You’ve Been Using AI All Along – It Isn’t What You Think

But when I saw what GPT-3.5 could do, I realized this was the future. I realized I’d missed the biggest piece of the puzzle; AI wasn’t what everyone was claiming. However, I wasn’t ready to reveal my take to the world. Also to complete the puzzle, I knew I had to embrace AI, or I’d quickly fall far behind the productivity expectations of our clients. 

So, I set off to work and began researching what the tool could do, its weaknesses and strengths, the flawed logic in word choice and extreme overuse of cliches, and how with some better input and direction, not unlike choosing specific styling functions in tone analytics tools, it could create content that mirrored my own, sweat- and research-laden content. 

Over time, I realized that the future of content creation was evolving, and when GPT-3.5 first gained the ability to start adding formatting, I realized this tool actually was based on some things I’d come to take for granted over the years. I realized that many people, including you, have been using AI for years. You just didn’t know you were using AI. 

Before the endless GPT cheat sheets and infographics, and well before the results-promising, widely reshared prompt techniques of thousands of people on LinkedIn…the functions were already alive and well. These people who were claiming to have the inside scoop seemed to have all the answers, but they hadn’t been on the AI bandwagon for long and seemed to know what was best. Many were marketers, and there’s a reason for that. 

Marketers were among the biggest group of people using AI every day and waking moment before it was trendy and before AI became mainstream. There are people–maybe even you, reading this very piece of content–who in the late early 90s and most definitely after 2000 were using AI daily. You just didn’t know it. 

A Brief History of AI in Modernity and the First AI Winter

Rewind the clock to 1955. John McCarthy, Marvin Minsky, Nathaniel Rochester, and Claude Shannon effectively make the first use of the term, artificial intelligence, in a proposal for a 2-month, 10-man study. 

But if you go a bit farther into the past, the first tangible, do-it-yourself reference to a computer-controlled program was made by Claude Shannon, providing theoretical instruction on how advancing technology could think in a specific setting in response to input. But one of the first to create a program of such specificity, giving a computer to act on its own volition was in 1952, when Arthur Samuel developed a computer checkers-playing game. 

But 1952 is hardly what we’d describe as modernity. And computer checkers is hardly AI…right?

Wrong. So let’s fast forward to when man set foot upon the moon-1969. 

Arthur Bryson and Yu-Chi Ho published a paper on something called, backpropagation. Considering the first deployment of the latest tech craze-robotic machines– only a few years earlier, the pair described a dynamic system optimization method that would learn from a multi-layer artificial neural network. 

But much like Shannon, the idea was present, but technology had yet to progress enough to test the theory. All the computing power in the world–maybe–wouldn’t have been sufficient for such a test. 

Enter the 70s

Money poured into the up-and-coming Silicon Valley where everyone promised big returns and almost utopian futures. Even the government wanted a piece of the action, so the money flowed like Niagara. But then, the river dried up. 

See, there was a fatigue in the air. Movies like War Games and Science Fiction brought fear to the majority of civilization. Plus, even if we had those programming abilities, the actual technology and infrastructure on which to test the theories were far behind. So, the world needed a new shiny something that could draw interest in computer science while waiting on the physical equipment to catch up. And it all started with a cookie.

Mrs. Fields…I can practically taste them with my mind…opens a bakery in 1977. She begins expansion down the line and finds herself with a problem. She didn’t want the secret recipe to fall into the hands of competitors, meaning she couldn’t share ingredient amounts with other stores. She ended up measuring and prepacking ingredients to send to each subsequent location. It’s a labor of love, and it’s worth the drive. 

But with a growth trajectory that would hit 350 stores in 1987, no one could drive to an increasing number of locations every day or even every week to deliver for reordering lists. But Mrs. Fields had a secret weapon–Mr. Fields. 

Mr. Fields had began developing a means of communicating data on individual computers between two places, and in time, he built Fields Software Group, the first true business intelligence tool on the market that could predict demand and enable efficiency gains at scale. 

Ah, the intranet begins

At the same time, a similar system was growing in government development–meant to provide a way of sharing actual information, files, and photos, through a channel similar to telephone signals. 

Mrs. Fields grows, and Mr. Fields does as well. Combining their store-to-store data-sharing system with the business growth led to Fields Software Group–the first multi-stage learning system or business intelligence system on the market. And Burger King acquired Fields Software Group in 1990.

