Artificial intelligence was front and centre at DMEXCO last week - WARC’s Brian Carruthers considers the problems and possibilities for an era of AI-enhanced creativity that remains difficult to imagine. 

“Ladies and gentlemen … Rory Gallagher!!” 

The words are indelibly seared on my memory from the first live concert I went to. There’s little reason you’d be interested in the soundtrack to my youth in Belfast some decades ago now, but bear with me while I sit in a Leverkusen hotel and explain.

Raised on a diet of guitar-based, blues-based music – in part because Rory was one of the few musicians to actually gig in the north of Ireland in those troubled times – I grew up with a particular musical world view. We musical tyros were disdainful of anything that didn’t fit with the narrow boundaries we’d set ourselves. 

But around the same time, just 35-odd kilometers up the road from where I now write (or the same distance between Belfast and where I actually grew up), Kraftwerk were developing their unique sound, which was very far from the guitars my teen self had thought crucial to any “real” music.

That sound was based on synthesisers – and not just any off-the-shelf synthesisers; the band’s Kling Klang Studio was filling up with self-made equipment as they took seriously their self-appointed role as “music workers”.

So what, you ask? Well, Kraftwerk are now acknowledged as trailblazers, influencing artists from David Bowie to Daft Punk. Back then, however, there were a lot of people like me who didn’t appreciate the idea of electronic music and, if we thought about it at all, worried about what it meant for our guitar heroes.

Some of us eventually broadened our horizons – these days I’m more likely to stream Steve Reich than reach for my Kraftwerk vinyl – while still retaining an affection for our formational listening. 

“Technology didn’t destroy music – it gave us music that was richer, more experimental and more varied,” Microsoft’s Gaurav Bhaya, reminded a DMEXCO audience. “It gave birth to new genres of music, like disco, techno and hip hop.”

And he drew an interesting comparison: “New tools powered by AI have the potential to do for businesses what electronic tools did for music”.

Remove the hype and the existential angst about a Skynet future and in this analogy AI is effectively at the same stage as the four original Kraftwerk members building their own studio equipment. There’s a whole lot more to come and we can’t be sure of all the directions it will take.

But already there’s a clear need to build in safeguards against biases. At DMEXCO, it was hard to avoid talk of AI, and while certain biases were noted – typically the visual elements, as when a “beautiful woman” prompt inevitably returns a white woman with long flowing hair and a “businessman” prompt inevitably returns a white man in a suit – the audio aspect was less remarked (if it was mentioned at all – I didn’t hear anyone do so). 

Earlier this year, at the Nudgestock conference, Steve Keller, sonic strategy director of Studio Resonate, SXM Media’s in-house creative agency, reminded us that AI assistants have been found to have an error rate that is much higher for voices of colour than for white voices. “It’s almost as if the technology is saying ‘I would understand you better if you could sound a little more white’,” he said. 

That’s problematic, to say the least, and one of the reasons Studio Resonate has launched the Stand for Sonic Diversity with the aim of “redefining the sound of the American voice”.

Keller is also working with others on research into how humans react to synthesised AI voices – that’s something that may become increasingly significant as more businesses utilise AI video like that offered by Synthesia, who worked with Lays on an AI activation that saw Lionel Messi deliver custom messages in multiple languages based on user inputs. 

It’ll be hard for Keller and other researchers to keep pace with developments, though. It took Kraftwerk four years to go from being formed to producing seminal works like Autobahn, but advances in AI are going to be measured in months and it’s questionable whether the necessary safeguards will be able to keep up. There’ll be regulation at some point but that’s even slower.

It’s enough to make you hanker for a slower age of more limited knowledge when you could while away the time arguing about who was the best blues guitarist. The answer is of course that Peter Green was better than Eric Clapton, while Rory was “the best guitarist you’ve never heard of”.