Published on April 7, 2023
AI chatbots like ChatGPT won’t replace content strategists but will make marketers’ work more efficient and distinctive.
A few of us recall when calculators were not allowed in class and we had to learn how to use a slide rule. Keeping track of decimal points in our head wasn’t anything we hadn’t learned from a Radio Shack dot-matrix display, though, so chemistry class simply became that much more of a slog. Communicators have reached the same point with ChatGPT, a calculator for words that many public relations agencies are still reluctant to embrace.
Conversational artificial intelligence tools like OpenAI’s ChatGPT and its bot rivals—Google Bard and Microsoft Sydney among them—will not replace PR professionals but will make copywriting faster and focus content strategists on exploring unique brand values. That’s a win-win for us and our clients. The handy Tandy calculator eventually earned a place in the classroom and paved the way for laptops and tablets. The same is in store for AI, so it’s worth sharing some of what we’ve learned.
AI in PR: The First Rough Draft of Story
If the slide-rule analogy seems ancient, think of the new chatbots as autofill for an entire document. Their large language models simply predict the next word in a sequence, on and on for hundreds or thousands of words. The result is far from unique, though, and more helpful in structuring a document or checking grammar than saying anything that will attract a journalist or search engine’s notice. On the production side, AI chat also corrects coding syntax, generates images and produces working code snippets, shortening the “staring and swearing” stage of development. (ChatGPT debugged the referral box that ends this post.)
Some communications teams have been happy to turn over their thought leadership to a conversational AI tool and cut out the content writer entirely. However, they’re more likely to produce thought followership—the quick takes of those who just did a Google search and proclaimed themselves experts.
Chatbot copy will not top search engine results pages; it can’t even qualify for copyright under current law. At worst, it simulates expertise with fact-free copy that is more BS than PR, or plagiarizes its source material and puts its users in legal jeopardy.
To avoid falling into a galaxy-mind trap, AI should be in the hands of marketing professionals who can bring out the unique characteristics of their brands. For content strategists who already know their way around search queries, AI is search and spell check on steroids—a time-saver in researching a nuanced presentation. Its copycat content calls for rewrite from a communications pro with subject matter expertise and human perspective. Marketing AI cannot innovate. AI chat is not the end of storytelling but the start.
Social Media Conversations: AI Sentiment Analysis
For social media marketers, chatbots can identify the sentiment of posts, comments and messages and identify social media influencers. In ChatGPT, the Google chatbot Bard or a multisource AI tool such as Quora’s Poe, sentiment analysis might start by entering text with this prompt: “What is the sentiment conveyed in this text? Is it positive or negative?” In a social media tool such as Hootsuite Insights, AI can help trace the causes of a change in sentiment. Coca-Cola saw a 42% engagement boost using AI in analyzing metrics and conversations.
A simple way to learn about AI is just to ask the chatbot. For instance, you might ask, “How can I prompt ChatGPT to extract the aspect and direction of sentiment in a text?” ChatGPT will respond with a choice of detailed suggestions, such as “Prompt: What are the different sentiments expressed in this text, and in which direction are they leaning? Please provide specific examples of positive and negative directions.”
ChatGPT shared a detailed cheat sheet that reminds us of the instructions a new hire might need to perform a task. Spending more time on prompts may seem time-consuming, but the process is a reminder of how many factors go into well-crafted content. Conversational AI works much like search: The way a question is phrased makes all the difference in the results.
Conversational AI will help marketers personalize a customer service response (the bot is always polite), such as a follow-up message or survey after a purchase or interaction to gather feedback and address any concerns. Neurodiverse users have even used chatbots to rehearse social conversations. There are good reasons to be cautious about getting personal with a program that collects and shares information. But who knows where this kind of code-switching might lead?
Martech Meets AI: Software Startups
Dozens of new AI marketing technology products go to market every week. A Marketing AI Institute webinar presented 20 AI writing tools, and the daily AI newsletter Ben’s Bites often debuts new content production products. Among recent marketing AI software startups are Charlie and Copymatic, which write ad, blog and social media content; Copy.ai, Lavender and Rasa.io , geared to newsletters; SEO tools Demandwell, Frase and MarketMuse, and image and video generators GlossAI and Visla.
AI upgrades are in the works too for familiar research and writing essentials. Word, Excel and other Microsoft 365 tools are due for AI updates; GPT-juiced Bing search has made the Edge web browser a complement to the Purpose Brand content team’s favorite Chrome alternative, Opera.
Google has made more tentative AI moves, warning that Bard “may give inaccurate responses” or “offensive information that doesn’t represent Google’s views.” But Google’s adoption of snippets, maps and local results has set the pattern for AI integration—and a note of caution for businesses that rely on SEO. AI chat results eventually will make search engines even more of a self-contained tool and give fewer users a reason to click through to source material.
Rapid improvements are likely for conversational AI, but it won’t replace humans. Like Kazuo Ishiguro’s artificial friends in “Klara and the Sun,” future AI chatbots will run into trouble once they try to mimic empathy. When our research robots seem smarter than us, then we’ve forgotten how to tell a distinctive story and left our hearts out of our work.
