Creative Content and Sales in the Age of AI

We have a Slack channel for Maestro Group that tracks “ai-in-sales.” Most weeks, someone contributes a link to a resource that promises how artificial intelligence can help us—build your webpage! Leverage LinkedIn! Summarize your Zoom call! The tone is generally upbeat, suggesting that AI can help us do our jobs better, faster.

Meanwhile, my social media feed is streamed from a whole other planet, one on which AI is a quasi-apocalyptic influence worthy of bans and open letters. Teachers worry about catching ChatGPT-authored work in their comp classes. Artists beg us not to feed the machine by uploading screenshots of their illustrations to DALL·E. A science-fiction writer is bullied because her cover reveal led to accusations of using an AI-shaped design. Poets fuss when a book contest winner centers his project on AI-manipulated text. 

I’m a pragmatist when it comes to technological tools (thanks, Clippy), and AI is not that new. What’s creating impact in the culture now is that AI is more sophisticated, more “human,” than ever before and therefore, to some, more threatening. We don’t know, yet, how that will affect a dynamic where the most intriguing creative collaborations between human and machine, to date, have been explicit in the valuation of strange, inorganic, or uncanny moments. (Case in point to how using AI can be done right: Vauhini Vara’s essay, “Ghosts,” first published in The Believer and later anthologized in Best American Essays 2022.)

This isn’t the space for hashing out the core ethics of digitalizing faces, images, voices, or styles. That’s a complex territory for businesses squarely in studio and performing arts—literary publishing, music, stage and screen, et al—and their unions. But if you have even a moderately-sized staff, or if you budget for freelance or work-for-hire, you’re in dialogue with artist-creators who are grappling with these issues. What’s keeping them up at night?

THE WORST VENTRILOQUISM

On the writing front, low-hanging fruit for using AI at your company may include generating first drafts of correspondence, speeches, and sales copy. Note the significance of “first drafts”—editing for authenticity and accuracy is key, especially since ChatGPT 4.0 only reflects data from 2021 and prior. But otherwise, why not? Generating the prompts can be fun. 

“Write a speech to inspire someone to make more sales,” I asked ChatGPT.

And the results were…fine? The text used classic techniques of active syntax, parallel structure, and hints of figurative language, though no conceits—that is, sustained metaphors or similes—which strike me as requiring a significant (and distinctly human) level of insight. Of course, the field of sales or specifics of any company was unclear. The content was vague, filled with what Douglas Hofstadter called “generic platitudes and fluffy handwaving,” in a July 2023 think-piece for The Atlantic on AI efforts to imitate his writing style when generating a faux-introduction to his seminal 1979 book, Gödel, Escher, Bach: an Eternal Golden Braid.

“As a woman speaker, write a speech to inspire someone to make more sales,” I asked ChatGPT, just for the fun of it. 

“Today, I stand before you as a proud woman, not just as a speaker, but as a trailblazer in the world of sales,” this version of an otherwise similar speech began. “I am here to break barriers, shatter glass ceilings, and inspire you to reach new heights in your sales journey.” 

As we found with “Shalom from the mountains,” identity-led prompts are not the way to go in this interface. This experiment got me wondering, where does ChatPGT pull its language models from in 2023? One of the primary resources, it turns out, is a mass repository of emails from Enron, an American energy and commodities company that was headquartered in Houston, Texas. Enron ceased operations in 2007 in tandem with federal investigation, which is what resulted in the posting of their emails online. 

Another source is books, presumably, but which books? Here, OpenAI is willfully oblique in its sourcing. An early document simply references “Books1” and “Books2,” containing approximately 63,000 titles and 294,000 titles respectively. 

Perhaps OpenAI only used books legitimately available in public domain, such as those found on Project Gutenberg’s website, 70K and counting. If so, that presents the problem of who was privileged as authors in the era of 1923 and prior: white men. (Also, Texas 2007: white men.)

Even if that’s the case, where did those other 294K titles come from? No official digital repository has that size of archive. If OpenAI sourced contemporary books via a torrent site (an illicit site specifically designed to distribute a large cache of files, usually pirated, to multiple users), the good news is that they got a wider variety of voices in terms of gender, race, and other social values. But they probably did so in flagrant non-observance of author copyright, which does not bode well for their relationship with authors going forward. 

THE FILTERED HORIZON

Although it’s amazing when someone trained for art is legitimately employed in-house for art, what’s more common is having someone who is just “good with that stuff,” meaning the person you have check image embedding, resolution, cropping, and contrast. Maybe you have a go-to hire for higher-level design, such as rebranding and changing your logo, or as part of proof-of-concept for a new product. Maybe you have a friend with a camera who you throw some freelance money, in return for them showing up to a company event and snapping photos for your website. 

AI technology can affect how you interact with all these artistic resources, and the people in question know it. In some respects, this goes back to the breakthrough of Adobe Photoshop’s filter technology, which was priced for marketplace consumers to play with what had previously been advanced techniques among photographers. Photoshop has more recently been updated to include a number of AI-aided techniques, including significant options to erase and fill. 

I remember hearing film-based photographers bemoan how digital cameras could be the death of their industry. They adapted, just as professional writers adapted to the convenience of spell-check and losing other “easy” signifiers that separated pros from amateurs. Part of what photographers realized is that no technology could replace the value of their facilitating presence on-scene, e.g., recognizing the pathos of a shot that no automatic lens could spot. Or saying the right thing, at the right time, to make someone smile in an impromptu portrait. 

Now, sales spaces are flooded with alternate options to an outside or in-house artist. We can circumvent stock photography with a handful of keywords or a sample image. We’re invited to upload seven or eight snapshots to generate a professional-quality headshot with a background of our choice. You can design a logo with eight to 10 clicks versus an hour-long conversation. Why go back? 

PEOPLE, NOT PIXELS

What I realized in the course of writing this blogpost, is that I’m protective of people who, like me, are creatives who are agnostic about the advent of AI technologies into everyday practice. We’re not freaking out. We might even be in favor of certain conveniences or starting points for edits. Yet we don’t want to be backended into compliance, either. 

So please, be considerate in your attitude toward AI in the workplace. Don’t presume that everyone is ready to embrace it in the same ways you do, and be careful about moves that cannot be reversed. Three best practices:

• Give clear, advanced notice—subject to approval—before entering someone’s intellectual property into an AI interface, even as an “intermediate” step for brainstorming or revising. 
• Don’t underestimate the social value of in-person meetings for creating branding materials such as logos, documentation of employee culture, or individual staff photographs. Collaborating builds culture. 
• Count fingers. And teeth. And abs. All these values get weird, fast, in the AI landscape. 

Technology may be able to replace human competency, but it can’t replace human creativity (or heart), and within that will always lie opportunity. We just have to hold open space for it. 

AI can’t hold a candle to our light. Reach out at mastery@maestrogroup.co for more information on training, assessments, and more.