I heard about a tool called ChatGPT, played with it a bit, and was mystified by its magic-like qualities. There’s a famous quote from Arthur C. Clarke that I remember hearing in a college lecture, “Any sufficiently advanced technology is indistinguishable from magic.” Generative AI felt like this to me—and, frankly, to most of the world.
CLM Marketing and Advertising hosted a speaker series in June 2023. By this time, I had dabbled with the tool but nothing extensive. The speaker was Dr. John Sviolka, a professor from Harvard, and a long-time educator and researcher in Data Science and Artificial Intelligence. Eager to learn more about this magic from a distinguished academic, I couldn’t wait. Thus began my journey to learn about and embrace the momentous technology that is Generative AI, and to help demystify it for my colleagues and clients.
AI: What is it? And, why you shouldn’t be afraid. AI is just a set of instructions or an algorithm that has been shown a lot of examples so it can make predictions on a massive amount of data to build a model.
A model: is the program that uses these learned patterns to make predictions, all based on the billions of parameters on which it’s been trained.
Machine learning: A subset of AI that enables computers to learn on their own. It uses algorithms to guide a model to perform a task. It can be trained through a few different techniques—supervised, unsupervised, and reinforcement learning.
Deep learning: A type of machine learning where the algorithm is written in such a way as to understand a consumer’s behaviors and habits.
Natural Language Processing: A field of computer science and a subset of AI that allows a computer to learn, “understand” and clarify human language. Most of us know about this via the frustration of having our message changed while we’re in the act of texting, often to something we didn’t intend. It’s gotten better over the years, but machines can still get it wrong.
I would bet that most, if not all, of you reading this have been influenced by AI thousands of times in the last 10 years. Netflix is one of the most famous for this. It’s a Sunday afternoon, and you’ve just binged your favorite show, then Netflix gives you a few options and all of them seem to your liking. Next thing you know, you’re in front of the television, for better or worse, for a new show that you plan to finish next Sunday.
Another example of AI impacting our daily lives is when your favorite package arrives on your doorstep just hours after you ordered it. Amazon’s fulfillment centers and online storefronts are heavily powered by AI, and it’s not just limited to product recommendations. For instance, Amazon knows that when Sarah views an item three times, she usually buys it. To capitalize on this, Amazon ships the item in advance, betting that Sarah will buy even more frequently if she sees that her package can arrive in just four hours. It’s an impressive service—fascinating, a bit unsettling, but undeniably convenient.
Enter stage left Generative AI. Generative AI (Gen AI) systems can make predictions based on existing information they’ve been trained on; the key difference for Gen AI is using patterns in data to generate new content. This can be extrapolated further if you were to prompt the same thing twice; you won’t get the same answer. Generative AI is a probabilistic model, not deterministic. Deterministic means the outcome is always the same from a starting point or prompt. Probabilistic means the outcomes can vary, and you can only guess the chances of a result from the starting point or prompt. Prompts and the skill of prompting constitute a vast topic; I will talk about prompts more in a later blog.
So why should we be interested in Gen AI? For starters, AI is the dumbest it will ever be the day you read this. This technology is here to stay, and it will only get better at what it does, remember it is learning. I’ve heard stories of people scared to lose their jobs, which is a fair statement. This is magic, and if you don’t know how it works and don’t start using it, you will likely be left behind. I’m of the opinion that it will allow more time for innovation at work and in our personal lives, but it will also be a brainstorming buddy and the grammar police. There’s another great quote I like from Nicholas Boucher, a LinkedIn Finance personality: AI is poised to, “bring back humans to human tasks, letting robots do robot tasks.” It will not remove you from the process but allow time for you to keep overflowing with new ideas of what is possible.
This sounds great, but it’s not perfect, at least not today. Gen AI can hallucinate (make things up and be declarative that something is real when it is not) and sometimes the output is awful; it may say “delve” a thousand times in your rough draft copy. I tell people to think of Gen AI as an intern. It’s newer than you at your company. It’s smarter than you—these models are trained on billions of inputs of data—and it’s incredibly quick. But all this is to say, it sometimes gets things wrong. Experience has value, our experience in tandem with this tool has value, so use it!
I’ll be writing several blogs in this series to help explain techniques so we can get more out of this tool. So, we can begin to understand this magic and appreciate it. In the end, Gen AI is just a new technology much like the internet of the ’90s. I hope at the end of this blog series, you’ll embrace Gen AI and we can have a conversation together about it.