BLOG
As part of our blog series on AI tools, Mithu Lucraft shares her feedback on AI for writing content.
Whether you write for social, web, email, blogs, or create other long-form content, it’s unlikely that you made it through 2023 without looking at what impact generative AI will have on content creation. Research and news reports suggest that writing is one of the most impacted areas from the explosion of genAI tools. Indeed, I’ve been in presentations that have effectively claimed that AI should put me out of a job. They have told their avid listeners, “Yes! These tools can create your social posts, and your blog posts! And hey guess what, they can even create a workflow for you to push this content live straight from their systems!”
Let’s be honest, I have a vested interest in showing that AI is not ready to replace me as a content creator just yet. I write pretty much all day, every day. Whether I’m crafting project summaries, writing a white paper for a client, or making sense of market research to provide strategic reporting, I make a living by delivering words that tell a story.
So, here’s the sticky question for the scholarly and professional publishing community: when we write for communities for whom authority matters a great deal, how authoritative is the content generated from AI? And how much time can these tools genuinely save in-house marketing teams? Having spent some time reviewing a variety of content tools, including free services like ChatGPT and specialist paid services, I’ve put together some thoughts for anyone who creates content to consider.
1. Who’s in charge?
In the words of one of my favourite authors on writing, Ann Handley, “AI is a robot perched on your shoulder. AI is not the creator at the keyboard.” Yes, AI can increase efficiency and supercharge your team’s content creation. But only if those tools are used correctly. Remember how back in the early days of social media, many an intern took the blame for inappropriate content shared via a corporate Twitter account? Lessons were quickly learned back then about losing control of a brand’s messaging. Now, we’re in a position to avoid those same mistakes by treating our AI tool as the company intern. By all means, give your AI tool a prompt to craft ten headlines; or plan out a series of blog posts on a topic. But giving any tool free rein to create and post that content on your behalf is a high-risk strategy. Be prepared to invest the time you would have spent training your junior marketer, and on evaluating their content. Specifically:
- Fact-checking is essential. It’s been well-documented that large language models like ChatGPT will hallucinate, or provide made-up information. The nature of absorbing training content that is then re-formed as part of outputs from AI tools means that authority can only be achieved through human intervention.
- Good writing is a creative skill. When our team at TBI discussed AI content outputs, we began to notice a pattern in the shape of what we were getting. It was repetitive. For example, paragraphs would often start with the same structure, or there was little variation in sentence length. There still needs to be a human in the process who understands what makes a story compelling, and ideally one who understands storytelling for a particular community.
- Accept responsibility. At the end of the day, much as several publishing policies now make explicit, it’s perfectly acceptable to use AI to help improve your content, but remember that it is a tool, not an author. That policy should be strictly adhered to, by all content writers across the organisation. Wired’s policy provides a helpful model to consider.[1
2. Tame your tools
There are many ways that AI tools can be used to boost productivity for content writing, but only through taking the time to use them effectively. Whether using ChatGPT[2] or paid services such as Jasper.ai or copy.ai[3],here are some ways you can harness your AI tools of choice:
- Establishing a tone of voice. Many of the writing tools available, including ChatGPT, can write to a specific tone and style. Using AI to edit to a specific style can be helpful where a team has multiple writers, to ensure there is consistency across the group. In some cases, where a brand has not developed a tone of voice guide, it’s possible to use inputs such as existing copy or webpages to help the tools tune their style. Some paid-for tools will save specific tone of voice prompts in their workflows which can be helpful for teams that work across multiple brands simultaneously with different voices. It’s worth reverse-engineering the process to ask the tool you use to summarise a tone of voice, to see how this compares with official brand guidelines.
- Optimising prompts. Without taking the time to provide the right input, outputs from these tools won’t be valuable. There are many blog posts, guides, and resources that identify standard prompts that can be tailored to support marketing and communications teams. These offer a great way to start using free tools such as ChatGPT, meanwhile paid services will have large libraries of generic prompts for marketers. With customisable GPTs[5], it gets even more exciting, with a way of tailoring prompts into a custom tool you can build and use for your specific requirements. Education and training are essential to get the most from the tools available.
- Introducing scale. For larger teams, paid options such as Jasper and copy.ai [3] offer a means to streamline workflows for repetitive copy creation such as product listings, shifting time from composition to editing. Here again, the emphasis is still on assisting marketing experts, rather than removing them from the workflow.
3. Look at efficiency gains
There are some uses of AI that require very little input, and still provide immediate benefits to content writers.
- Proofreading. I know I’m not the only one who increasingly edits all their content on-screen, instead of printing out every piece of writing I do. Tools like Grammarly[6] are part of a growing suite of embedded tools that help me to ensure I present the best version of my copy.
- SEO. Something that has typically required time and effort to do alongside creating copy is to optimise it for search. AI tools can help speed up the process, making sure copy is search-friendly. There are also several SEO-specific services available that look specifically at website optimisation such as Diib and canIrank [7].
- Expected performance. Some of the paid tools will also optimise content for specific channels, giving feedback on expected performance before ads have even run. Anyword [4] is an interesting example of this.
Getting started
AI will continue to be infused into every piece of software we use in the office environment, such as Microsoft Co-pilots. As writers, these tools will get better and better and it’s important to continue to experiment and embed the right tools into our processes for writing. In putting this blog together, I reviewed nearly 30 different AI tools for content. Many of these overlap in their functions and effectiveness. Pricing varies from free to around $500 a year, but several paid tools are offering a free trial. Starting with the free options and these free trials can be a good way to test out what might work for your team.
Ultimately, remember that in all instances, these are tools to enhance, not replace, your content writers. Used in the right way, these tools can be an excellent assistant to your human experts! As someone who remains passionate about writing and storytelling, I am excited to continue learning and experimenting with AI for writing.