Can AI solve B2B marketing problems? Of course. The real question is how to get the most out of AI tools.
When used properly, AI tools offer enormous advantages. The trick is inserting these tools into key parts of an overall content workflow that is guided by people who know their subject matter well. The other trick is NOT to treat generative AI tools, like ChatGPT, as a friend, but rather as a tool.
Most people are familiar with the concept of AI hallucinations, ie, AI tools generating fictional answers to questions when the AI tool cannot find online material to answer a question.
Yet another AI trait to be wary of is sycophancy, which in terms of AI refers to a chatbot adapting responses to align with what a user wants rather than seeking truthful answers. As people continue to treat chatbots as friends, partners, or therapists, it is unlikely that sycophancy will simply disappear.
But for B2B marketing, AI presents some interesting opportunities.
Research with AI:
Research is the most obvious possibility, but also the most prone to error. This is the area in which sycophancy is most likely to occur. Most AI tools will provide links where they can in their answers, but not always.
When you want to discover media sentiment around a product, understand the pain points of a particular industry, or even to develop a customer profile, AI tools like Perplexity can help discover trends or popularly expressed views without having to spend hours digging through documents online.
However, open ended questions that follow from existing chats can colour the response from an AI tool. Switch off the AI and restart or begin a new chat. Do not provide any leading information or context, unless you can provide specific material that you want the AI to consider while formulating an answer. Keep your information requests on point.
Another idea is to begin by finding and verifying all the information sources you want your AI LLM to use when formulating answers. For example, are there industry associations that publish a lot of online information that you can access? With tools that support live browsing or document upload, you can direct the AI to authoritative sources to better inform its responses.
In short, treat the AI you choose to work with not as a partner, but as the hammer in your hand.
Writing and editing with AI:
It is all too easy to just let ChatGPT do it. Why bother struggling with drafting a headline or a blog or strapline when ChatGPT can do it for you?
ChatGPT can write and edit, but marketers and copywriters need to participate. Well-constructed prompts that logically set the parameters and intended results are important. But so is a background or knowledge of the field in question. Relying on ChatGPT (or any other such AI tool) solely to write something may result in posts or headers that don’t make sense to industry insiders.
I prefer to think of myself as the Editor when crafting content using an AI tool. Let it do the draft or switch it up – you do the draft and the AI tool is the editor.
If other people on a team are involved in the process, they may also have their own AI generated results. If so, compare the prompts as well as the results. Not all results may be equal and the prompts will give a better idea of whose results to pay more attention to.
This topic is evolving rapidly as AI tools evolve and as human users evolve as well. The ideas here are just the beginning of how to incorporate AI tools into content marketing workflows.
Leave a Reply