If you work in marketing, you’ve lost track of how many headlines about AI marketing and ChatGPT you’ve seen by now. It’s interesting, fun to experiment with and frankly, if you’ve read Asimov or watched any of the Terminator movies, it’s a little scary. As AI technology develops, it is going to increasingly touch almost every aspect of our lives, in ways we cannot even anticipate. It will be disruptive and nothing short of world changing. The question for now is, what does the emergence of AI technology like ChatGPT, Google’s Bard and others mean for marketers?
Marketing AI has not made us obsolete … yet
First, let’s establish what it doesn’t mean.
For the foreseeable future, human marketers are still relevant and needed. Technology, even something as cool as ChatGPT, is not a magical marketing panacea. I’m reminded of the story of a friend who was leading a marketing automation installation for a large company. When she asked about the company’s plans for developing their marketing automation campaign strategy and the content to power it, the executive in charge sheepishly replied, “Doesn’t the software, you know … automate it?”
Now with AI, maybe one day that misguided executive’s dream will come true and marketing will truly be “automated.” “Dear AI – I would like more hot marketing leads please.” We’re not quite there yet.
For now, AI is a tool that takes expertise to utilize. While a Google engineer famously stated that their AI had become self-aware last year, and then was summarily fired, we’re still a few years away from achieving a dystopian technological singularity. Maybe at least we’ll finally have self-driving cars by then.
The AI MarTech landscape
OpenAI with ChatGPT and Google are the predominant players. More players will emerge with their own AI, or more likely will build out solutions that leverage ChatGPT or Bard. At this writing there are approximately 40 AI writing tools in the marketplace. Most are using GPT-3 (predecessor to ChatGPT) or a similar natural language generation (NLG) AI as the underlying engine and then layering their own development and features on top.
How marketers can use current marketing AI tools
As ChatGPT or Google Bard are not currently offering API access, the ability to integrate and leverage the tools in your organization are presently limited. However, there are a few use cases where we are finding benefit in using ChatGPT.
Here’s where we’re currently seeing some benefit in using ChatGPT:
- Topic generation
- Title brainstorming
- Email outline inspiration
- Social posts
- Ad caption drafting
The challenge with all of the above is that to make them truly useful and effective, human editing is required. Once we revise, fact-check and edit to be compelling to a human, we’re not seeing significant efficiency gains.
Current limitations of marketing AI
Beyond lack of API access, there are several limitations of marketing AI content. However, we view some of these as potentially temporary limitations, as AI will continue to evolve and get better over time.
We had our writers and editors put ChatGPT through its paces, and while it’s blazing fast, it doesn’t match a skilled human writer on quality. Here are the specific areas where we found it lacking:
- The output is often robotic (understandably) in tone and style. The writing lacks the rhythm, creativity and sentence variation that makes good writing interesting and engaging.
- It does not have the ability to understand brand voice, product priorities, key messages, editorial preferences, etc.
- We found that ChatGPT struggles to scratch below the surface. In our testing, it rarely pulls hard data to support a point, which is often an essential element of building a news story. A human will have to do the research, or, if something is provided by the bot, they will need to fact-check to avoid publishing a mistake.
- Finally, it does not include quotes from a subject matter expert or an individual’s experience, which is a tactic we often use to strengthen our work.
10X content and a race to the bottom
I recently had the opportunity to sit in a room full of executives where ChatGPT was demoed. The speaker suggested that you could take the raw content that ChatGPT spits out, post it to your blog and call it a day. I could see the sheer excitement in the room to pounce on this tactic. This is a great thing for the rest of us.
You see, if everyone is using AI-generated content it’s a race to the bottom. You likely want to rank for the same keywords as your competitive set. If you and your competitor are using AI-generated content, you’ll have a morass of undifferentiated content that does nothing for your business. For content to perform and generate results for your business, it needs to be exceptional.
[Read More: What is High-Quality Content in the Eyes of Google?]