The Internet and the Dot-Com Bubble

The Internet and America Online became iconic names, and suddenly we have this doc-com bubble. It shuffles, and business intelligence takes a ride but latches on the techy-lingo of the age–business intelligence or analysis processes. The money returns steadily, until the late 1990s, when the bubble burst. 

Everyone hates the terms associated with the dot-com bubble, so they begin shortening the lingo again. They call it analytics, and it’s about to get a boost with a Garage-founded search engine duo and a company they’d one day acquire and what would become the basis for digital marketers. 

Google rolled out in 1998, and in 2005, Google acquired a software development company, Urchin Analytics that tracks page activations of visitors, essentially tracking web server logs and identifying returning visitors. It became the original website analytics tool, and with the acquisition, Google Analytics is born. Urchin’s history is still alive and well today in the form of UTM tags, and in fact, the “u,” which most assume is meant as a universal tracking module, is actually the shorthand of the Urchin Tracking Module. So, it’s still here. 

Everyone, including investors, latch onto the upstart that’s breaking record after record. Google Analytics quickly becomes the gold standard in advanced technology around marketing and business by extension. Computer equipment has reached sufficient levels, and theories are proven. 

But returns were challenging. The bubble had just burst, so the money started to dry. Various companies started large language model development a decade or so earlier, and now was the time to ride the wave. Business intelligence got much of the notoriety for a while, but then, the AI shift began again. OpenAI released GPT in 2018, and everyone is in dev mode. In 2023, GPT-3.5 hits the net. 2024 brings GPT-4 and dire warnings. GPT-4o eventually arrives, and now, it’s hot to be AI. 

The Age of Analytics Meets Generative AI

The tides shift back, and all the analytics companies of the past, the brands that valued business intelligence, realize that by embracing their roots, they can tap the power of the AI conversation–because what they have is what the world wants and now with AI back in the spotlight, it’s what startups and new competitors claim to be the first to achieve. In reality, those who were doing analytics and insight before AI are the rightful owners of the modern AI world. It’s not all generative AI and cupcakes, at least that’s the way it would seem. 

I’ve met loads of people in recent years who have either been all-in or all-out on AI. Yet, they were confusing what AI actually is. 

There’s another missing piece of the puzzle, the ability to intervene inside of systems based on data that seems entirely new. But if you go back to the origin story–the point at which a computer was first given the ability to engage with a person in a game of checkers, you can see the outcome. 

You can see that the computer is acting as an agent of the game. 

The computer has acted on information and given the user an option and opportunity to respond. That’s the agentic side, and it’s the simplest form of looking at what we all know now is agentic AI. But a checkers game isn’t really what I’d call agentic AI. 

So, what makes agentic AI? 

Agentic AI Rapidly Advances

Agentic AI has always seemed like a great idea on the surface. Take some input, act on the information, and move along. But what differentiates the agentic AI systems of today is rooted in another fundamental capability, how a computer interprets input across a non-binary format. In other words, can the computer recognize something beyond text, and the answer is a resounding yes. If the information is kept inside a given system, the functions are already present. 

But if you wanted to go beyond a single system, you have to consider the volume of data shared, the response data, and a series of compounding actions that occur based on the results of this one interaction. 

In this context, a system could mean multiple systems, thousands of computers, and trillions of potential sites, applications, or programs. 

It’s a chance to connect systems together that would make even the most seasoned of developers giddy and make them interact in entirely new ways. Combine that with the power of generative AI, wherein a machine is essentially having a conversation with itself and other machines. And BAM! You get an AI tool that can act on your behalf, you get a computerized set of processes that take on the characteristics of an employee, a contractor, or more notably, an agent. 

Thus, we get the agentic AI that’s been all the rage for the last year. Again, you’ve been using this all along. 

You’ve likely had conversations with chat it’s in the past, some pretty limited in their responses. Known as robotic process automation (RPA), systems were routing information, making decisions based on that information, trigger downstream activities and then giving you what you asked for. 

The difference between the chatbots of 2019 and today is that the conversational side, where the computer takes that information or even an idea, is that the computer could have already taken these steps, but no one wants to listen to a computer talking in the old Hawking-like voice of what happened at the coding level or execution-level. People wanted a simple answer, and a generative AI capability forms the last piece of the puzzle I mentioned earlier. 

