fashion, very startup-y: fast, unexpected, and occasionally fueled by oat milk lattes.

from rule-following to pattern-learning

Early AI was rule-based. That meant if-this-then-that logic—great for programming microwaves, less great for anything human. But around 2015, a shift happened. Thanks to advances in deep learning, AI models began training on huge amounts of data and learning to recognize patterns on their own.

Instead of programming an AI to know what a stop sign is, you could show it thousands of examples and let it figure it out. In practice, that sometimes meant it also thought octagons in red T-shirts were dangerous intersections, but hey, progress isn’t always clean.

2018 and the arrival of transformers

In 2018, researchers introduced transformer models—giant leap. These allowed AI to understand context and relationships in language better than ever. It meant AI could move beyond finishing your sentence to finishing your homework (which you probably shouldn’t let it do).

This paved the way for language models like GPT and BERT to do things like explain metaphors, draft essays, and write song lyrics that sound oddly like your English teacher after a double espresso.

2020–2022: when fake got real

As language models took off, generative image models weren’t far behind. GANs (generative adversarial networks) made it possible to create incredibly realistic images of people, animals, or cats that look like croissants. Deepfakes moved from meme territory to mainstream, and tools like DALL·E started turning words into pictures faster than art students could find their sketchbooks.

In our high school AI club, someone typed: “An alpaca on a skateboard drinking boba tea.” The result? A new club mascot. We named him Alpacino.

2023: chatbots go viral

When ChatGPT launched, it felt like everyone had suddenly gotten a new best friend—who happened to have Wikipedia-level recall and the tone of your favorite camp counselor. By the end of 2023, students were using it for everything from writing outlines to figuring out what “stoichiometry” means ten minutes before chemistry class.

Other bots joined the scene. Claude showed up like the thoughtful cousin who writes poetry and takes gap years. Gemini dropped in with vibes that shifted between cool and “wait, why are you linking a 2017 Reddit thread?”

2024–2025: the year of video AI

Enter Sora.

In 2024, OpenAI launched Sora, an AI that could create short video clips from text descriptions. You could type “a penguin in a detective hat solving a mystery in Venice,” and it would spit out something surprisingly close. Then came Sora2—more stable, more realistic, and fewer glitchy fingers.

In our club, we used Sora2 for mock trailers and creative projects. One student recreated their family vacation using only text prompts and exaggerated memory. Another made a mini documentary narrated by a squirrel. Was it realistic? Not really. Did it have emotional depth? Surprisingly, yes.

student opinions: divided but curious

Most high schoolers I know are somewhere between “this is the coolest thing ever” and “I miss when my calculator just did math.” Some use AI to brainstorm. Some use it to procrastinate by asking it increasingly weirder questions (“What would a Shakespearean sonnet about bubble tea sound like?” Answer: amazing).

There’s also a growing interest in learning how it works. Students who weren’t into coding are now poking around with prompt engineering or joining AI ethics discussions. If nothing else, AI has made being curious cool again. Well, cool-ish.

what sora2 means to teens

Sora2 hit a nerve—in a good way. It blurred the line between storytelling and animation. Students who didn’t think of themselves as “creative” were suddenly making short films. Sure, some had floating eyeballs and doors that led to nowhere, but they were visual ideas brought to life in minutes.

One sophomore turned their fantasy novel into a trailer. Another created a “day in the life” of their dog from the dog’s perspective. A junior made a fake documentary about aliens hiding in the San Mateo public library. It didn’t win awards, but it did go viral on our school’s private Discord.

it’s not magic, but it’s weirdly useful

At the end of the day, most of us don’t think AI is going to take over the world. We do think it might accidentally schedule our dentist appointments for midnight. It’s a tool—one that gets more interesting (and sometimes chaotic) the more you use it.

In class, it’s helped with tutoring, language practice, and project ideas. Outside of class, it’s helped write raps about mitochondria and generate memes featuring our principal as a Jedi. We haven’t figured out the full ethics of that last one, but it got a laugh during morning announcements.

what’s next? we don’t know either

There’s no clear answer about where AI is heading next. But students are watching. We’re testing new tools, figuring out how to use them responsibly, and asking the big questions. Like: can an AI really understand what it means to be human? Or just what it means to want a burrito at 2 a.m.?

The last ten years have been fast. The next ten will probably be faster. Whether AI is helping you animate your dreams, pass your math test, or argue with your dad about who invented the toaster (spoiler: it wasn’t you), one thing’s certain—this is just the beginning.

Want to bring AI to your school the way we did? Check out this step-by-step guide on How to Launch a High School AI Club in 10 Easy Steps.