I’ve long been a fan of Rand Fishkin and he introduced the concept of 10X content. Simply meaning that if you want to show up on page one of search results and beat out the incumbent, your content needs to be ten times better than what’s already out there. This is as true today as it ever was. AI gives you and your competitor largely the same content and quality; it cannot create truly distinctive work.
Marketing AI and SEO
One of the primary goals of the content we produce at Brandpoint is to aid our clients in SEO. As most customers are searching for a solution to their problem, not your brand, this is very important work. As it relates to using marketing AI like ChatGPT to generate content for your website, Google is very clear on the subject: Don’t do it.
Search Advocate John Mueller says content automatically generated with AI writing tools is considered web spam, according to the search engine’s webmaster guidelines. To test this concept, Neil Patel recently ran an experiment on 100 websites using AI-generated content. He posted both raw AI-generated content and AI-generated content that was edited by a human. The raw AI content yielded a 17.29% decline in traffic on average and in some cases, the traffic dropped as much as 80%. The human-edited AI content fared much better, but still saw declines of 6.38% on average.
As Google rolls out Bard, it remains to be seen whether their definition of web spam will change. Until then, if you want to rank well in search, don’t use raw AI-generated content.
Marketing AI content accuracy
In our testing of AI writing tools, we discovered they are prolific liars. The technology is programmed to give human-like responses to our questions. In doing this, AI can generate entirely false content or inaccurate content that appears correct. While AI will often provide correct answers, if you are creating content with AI writing tools, you will want to fact-check everything.
Marketing AI content legal considerations
Using marketing AI writing tools to create content for marketing purposes is murky legal territory. The big issues are copyright and IP infringement. It gets to the question of who owns the content in the data set the AI is drawing from. There is pending litigation on this subject that could create precedent on the broader topic of AI and copyright ownership.
Further, it appears that you would have a difficult time under current law securing any copyright protection for AI-generated works. AI is the future, and this will all be sorted out legally and standards will be established. Until then, we advise proceeding with caution when using AI-generated content.
Establish your position and process on marketing AI content
Whether you’re using marketing AI internally or for client work, it’s important to develop a set of guiding principles for how you will use the technology in your organization.
Here’s our current policy at Brandpoint:
First, Brandpoint will not submit raw untouched AI-generated content as finished client work.Brandpoint will always utilize the best tools and technology available to inform our work and generate desired client results. We believe that human emotion, critical thinking, experience, knowledge and imagination, coupled with human expertise in utilizing tools, are what drive value, not the tools in and of themselves.
In addition to policy or position, we need to consider how AI marketing tools will require new processes. As an example, if you use freelancers, do you need a policy regarding their use of AI writing tools? Do you need to incorporate AI detection tools prior to using content? Your business will be facing these and many other questions related to marketing AI and writing tools. It’s a good idea to start working through them now.
The future of marketing AI
The leap from GPT-1, to GPT-3, ChatGPT and now Google Bard in data set and processing power is both profound and difficult to conceptualize. Suffice it to say that we’re just getting started and it’s going to be mind-bogglingly good. The shortcomings that I discussed earlier in this work will likely be resolved to varying degrees over time. Where the technology likely becomes disruptive and more creative is when it can make its own inferences and be trained on internal databases in addition to larger external data sets.
It’s also going to be about a lot more than writing. I see everything we do in marketing being enhanced by AI technology. Customer service chatbots that universally suck now will become wildly smart and actually helpful. Your marketing automation software will automatically create lead nurture programs that are significantly more sophisticated than today, and it will optimize on the fly with real-time data. Social posts will self-optimize based on user engagement. Ecommerce platforms will know what I need before I do. AI will be able to predict what content topics will generate the most user engagement and results based on the current cultural zeitgeist. AI is going to touch every aspect of marketing, and our lives.
Marketing AI final thoughts
We are living in the world that I read about in sci-fi novels as a kid. It’s really cool and a little scary. We are going to see a lot of change in the coming years. As marketers we need to embrace the rapidly approaching changes and define how our uniquely human creativity will best utilize these tools as we meet the future.
And finally, an Author’s Note: An actual human wrote this article.