The tagline was catchy. The premise was all wrong.
In 1971, Memorex released a television ad challenging the viewer to guess whether singer Ella Fitzgerald was singing live or if her voice was playing over an audio recording. The ad asked, “Is it live, or is it Memorex?“.
First, for those watching TV from our middle-American homes in 1971, BOTH versions coming through the TV were recorded. Ella’s glass-shattering ability was never real to us, no matter what Memorex wanted us to believe.
And second, what did it matter to the viewers if Ella was standing in a studio belting out a high note, or if it was a replica saved on a thin strip of plastic, coated with magnetic particles? For those of us sitting in our cigarette-stained, wood paneled living rooms, on avocado green couches, burnt orange over-stuffed recliners, and mustard yellow bean bag chairs, that bold voice became sharp and shrill when emitted through a poorly built, scratchy, monotone speaker. Her thin, tinny voice died a painful death as it slowly absorbed into the harvest gold shag carpeting, the black velvet painting of Elvis hanging above dad’s chair, and the single-pane window darkened by heavy polyester curtains. The only bright spot in the room was the glow of the TV and the lava lamp on the side table.
While the television speaker could be blamed for the inability to tell the difference between truth and fiction from our homes, the reality was that Memorex audio cassettes weren’t that good. The market was flooded with better options and higher quality; however, Memorex convinced us with a single ad campaign.
Close to real was good enough for us.
Fifty-five years later, we continued to resign ourselves to blur the line between reality and fantasy. We don’t even ask the question anymore.
We are foolishly fooled by the fake.
We simply settle for the substitute.
Memorex is replaced with AI and we can’t tell the difference between the real or artificially generated version. Recent studies expose the extent of our failure to distinguish between the two. In one study, only 64% of those tested could tell which image of a face was real and which had been generated. That number dropped to 59% if the image was of nature. As the authors note, “No better than a coin toss.”
In another, participants struggled to discern between an AI created voice and the real version: “Strikingly, on average, for 41% of Generic AI-generated Voices trials and 58% of Voice Clones trials, listeners labelled these AI-generated voices as being human… These data thus underline how convincing Generic AI-generated Voices can already be…”
Perhaps most concerning is a study published last month revealing that of the 9,000 listeners tested, a shocking 97% (8,730) were unable to tell the difference between music recorded by musicians and AI generated tracks. The Deezer/IPSOS survey was an extensive dive, covering eight different countries and exposing a scary truth about the direction of AI. It highlighted the importance of transparency when it comes to the industry’s labeling of AI generated products. If the article is to be believed (and at this point, who can believe anything), Deezer, a “global music experience platform” reports 34% of all music uploaded to the site every day is AI generated. That’s 50,000 new tracks a day!
While our technology has advanced from from analog to digital, from cassette tapes to microchips, our ability to distinguish between the genuine article and a cheap imitation has failed to keep the same pace.
The implications for this recognition gap are staggering. Deep fakes drive our perspectives on everything from entertainment to politics. They feed conspiracy theories and foster social discord. They replace artistic creativity with digitized imitation. We are easily convinced and we like it that way.
The technology, once designed to make life better, easier, and more complete, may in fact be fooling us with its falsehood into an unrecognizable and tragic reality.