So where does that leave us? Quite simply, AI only looks and feels like AI to us because the generative side now makes us feel like it’s not really a computer. The agentic side makes us feel like we have someone there inside the machine to do our bidding. And as computers process even larger volumes of data, the advancement of AI is all but uncertain. 

For core business functions, reporting, maintaining the mental well-being of your employees, hiring, sales enablement, and yes, marketing, the AI revolution is by definition only the progression of what was hypothesized in the context of computers nearly a century ago. And it’s volatile, it makes mistakes. But so do people!

Take this final example; a programmer could create an interface with all manner of bells and whistles, and the computer can do a lot, but it’s still subject to error. The fact that the idea behind the design is inside the mind of the coder is what creates the distinction between the two. But if another person, around the globe, in some archaic journal, report, blog or social media post had a similar idea, the computer can now make the leap to put the two together and create an even more capable program. And that specific information, gleaned from an unimaginable volume of information, my friends, is power. 

So yes, it can and does make mistakes because it’s based on probability, and the probability of an advanced coder making a mistake is still present. It’s only when you try to preview or test the functionality of the app or process that the errors become apparent. Again, we have a new feedback loop, where the computer has learned something that doesn’t work, and that information is stored to help you down the line. 

So in marketing, your perceived brand among prospects is everything. In business development, your perceived brand among partners is everything. In sales, your brand’s ability to solve problems is everything. And since we all use technology nonstop, machines are gaining more information about what does and doesn’t work. But sometimes, you still need the test, the people who can look at the output and say, it’s not what I imagined. The output could be varied, from a video to an action inside of a BI report. But you’re still using AI, even if you don’t know. 

Have a spell checker running? It’s an AI program. Have a BI dashboard that shows recommendations to your clients? That’s an AI. Have a smart thermostat? AI. Heck, even the analog calculator on my desk carries characteristics of what was known as AI back in the 1950s. And you know what else? I’m tired of the endless prompt sharing. 

I’m tired of everyone claiming to be the greatest expert in AI. I’ve been using AI for more than a decade; I was raised in the transition between the dot-com bubble’s formation and machine learning. I was taught to use technology from an early age, and it’s given me a clear outlook. AI has been with me my whole life; I just called it a grammar checker, a game, and a way of accessing the internet. 

Choose an AI-Inclusive Marketing Agency to Get & STAY Ahead

So don’t worry. If the machines were going to rise up, or if any core business function was going to be entirely replaced by AI and eliminate jobs, look no further than the phone in hand or pocket. You’ve wanted to do more with less all along. 

You’ve been creating more powerful AI functions all along, and you’ve been training it to do what you want. You might have called it a saved style, a preference for serial commas, a change between light and dark mode based on the time of the day, or maybe you’ve been avoiding it in every business action except for when you browse the internet. 

But here’s a secret you may not have considered. Since an AI is simply a computer acting on information and applying reasoning to find information, a search engine meets that interpretation, too. After all, a search query is nothing more than giving a computer a tidbit of information, where the computer and applications look into their memory–their files and records of indexed pages–to find the one that most closely matches the information you need. So if you use the Internet to search for something, you’ve been part of the AI revolution, too. You just didn’t realize it. 

JCI Marketing gets it, and we’re ready to help you embrace the next age of technology. We’re using it to improve our capabilities every day, and we know what works and what doesn’t. Yes, you could, with enough time and patience, have an AI, like GPT, do most of what you need. But business is all about speed and results, and few have that much free time to figure it out and apply GPT, whether generative in GPT Plus or agentic like GPT Pro, to get what you need to succeed. And hey, it helps to know when something that a computer thinks will work is at odds with something else, something you hadn’t discussed, shared or realized. That’s where JCI Marketing comes in, becoming the bridge between you and what everyone wants from the digital future. 

So maybe it’s time to stop trying to figure it out on your own and let us help your marketing and sales efforts get and stay ready for the future of B2B through AI. And lastly, bear in mind that there are many AI tools and platforms out in the world that can power through the information and help to grow your brand by turning all the millions of data points into actionable items. 

Connect with a JCI Marketing team member to get started.