Human Ingenuity versus Deepfake

Recently, I attended a live performance of Wolfgang Amadeus Mozart’s Requiem, an experience I’m not likely to forget. It was exhilarating and intoxicating, the combined talent and cohesion of a full symphony orchestra, two choirs, and four soloists.

When I got home, I read the notes in the program, and discovered that Mozart died before he completed the requiem. The piece was completed by three other musicians: Mozart’s favorite student F.J. Freyst?ydtler, close friend Joseph Eybler, and Austrian composer Franz Süssmayr. The latter claimed he wrote the last few passages of the requiem himself; although, musical scholars doubt this fact since Süssmayr other pieces didn’t match the brilliance of Mozart’s other compositions.

Masking fake for genuine

If Mozart had lived today, and suddenly died, would his widow, Constanze, have considered using generative artificial intelligence (AI) to complete the requiem? And would the outcome be a fake?

This question lingered in my mind as I also wrestled with a word, recently added to my vernacular, deepfake. When I first saw the term, I wondered if it was an oxymoron or simply an imitation Picasso sequestered in a basement.

No. Deepfake is the use of AI to make bogus content by replacing an existing image or soundtrack with someone else’s likeness, voice, or words. The outcome is nearly impossible to distinguish from reality, putting politicians, celebrities, and everyday people in places and situations where they’ve never been, and saying what they’ve never spoken.

“I cannot write in verse, for I am no poet. I cannot arrange the parts of speech with such art as to produce effects of light and shade, for I am no painter. Even by signs and gestures I cannot express my thoughts and feelings, for I am no dancer. But I can do so by means of sounds, for I am a musician.” ?Wolfgang Amadeus Mozart

With generative AI now able to create text, write poetry, compose music, paint and draw, and respond to prompts in a way that is indistinguishable from that produced by a human, how is it not also deepfake? Yes, it’s not “replacing” one image or sound for another, but unless you explicitly say that the resulting creation was produced by AI, is it not an imitation?

Afterall, it’s common practice to credit reporting, publications, photography, videos, inventions, and other intellectual and scholarly works and assets to a person, organization, or company. These practices encompass citations, patents, copyrights, and trademarks.

Pretending to be something other than myself

Just for fun, I used Bing chat to re-apply for a job with a company that continually advertises the same openings, which never seem to be filled or more likely, they haven’t discovered the perfect candidate. Their application process consists of writing blurbs about one’s expertise and marketing savoir-faire.

In the past, I’ve written insightful responses that touched on my successes, delving into the many avenues and approaches that I’ve used to generate awareness, drive engagement, and turn casual prospects into delighted customers. ?

This time, I enlisted Bing chat to help me complete the application. It dutifully spit out a word salad for each prompt that I lightly edited for continuity to make it sound as if the words were streaming from my typing fingers. I then pasted the responses into their application form.

Instantly, as expected, I got a thank you for applying, and the milquetoast assurance that “a real human reviews every application,” but not to expect additional communications [unless their application tracking system (ATS) fortuitously matches my skills to that of their ideal candidate].

While I expect this application, like those past, to be spurn by their ATS, it brings up an interesting question: Given the ability of AI-powered chat to churn out remarkably lucid, insightful, anthropological responses will it eventually become a preferred replacement for human endeavors? ?And if replacing someone’s image for another is consider deepfake, does this deception extend to written, composed, drawn, painted, coded, engineered, and architected content?

Claiming authorship

The penalty for taking credit for one’s work – whether produced by a human mind or AI engine – depends on its intellectual value and relevance. A student who copies CliffsNotes isn’t as injurious as someone selling counterfeit Andy Warhol and Roy Lichtenstein prints.[1] In the grand scheme, with Mozart having composed fifty million bars of music in his thirty-five-year life, does it matter that three of Mozart’s violin concertos may have been written by his sister Maria Anna Mozart?[2] ??

“I am never happier than when I have something to compose, for that, after all, is my sole delight and passion” Wolfgang Amadeus Mozart

There’s a third element, which is the glue that binds successful societies. It’s integrity. Without truthfulness there is chaos because there’s no way to know what is authentic and what is emulated, exaggerated, or conjured.

Undoubtedly, AI has the potential to solve humanity's toughest challenges, but we must realize the need to distinguish and “label” what’s produced by human ingenuity, and what’s a product of AI.

Thank you to Radek Grzybowski for his amazing picture on Unsplash

?

?

[1] Palm Beach art dealer pleads guilty to selling counterfeit Andy Warhol prints, Adam Schrader, U.S. News, February 24, 2023

[2] An Australian professor has found evident that one of the most famous composers to ever live, claimed credit for some of his older sisters compositions, Sophia Alexandra Hall, Classic fM, January 10, 2022