Should You Replace Your Copywriters with ChatGPT?
There’s been a lot of talk about ChatGPT …?maybe, maybe too much talk. And you’ve definitely seen ads like this:
For years, tech companies have promised us magic wands that solve for all sorts of marketing problems. Analytics, automation, scheduling, etc. But now, a technology is being offered as a solution to a creative “problem” — namely, the problem of copywriting.
So let’s tackle the 64-dollar question:
Should you replace actual, real-life copywriters with ChatGPT??Can?you?
First, let’s get a handle on the technology …
How does ChatGPT work?
ChatGPT uses a neural network architecture to generate responses. By tapping into a large dataset of conversational text as its training data, ChatGPT learns the patterns and nuances of natural conversation to produce responses that are highly likely to be relevant and coherent. And by tapping the vast dataset that is the internet, ChatGPT is able to find common patterns in information and use them to piece together narratives that have a high probability of being relevant and coherent.
In a sense, ChatGPT is leveraging the internet to “remix” explanations, arguments and narrative forms on a wide variety of topics.
If you were to ask ChatGPT, for example, to make an argument for why Michael Jordan was better than LeBron James, you will get a strikingly accurate and reader-friendly breakdown. That’s because tons of material has been written on that subject. With loads of data to work with, ChatGPT finds the most common arguments, phrases, transitions, etc., and mixes them into something technically new … even though the information this “new” article is built upon is all rehashed.
(If, by contrast, you asked ChatGPT to explain why chocolate is commonly used as a substitute for motor oil in Zimbabwe, it gets stuck in mud … since no one has ever written anything about that fake thing I just made up.)
This is where we see the key limitation of ChatGPT. In a sense, it’s like a really good DJ who doesn’t know how to write original songs. It can play really good music all night long, all of which has technically never been heard before … but it can’t do anything without sampling other people’s content.
领英推荐
Good writing is good thinking.
There’s an old adage that?good writing is good thinking. And the truth is, ChatGPT doesn’t think. It curates.
IBM’s Watson — another AI product —?smoked every Jeopardy contestant?who ever dared face it. But Watson doesn’t think either. It?processes information, and does so at speeds no human could ever match. It’s like a math whiz working by hand, competing against someone with a calculator.
The difference is, the advent of the calculator didn’t drive people to stop teaching math. We think of a calculator, correctly, as a wonderful tool to?help?us do math.
The same should be true of ChatGPT and writing. It can serve as a shortcut for research. It can put together coherent explanations of things (again, based on previous writings). It can build the structure of an email or blog post [1] or term paper [2]. But it cannot?write. And if you use it to “write” for you — to take over the entire process — you’ll be with remixed content devoid of original thought.
It may pass for writing, but it’s tantamount to clever plagiarism. Like Styrofoam peanuts: useful filler, but nothing more.
Case in point: Check out this much-talked-about Mint Mobile commercial, where Ryan Reynolds reads an ad he created using ChatGPT:
The ad copy reads astonishing like a “real” ad copy. And having worked as an advertising writer — and copy director and creative director — for a long time, I can assure you that this ad is not dissimilar to what many,?many?copywriters would turn in if given this assignment. So looked at that way, sure, ChatGPT is a fine replacement for a real-life human copywriter.
But what makes this ad good isn’t the ChatGPT-produced copy. It’s that someone thought it would be interesting, and potentially funny, to have Ryan Reynolds read such copy aloud and react.?That?is what makes this spot different and special: the idea at its center.
An idea created by a human being. A writer. An artist.
The late, great Bill Bernbach once said: “Imitation can be commercial suicide.” He also said: “The difference between the forgettable and the enduring is artistry.” The ultimate goal of any promotional copy you produce — whether that be an ad, a social media post, a blog article, etc. — is to make the audience stop and go, “Huh.” To make an impact. To drive an action. To be remembered.
You can’t achieve that with technology. You can only achieve that through artistry.
1. Blog posts are a great example of ChatGPT’s two-edged sword. On the one hand, a marketer can use the tool to generate a coherent article on a subject your company is supposed to have expertise in. On the other hand, it can only “remix” content that already exists. So in using ChatGPT to produce blog content, you’re not producing anything original. That’s the opposite of thought leadership. Also: Since the content you’re using to fuel this new content is already indexed in Google, good luck ranking high in search.
2. ChatGPT has just about killed the student essay. As one professor wrote: “You can no longer give take-home exams/homework … Even on specific questions that involve combining knowledge across domains, the OpenAI chat is frankly better than the average [student] at this point.” The implications here are enormous. Again, good writing is good thinking. And when you give kids a shortcut to writing (which, let’s face it, most are going to take), you remove one of the best exercises for constructing an argument and thinking creatively — writing.
I am a writer and editor who works in healthcare, health tech, science, and business-to-business (B2B). I've worked on websites, email, print, video, audio, white papers, and other long- and short-form content.
1 年I asked ChatGPT "What are the pitfalls of using ChatGPT?" This is the response: There are several potential pitfalls when using ChatGPT: 1. Language bias: ChatGPT, like other language models, may perpetuate and amplify existing biases in the data it was trained on. 2. Lack of context: ChatGPT may not have enough context to understand the full meaning of a statement or question, leading to inaccurate or irrelevant responses. 3. Misunderstanding: ChatGPT is a machine and can misinterpret or misunderstanding the intent or meaning of a statement or question. 4. Lack of common sense: ChatGPT is not able to reason or make inferences based on common sense knowledge, and may therefore give incorrect answers to certain questions. 5. Lack of domain knowledge: ChatGPT may not have knowledge about certain specific topics or domains, and may give inaccurate or incomplete answers.