By now, you’ve almost certainly heard about ChatGPT, the recently released artificial intelligence chatbot from OpenAI that functions like a cross between an encyclopedic second brain and instantaneous writer.
If you somehow have not heard of ChatGPT, then by all means, please meet this potentially revolutionary technology that, depending on whom you listen to, has already killed the college essay, will destroy high school English, is coming for the creative class next, and might just be a virus unleashed on society—one that suddenly feels quaintly not-so-modern.
Those are some of the darker takes. Some more optimistic perspectives include that it can enhance both productivity and creativity, can do all sorts of coding, does great impersonations, and can potentially be a tool to help novelists perfect their craft. And while we’re looking for reasons to be excited, not worried, it’s worth noting that advancements in knowledge and technology have not yet led to widespread unemployment and have generally been good for society.
However you slice it, artificial intelligence as a significant factor in our daily lives, it seems, has officially arrived, catapulting from somewhere near the horizon to right here in front of our very eyes.
That’s particularly true for people who deal with the written word, including content creators and marketers.
So, how can marketers use ChatGPT right now? And what are its limitations? Here are a few things it can do, some that it can’t do, and how we, at least, suggest you can put it to use right now.
First, let’s start with what Chat GPT can’t do.
6 things ChatGPT can’t do right now
- Write things without error: The OpenAI folks readily acknowledge that ChatGPT will get things wrong. If you’re looking for it to create a flawlessly researched and written piece of content, it can not do the former.
- Write about anything current: The platform’s large language model has not been updated since 2021, so it does not know about anything since then.
- Provide rich content without a lot of input: The more input you put into ChatGPT, the better it gets. But that, of course, requires a high level of effort. By the time you’ve thought through the various parameters for writing a distinct, high-quality piece of short-form content, you might have been better off writing it yourself.
- Build credibility with Google: Google has worked to identify artificially created content for a long time, and it penalizes content it determines was not created by humans. And ChatGPT is working on creating a watermark that will make its content easily distinguishable as non-human, so even if Google can’t recognize what is made by ChatGPT right now, it probably will in the near future.
- Provide individual perspectives: ChatGPT can mimic people who are well known (you can ask it to write anything you like in the voice of any celebrity or historical figure, and the results will be impressive), but it can’t write something from, say, the perspective of your company’s CEO.
- Write in a way that sounds truly human: This is subjective, but the quality of the writing is probably best described as a little above average, but there’s something lacking, a certain amount of truly natural, human-sounding tone to the copy. The best examples I’ve seen that sound the most natural are parodies or things written to sound like specific other people, like Mark Twain or Jerry Seinfeld.
25 years later, the promise of Ask Jeeves is finally fulfilled.
4 opportunities for content marketers
Generally speaking, ChatGPT is probably best used for the ideation process or for sketching out a conceptual framework in a few different ways.
- Use it to generate ideas for blogs: Stuck on how to get started for a blog or other piece of short-form content? Ask
JeevesChatGPT. That said, without a lot of specific input parameters, it’s going to give you a very generic blog. And if ChatGPT kills off the generic listicle, that’s probably not a bad thing. How many more versions of “10 Ways to Save Money” do we need?
- Use it to generate social media copy: Last week Electric Kite’s Louis Miller posted a short blurb about his opposition to influencer marketing, created by ChatGPT. As he acknowledged, it wasn’t half bad. Social copy generally packs a few ideas into a short space, something ChatGPT does well. There seems like a lot of potential to use these tools to create baseline social copy, modify it to match your brand’s voice, and post.
- Use it to generate short-form ad copy: Similar to the social media copy, there is the potential to use ChatGPT to quickly create a lot of options for short-form ad copy that you can then edit or modify as you see fit, and select the best ones you want to use.
- Use it to create basic SEO and content plans: Much like a listicle, if you ask ChatGPT to give you ideas for creating SEO and content strategies, it can create the kinds of high-level outlines that you could get from any number of content marketing listicles. Give it a specific subject matter to plug into those outlines, and it can give you some specific SEO keywords to target and content topics to tackle. Again, these are useful starting points, but not full-fledged strategies or granular analyses based on a specific company’s needs and goals. As Zain Kahn noted, this type of analysis is roughly the output you would expect from an SEO marketer with 1-2 years of experience.
Want to write a social media post but don’t have the time or want to do it yourself? There’s an app for that.
One significant risk that is probably clear by now is of people misusing this new technology to create generic, uninspiring content. This is a powerful tool that, ethical implications notwithstanding (put a pin in that thought for one more minute), can create good content quickly; but the output will only be as good as the input.
Ultimately, while this first publicly available version of ChatGPT may be useful for creating baseline drafts or getting ideas going for various copy and marketing collateral, to create something valuable and relevant to your users, you still have to do the heavy lifting.
It’s not that ChatGPT can’t be creative, or at least witty—if you want it to write a very witty poem about, say, a man getting a peanut butter sandwich out of the VCR, it can do that. But to create copy that is accurate, relevant, incorporates your brand’s voice and tone (which likely won’t be included in a large language learning model… yet) and the specific need states of your customers, you would, at the very least, need to give lengthy query parameters and then heavily edit the byproduct. And at that point, from an efficiency standpoint you’re probably better off doing it yourself.
Beyond that, of course, there are ethical and moral considerations. ChatGPT is new, and has captured a ton of attention. But there are other similar products out there that claim similar capabilities, and there will surely be more coming soon. Depending on how quickly these tools are adopted, it may be necessary to be transparent about the use of these technologies for creating content. That would change if we, as a society, quickly adopt and accept AI-generated content as a representation of true, original thought from the person presenting it, knowing and accepting that the presenter used AI as a tool in their production process.
But second, and more immediately, is the fundamental question: how much do we value the work of humans for the sake that it was created by humans? We, as an agency, believe strongly in the inherent value of humans; it is an actual pillar of our internal branding. We value humans, in general, and the people that we work for and with, in particular. That doesn’t mean we won’t use new tools and technologies to help us do our jobs better, but we value the creativity, passion, experience, viewpoints, and skill that we, as humans, bring to the table—and, at least for now, we don’t plan to replace them with robots.
Ultimately, we don’t know what’s coming next, but we do know that embracing change and looking forward is always more helpful than hand-wringing, lamenting, and digging in our heels. So with that in mind, we’ll take a glass half-full approach to the coming of AI. And we’ll make sure it’s a stiff drink.