- There has been a sudden rise in Large Language Models (LLMs) like ChatGPT and their impact on the creative industry
- The age of the prompt has finally dawned on mankind, with creativity starting in the machine and working its way outward
- There is an urgent importance for expert drivers to guide LLMs and avoid obstacles like biases and capability overhang
Somewhere between the .1% and an algorithm, here we are, worried ChatGPT’s after our jobs and houses. Something man-made forced us to self-assess. Someone flipped a switch. Somewhat loudly.
Across the brandscape, professionals wonder aloud about their value and “place.” Generative Artificial Intelligence (GAI) like OpenAI’s ChatGPT and Google’s Bard are powerful proofs-of-concept, and yet, pretty much automated versions of the childhood game 20 Questions. Only now it’s 175 billion questions.
Welcome to the Age of the Prompt. Just like playing 20 Questions, we use Large Language Models (LLMs) to arrive at a satisfactory output only through expert interrogation—making what’s inside the questions our new brand creative frontier.
Nearly 70 years ago, men with pipes and/or suspenders agreed that a binary computing architecture would double itself until interconnections surpassed the human brain’s and appear to “think.” Not long after that, we started allowing machines to record our inputs to make work faster.
Today, after generations of recording, ChatGPT’s just playing the sound back. And we’re scared of our own voice.
Hardly the first time a new technology passed fear on the road to mainstream. The telephone was “an instrument of the devil” that let people talk to the dead or could get you struck by lightning, just as “Computerphobia” was a thing in the early ‘80s. It’s not wrong to be skeptical of early tech, but it is a shame to dismiss it before the world tries its hand. Responsibility may just prevail.
To get where we are now required trillions of “queries,” another way of saying people gave computers commands in “machine language” for decades. Now that we can prompt machines like ChatGPT in our own natural language, the query can go back to where it came from (the bar exam).
Despite its unnervingly human veneer, ChatGPT is one more decision-support machine for brand owners and their teams, wholly dependent on a flesh-based expert to safely navigate it to its destination. Just as horseback riders became “drivers” with the arrival of the combustion engine, I believe our lexicon is about to change again—this time from creative, designer, and writer to something like “driver.”
A key difference between humans and computers has always been nuance; people get it, machines don’t. So as a driver with a point to make, I’ve just been hypothetically assigned to a pitch for a struggling online food delivery service, and entered one of two closely related prompts:
Prompt A: Write about brand challenges in restaurant delivery generated a story that began with food quality but shifted abruptly to brand reputation, competition, and customer loyalty.
Prompt B: Write about last-mile challenges for meal delivery brands focused on traffic congestion, parking, weather, and the like.
The order and specificity of the prompt components made all the difference; as you’d imagine, the relevance of ensuing outputs in the same chat depended on this foundation.
For the first time, brand creative will start in the machine and work its way outward. All creativity has ever been is making original associations that are relevant to an audience and objective. To do this skillfully takes time, and not all that time is spent doing creative things, e.g., data-mining, transcribing, basic coding. Just think of what we can do with the time and headspace saved when a machine eliminates workflow minutiae.
Case in point: how many phone numbers can you rattle off? Eight maybe? Our phones remember hundreds of them for us and life’s simpler for it. If machines can perform my preliminary research, experiment with a tone of voice, or slay writer’s malaise in milliseconds, I can direct my energies toward further value for the customer.
At the brand level, language-matching is an art form. Successful brands communicate simply, using the customer’s words where possible. With GAI, identifying and acting upon patterns and themes in huge bodies of unstructured text can be performed in seconds with a series of prompts, not days spent capturing and comparing. Now we have room to pursue bigger concepts, solve more pressing problems.
That requires an expert driver, as does avoiding obstacles in the road like biases, hallucinations, and capability overhang. Contrary to press portrayals, this thing doesn’t think. It has no feelings or opinions. It processes words as tokens and acts yes or no according to billions of inputs alongside a driver’s guidance.
Sure, people are worried today about what GAI represents, but we were too giddy to consider 15 years ago—while sleeping in line outside Apple stores for the first iPhone—where mobile data-generation at scale would lead for customers and brands alike. Instrument of the devil?
Like it or not, the Age of the Prompt has dawned. In fact, humans love “prompt” so much we made it a noun, verb, and adjective. The prompt prompted a prompt response. Take that, query.
Unlike most humans, LLMs are “forthright” about their own limitations. Good. Now’s the time to learn how to eruditely guide these vehicles—to drive literal value for customers, brands, and agencies—while they’re still being tuned.
So let’s go, ladies and gentlemen. Start your engines. Drivers wanted.
Steve Susi is the Director of Brand Communication at Siegel+Gale. He is a 360° content generator, Customer Obsession advocate, architect of innovation cultures, and an unabashed poet. Steve is the author of ‘Brand Currency’ and has served a leadership role at Amazon.
Subscribe to the ClickZ newsletter for insights on the evolving marketing landscape, performance marketing, customer experience, thought leadership, videos, podcasts, and more.
Join the conversation with us on LinkedIn and Twitter.
The post The age of the prompt: Brand drivers wanted appeared first on ClickZ.