ld AI try everything? //
Artificial Intelligence can be a wonderful thing. It’s been around a while but it has only recently started to make a noticeable impact on our lives. Martech platforms, for example, are releasing tools that let users generate campaign ideas.
This raises an interesting question. Should an AI tool always write creative copy? There are some AI ethics questions but there are other questions that are worth asking.
Drums, ducks and dinosaurs
I used a platform for managing marketing campaigns recently. It has an AI tool that allows you to provide some keywords or prompts and it will generate content.
It uses the best practices of past campaigns from clients of the company to make its suggestions. While the suggestions weren’t spectacular, they were not terrible either. This makes it an excellent tool for overcoming writer’s blocks and brainstorming.
I was intrigued and decided to play around by entering words that had little to do with the situation. I was curious to know what kind of hotel campaigns could be generated by a prompt like “drums dinosaurs and ducks” and the tool suggested that:
- “Looking for a unique getaway? Discover our themed hotel with ducks, drums and dinosaurs! “Book now to experience an unforgettable experience.”
- Get ready for an epic journey! Discover the ideal destination for ducks, drums and dinosaurs. “Book your stay today and create unforgettable memories.”
The results, while I do appreciate the effort made, are not really useful.
The largely harmless but harmful
The tool’s suggestions are not always wrong. In some cases, even seemingly unrelated terms can be relevant (such as the Peabody Ducks located in Memphis, TN).
Ducks in the Peabody Hotel, Memphis. Accompanied by the Duckmaster.
Pet-friendly hotels are also important for customers who travel with their pets. While “ducks”, for example, may not be relevant the majority of the time, prompts such as “cats and dog” are.
Although such suggestions may seem harmless, wouldn’t it be better if the tool asked more questions about an apparently nonsensical question before offering suggestions or guidance? If “drums” was in the prompt, the tool could respond with “Check out our guide on targeting niche audiences.” This would provide greater value to the user.
Some prompts are not well-equipped
Consideration of data is also useful. Data in the hands of untrained people can lead to ill-informed conclusions. Untrained people can sometimes have disastrous outcomes when they are given data. There are many situations where someone who is not a data expert can make valid conclusions. However, this doesn’t always happen.
Giving an AI tool a set of instructions that it’s not equipped to handle could also have questionable outcomes. This is also something to consider. Asking an AI tool to perform a task it is not trained to handle is a bad idea.
Think before you act
In my career and in life, I have learned that it’s sometimes wise to step back and consider something, whether it’s a question, a task, or a fork in the path. It’s nice to get a quick response and cross something off a list. But a hasty answer or decision may not be the best one.
This is a good example. Although AI is able to “think” faster than mortals, it may be necessary to train AI systems to think beyond the obvious.
The example above shows that inserting random phrases in the sample campaign text may not be what the user needs. A link to a guide on how to target niche audiences is more helpful than just trying to cram anything in the sample campaign text.
Real intelligence
You can have fun letting AIs do their thing and see what happens. It is wise to use one’s real intelligence.
Does this task suit this AI tool well? Is it really necessary to have the AI tool spit out a result? It seems sensible to ask and answer these questions.
MarTech is here to help! Daily. Free. Free.
input name=”email”, aria-label=”Business Email Address” class=”inlineEmail form-control rounded 0 w 100″ id=”nl”-inlineEmail> placeholder=”Get MarTech into your Inbox.” required=”” type=”email”/>
The article Should AI try everything? first appeared on MarTech.