In part one of this series, I explained the ethical and potentially legal issues surrounding generative AI in creating content. In this post, I will discuss how I use AI best practices to help make better content for Blue Star’s clients while avoiding the ethical and quality pitfalls I described in the last post.

It’s essential, however, to set expectations. AI is not a person, and it is not magic. Using it will not suddenly make you ten times more productive.

How I use AI now: Best practices

Lately, I’ve started thinking about generative AI as a “word calculator”. I’m historically terrible at math. I lean on the calculator app on my phone a lot. But people who are great at math, say researchers or professional mathematicians, also use calculators. In their hands, a calculator becomes an excellent tool for enhancing what they are already good at and tackling immense or infinitesimal values like the Planck length in quantum mechanics.

Calculators also help professionals communicate with non-professional math dum dums like me. In the same way, AI helps me write at a greater volume, check my work, brainstorm, and communicate with non-writers.

Grammar checking

I’ve used Grammarly to check and improve the correctness of my work for years. This AI-powered system isn’t generative–Grammarly Go is their answer to ChatGPT.

Summarizing research

An incredibly powerful prompt is “Pull out the main points.” Sometimes, I accumulate over 50 pages of notes for a single blog post. This prompt is an effective way of gathering my thoughts, beginning an outline, or reviewing what I’ve learned so far. “Summarize” is also an effective prompt that achieves similar results.

Brainstorming

I suck at thinking of blog post titles. Every writer has their weaknesses, and this is one of mine. But I can prompt, “Give me ten titles for this blog post,” and Grammarly spits out ten pretty good titles. They aren’t always perfect, but they get the juices flowing. I also rarely use the title “as is” from Grammarly, but one or two are usually usable with a bit of tweaking.

You can also brainstorm topics if you are in the campaign planning phase or doing keyword research. For example, I asked Grammarly to generate blog post topic ideas for the keyword “kraft paper packaging”. Here are the top five results (it gave me 15):

  1. “The Environmental Benefits of Choosing Kraft Paper Packaging”
  2. “5 Creative Ways to Use Kraft Paper Packaging in Your Product Design”
  3. “Kraft Paper Packaging: A Trend That’s Here to Stay”
  4. “How Kraft Paper Packaging Can Elevate Your Brand Image”
  5. “The Ultimate Guide to Customizing Kraft Paper Packaging for Small Businesses”

Nothing is outstanding there, but again, the point is not to get a perfect idea right out of the gate but to get the creative mojo working.

Project initiation

The first time I received a creative brief from a client that had (very obviously) been generated by gen AI, I thought I might throw up. “Welp, the machines have come for my job,” I thought while humming the theme song to Terminator 2.

However, I quickly realized that Gen AI helps people who are time-crunched to communicate quickly what they are looking for in a project. I use little of the original AI text, but occasionally, a line or two will make the cut. Again, I treat AI as a source, not a writer.

Spot gaps and weaknesses

“Point out any gaps” can spotlight any high-level logic flaws or point out times when you should give examples of the point you are trying to make.

AI best practices

Here are some guiding principles I’ve given myself to :

  • I don’t use AI to write first drafts. I feel like that is only one step removed from copying and pasting from Wikipedia. While it’s possible to work this way, I find it more complicated than just writing from an outline I created or one I knocked together with the help of AI.
  • I stick to using it on text that I write or the client owns that for repurposing.
  • Always double-check for copyright flags in case something snuck in during the process.
  • Avoid using Gen AI to create graphics for your text (and that is a topic for a different blog post).

Why AI doesn’t replace marketers/writers

Some predict that 90% of content will be AI-generated in a few years. Standing out amidst a sea of clunky, off-putting, Shrimp-Jesus-y, spammy content slop will become more crucial than ever in the history of business. Differentiating your brand with quality, thoughtful, empathetic content sets your organization apart.

AI doesn’t have imagination or empathy. It might sometimes make a pretty convincing mimic of those two things, but ultimately, it’s an illusion. It doesn’t understand, not in the way people do.

Ultimately, it is possible to use AI ethically in its current form, but you must be very careful to avoid the quality, legal, and bias issues that crop up. I hope that future iterations of AI have more guardrails and that creators can opt in or out of AI training. On the whole, AI needs more transparency and rules.

Need help sorting out AI ethics and best practices? Get in touch.