Picture it, summer 2023. It’s hot, the air quality is low, and everyone is consumed by holidays, vacation, and news. Your client emails you at 4:30 in need of a POV article on a topic you have only a vague recognition of. Deadline ASAP.
Content needs and requests can span many departments and teams, from design to development, to times you just need to fix or add a few words, or write an intro but you don’t have a copywriter to engage.
“I know it’s last minute, could we just use ChatGPT to get it done?”
You have the option of ‘just using ChatGPT to get it done. It will be good enough, you rationalize, in a moment of weakness. But then you remember … Not all content is king.
Risks of using ChatGPT to create content
There are certainly instances and ways to use ChatGPT in the content creation process, but it can have risks, especially if relied on to deliver automated content.
Would you crowdsource your client’s article to users on reddit, or folks on your local neighborhood app? Of course not. You have no way of knowing if the information they give you is true, real, or even if it’s original.
Same goes for ChatGPT. Because that’s where ChatGPT got some of it’s machine learning from!
1. ChatGPT (and all Generative AI) is Crowdsourced Content
AI/GAI is trained in information it is fed, and we do not have insights into what information it was initially taught on, nor do we know what is being continually added by the masses of users now. We’re familiar with the phrase “garbage in, garbage out”, well, what we do know is that the information AI and the content it was initially trained on is online content, including potentially less desirable, less curated sites that didn’t necessarily ascribe to civilities or decorum.
2. GAI content is biased, really biased
We all have biases, known and inherent ones, ones that are so deeply engrained in us culturally we don’t even realize we have them. As people we can respond to that and be more alert to look for and avoid them. We know this because we live in reality, in the virtual and real worlds. AI doesn’t.
The content repositories that make up GAI are enormous, so enormous that there are only a few companies in the world large enough to handle the content, so all GAI apps use those same repositories. And that content comes from the web. It’s all digital content fed into the machines, to learn from. Machines that don’t know nuance or context.
“These algorithms are trained on data that reflects not the world, but the internet – which is worse, arguably, That’s going to encode the historical and present-day histories of marginalisation, inequality, etc. There isn’t a way to get out of that … because data is authored by people. It’s always going to be a recycling of the past and then spitting that out, projecting that on the present.”
Meredith Whittaker, President of Signal, former AI researcher and academic at Google, Co-Founder of NYU’s AI Now Institute, and advised the FTC on AI. Interview in The Guardian 6/11/23
3. GAI Content is Old
Content created for a professional endeavor strives to be new, innovative, applicable to a business’s clientele. It is designed to both elevate the company and educate the audience, by offering something new; a new perspective, a new angle, a new approach.
AI uses existing data and algorithms fed into it, and uses statistical and machine learning, it does not generate new ideas, so will never deliver anything new, groundbreaking or unique, at least for long. Whatever it generates for you it will use to generate for others. And it won’t represent you, it will only reflect the millions, billions of data packets it’s drawing from.
Benefits of using ChatGPT to create content
It’s not all doom and gloom, though. ChatGPT can be used in content creation. It can be a great tool in improving efficiencies and increasing productivity.
1. Research
As content strategists and creators we spend a good amount of time getting up to speed and learning the nuances, obstacles our clients, their industry, and their offerings face. ChatGPT can give you a summary of key points, obstacles, and an overview of all of these with the right prompts, saving you from hours of research.
2. Brainstorming
ChatGPT can help brainstorm perspectives, directions, and ideas for your work, which is especially helpful when you’re facing a deadline and writer’s block.
3. Initial drafts and outlines
You can use ChatCPT to create a first draft or outline to get you started. It can also be used to review and proofread your draft and make suggestions, beyond just spelling.
Making the most of ChatGPT when creating content
Being aware of all of this is critical because it sets the foundation of expectations for using ChatGPT, but also opens up the doors for how it can be used in the process of content creation. Here are a few tips to mitigate some of the risks while taking advantage of the power of ChatGPT.
1. Use for outlines and preliminary research.
This is perhaps one of the best features of ChatGPT, to provide quick and succinct research and output for topics quickly and effectively, a way to quickly learn the topic and find key topics to focus on. It can generate initial outlines and top points on a topic, just remember, these are general, and you’ll need to confirm and revise based on the specific needs of your work.
2. Confirm and validate the content generated.
Based on the recommendations, validate and find updated research, trends and content. You want to make sure that the content you’re delivering is fresh and applicable. In this day and age we know things can change quickly.
3. Review drafts for readability and appropriateness.
Review all output by ChatGPT for validity, but also for voice and tone. Be sure to follow established editorial guidelines, and make sure your content is inclusive and supportive and aligns with the voice and tone required. Be alert for any potential subtle biases that may remain in the edited content.
4. Develop a thorough review process with the teams.
As with any work product, it is important to develop and follow a review process. Aside from more eyes and input, it also makes the content more effective in meeting the needs of the full audience, and exposes any missed biases or outdated, missing content that may linger.
Conclusion
As consultants, we need to take this opportunity to not only understand the capabilities and risks moving forward with generative AI but also provide guidance and expertise in content strategy in the age of generative AI.