What is generative AI, really? I get that question all the time (no surprise!). But it’s not always an easy one to answer. For the past 8 years, AI is what I’ve lived, breathed, and thought about. How do you describe the house you’ve been living in without deep diving every detail? I mean, I would love to go on about Large Language Models, proprietary layers, and neural networks, but that doesn’t answer the question, does it?
So, let me step back now and give you the straight story, no frills, no tech speak, and, most importantly, no doublespeak (you can tell when an AI site is all hype by the number of cliches it has on its website. Seriously, go check out a few!).
OK, then, what is generative AI really?
Generative AI is a complex computer program that has been trained to be super creative. Imagine you're teaching it to be an artist, but instead of using paints and canvases, it uses data and math.
Here's how it works:
1. Learning from Examples: To start, you show the AI lots and lots of examples of things you want it to create, like stories, pictures, or music. It looks at these examples and learns from them.
2. Finding Patterns: The AI starts to see patterns and rules in the examples. It figures out how sentences are structured, how colors and shapes go together, or how music notes create melodies.
3. Creating New Stuff: Once it's learned from all these examples, it can use those patterns and rules to make new things on its own. So, you can ask it to write a story, and it will come up with something based on what it's learned.
4. Getting Better: The more examples it sees and learns from, the better it gets at creating stuff. It's like a robot artist that keeps improving with each new painting.
Now that part about teaching and learning is super important. Generative AI learns not only from all the examples it’s given (in general, kind of techy terms, the “data” it is exposed to), but also from the humans it interacts with. The questions you ask it and the feedback you give it (“Nope, that’s more Picasso than Monet. Give me more Monet.”) are just as important. Always remember, AI is only as smart as the data it “sees” and the humans that “teach” it.
Generative AI is already making a difference out there in the world. It helps scientists simulate experiments, designers come up with new ideas, and writers with inspiration. It's also used in businesses for things like chatbots and customer service.
And it’s what lies underneath our cyber assistants that have been trained (or can be trained) in the basics of all sorts of tasks normally done by humans (e.g. filling out forms, writing code, doing quality checks, providing reports). They truly are “assistants” who can be trained by you to do the tasks you’d rather not do (or that take lots of time), so you yourself can be more creative, productive, and, well, happy.
Check back here for more “doublespeak” free info about AI, generative AI, and what it means for the future!