Beyond perfect


“Perfect is the enemy of the good,” attributed to Voltaire, is a bit of wisdom about diminishing returns. Seeking perfection by improving something that is adequately good for the task at hand is a fool’s goal. We can all identify with losing hours, weeks, or years in pursuit of some imagined but elusive standard of perfection, struggling to embellish an already-satisfactory thing so it might better match an imagined but nonexistent ideal version.

The temptation to want to achieve perfection shows the well-intentioned but delusional tendency, often seen in firstborns and Virgos, to value ideals over practical effectiveness. Scaling human interaction with AI is no more human, and only slightly more sophisticated, than writing a form letter. As with any new communication technology, the first uses of AI invariably involve pornography (such as deepfakes) and grift (Amazon trash books).

But if it’s free, it’s good enough, right?

The counterbalance to “perfect is the enemy of the good” is “cheap is the enemy of effective” — a warning in the age of AI. Our senses are now swimming in trash culture — a Noah-scale flood of trash books, trash images, trash news, trash music, and trash marketing. Each of these is being generated in cynically large amounts by unscrupulous actors wielding AI tools, all dedicated to steering your attention in the direction of whatever money it makes its (non-)creators.

Today nearly everything that can been seen or said cannot be believed at first glance. Therefore, your identity, as someone or some entity who wants to communicate, is now in jeopardy. Trust in institutions has always been a necessary component of business and commerce, but its essential importance is growing daily. 

Where do creators fit into this world? When word processors were widely introduced in the 1980s, everyone with a copy of WordPerfect or related software thought they were now writers. As it turns out, the ability the generate text is not the same as ability to generate interest, or to educate, or to persuade and motivate. Although it is true that AI may be able to assist in some of these activities, it cannot do so without adult supervision.

AI is capable of so many kinds of errors that generally do not apply to human writers, including understanding, context, creativity, and emotional intelligence. Any one of these is enough to make the average reader raise suspicions and dismiss the text as somehow bogus, although persuasiveness* scores seem to be improving somewhat (persuasiveness itself being a nebulously subjective and therefore all-too-human metric).

“Further research and responsible deployment practices will be required to mitigate the potential risks of rapidly advancing, and increasingly persuasive, language models,” stated in the research cited above*, implies the potential risks of generative AI to readers, while I am referring to the potential risks to marketers, who have a vested interest in producing responsible persuasive text. 

The risks, the limitations, and the potentially fatal pitfalls of unfettered AI in marketing are still too numerous for wide deployment. And everyone (except an AI) knows that easily avoidable risks are simply too expensive to take. Until AI improves by a several orders of magnitude, human copywriters are essential for risk management. 

Skepticism on the part of the audience rapidly converts to full-blown apathy. Using AI text in business contexts, unless offered by someone with writing expertise, is an invitation to danger. Are there good uses for AI text in business contexts? Yes, and I’ll talk about some of those next week.


  • AI is a miraculous tool that requires a strong editorial hand to guide it toward suitable results.

  • Its scope is severely limited — AI has no soul, intuition, morality or vision.

  • Only humans are capable of the nuances required to navigate multiple enterprise strategies, distinctions among intermediaries and diverse audiences, and potential conflicts of interest in routine business communication at the highest levels.

Curious? Contact me. Find me on LinkedIn.


Coming next week: Ways to use generative AI text in marketing.


You may find of interest:

Q & AI

Claude vs. ChatGPT


Image: Voltaire by Jean-Antoine Houdon, 1778. National Gallery of Art.