Author: Phil Barnes, Research Director, Beyond Blue
The 90s called, they want their dial-up internet back
As a child of the early 90’s, I’m old enough to remember that horrible noise dial-up internet made, painstakingly rewinding VHS tapes, and waiting for teletext pages to cycle through just to see the sports scores. But living through my teenage years in the early 00’s made me an early adopter of the digital world. And I loved it.
I remember the first time a classmate brought an MP3 player to school, how impressed we all were as he told us it holds a gargantuan 30 songs(!).
Back when the future felt like a game of Snake on a Nokia 3310
The future seemed limitless with the internet, and the world seemed smaller all of a sudden; everything at your fingertips, society had come to me rather than me having to go out and find it. At the time, I couldn’t imagine anything better, but as we age, perspective has a nasty way of dawning on us.
When the insights world hit its ‘Windows 95’ moment
Generative AI has taken this even further; in my decade in the healthcare insights world, I’ve seen things shift, but none more so than in the last handful of years. Efficiency and innovation are the name of the game; the old ways are dying, and I can’t help but feel a little perspective, and hindsight might be just what we need.
The day everything changed
I want to take you back to June 1st, 1995, a seemingly innocuous date for most, but it likely changed your life in more ways than you know. The streaming site Napster launched and changed the music industry forever.
Before this date, digitalized music was hard to find, hard to create, and even harder to play; the concept of downloading an album didn’t exist for the average music listener. But in one day, all that changed, music was now free, searchable, fast, and social.
The audience was now the distributor, as opposed to the record companies. Artists could reach their audiences directly without the need for radio or MTV. Now, all of this sounds quite liberating, in a ‘taking back the power from the man’ type way, but what came next wasn’t necessarily all ‘peace and love’.
Napster didn’t just change our music listening habits; it changed data. The world moved to song-by-song consumption; nobody had time for albums, and the metrics followed suit. Album sales replaced by skip rates, replay patterns, popularity of specific sections of songs, etc – the data was overwhelming.
When music found its algorithm
As the consumption of music changed, unfortunately, so did the creative process. With this new data providing insight into listening patterns, music producers were bound to notice. Enter Max Martin, one of the quintessential producers of the late 90s and early 00’s, think Britney Spears, Backstreet Boys, NSYNC, and yes, more recently, Taylor Swift.
I’ll keep my thoughts on these recording artists to myself for the sake of brevity, but Max Martin was a key figure in their success. He introduced the ‘melodic algorithm’ that would go on to define 21st-century popular music – this wasn’t music anymore, it was mathematics.
What RPM rate do we need for a hit? What tension-release sequence would attract listeners? How long should an intro be before you get to the catchy hook?
What Max understood was simple and powerful: you’re not selling music locally anymore, you’re selling music to the world. And with that, we said goodbye to local specificity, the Beatles sounded like Liverpool, the Smiths like Manchester, the Beach Boys like California; there was no place for this anymore, music didn’t need to appeal to your hometown; it needed to appeal to the greatest number of listeners possible, that was now success.
Hits without a heartbeat
All of this so far, depending on your particular music taste, might sound quite positive. We’ve got some catchy new tunes to bring in the millennium and a globalized music industry that made listening to new songs easier than ever, but here’s the kicker: things didn’t exactly go the way the industry assumed they would. With algorithms replacing DJ taste and unlimited access removing the need for critics to tell you how to spend your money, many assumed this personalized service would thrive.
However, playlists flattened genres into a single, all-consuming flavour profile, and artists struggled to find identity in the algorithmic feeds; hits emerged as they always did, but culture didn’t follow as predictably.
Music used to represent or drive social movements – Marvin Gaye’s ‘What’s Going On’, Bob Dylan’s ‘Blowin’ in the Wind’, Public Enemy’s ‘Fight the Power’ –, you’re there with them fighting the fight; but now music and culture had become stuck in a loveless marriage, present but uncommunicative.
Arriving at the point: what AI risks us losing in the insight world
Ironically, for the person who spoke earlier about brevity, I’d finally like to arrive at my point. AI promises scale and efficiency in how we conduct Qualitative analysis, but an overreliance on machine learning data risks losing the interpretive human layer that makes insights relevant.
In the same way that the music industry finally faced the fact that streams don’t equate to culture, AI-generated insights are not necessarily indicative of human understanding. While a song might go viral due to meme-cycle randomness as opposed to emotional resonance, an AI-generated insight might identify sentiment but misinterpret cultural or situational nuance. Just because something is prominent doesn’t necessarily mean it is valuable.
Both industries have the capacity to confuse high-volume signals with high-value insights.
Beyond the beat: the parts AI can’t hear
Where AI in qualitative analysis can fall short is in its overreliance, not in its utility. Simply put; to overcome AI’s limitations, we must not forget the jobs only humans can do.
The main issues I see with AI overreliance are the following:
- Loss of narrative cohesion: just as playlists broke albums into isolated tracks, AI breaks human stories into tokenized fragments.
- Overreliance on surface patterns: streaming data predicts hooks, not emotional resonance. AI predicts word patterns, not intentions.
- The context gap: music data lacks the cultural backdrop, AI lacks tacit knowledge and lived experience.
- The importance of the minority: the next genre isn’t born in the playlists, it’s born in subcultures that algorithms don’t see. Breakthrough insights often emerge in the margins where AI has the least training data and where only a human would recognize its relevance.
The revival of human curatorship
In the music industry, things have bounced back; independent curators are on the rise. Community-driven discoveries have brought localization back to the market. Vinyl is back. And people are even going to concerts again.
The industry has found a way to evolve, weaving in the promise of a direct-to-consumer feel with artist-to-fan storytelling via social media, creating deeper connections. These steps have emphasized a particular key finding: meaning is human-made.
And therefore, for qualitative research, we need to value and promote the role of Moderators who understand the nuance, of Strategists who sense cultural moments, of Ethnographers who can pick up on emotional subtext, and of Analysts who interpret contradictions. Only then can we see the full picture.
Don’t auto-tune the insight out
Augmentation, not replacement. We’re not looking to replace AI, we just need to figure out the best way to use it before the Max Martins of the qualitative analysis world strip it of its soul (sorry, I couldn’t keep my personal opinions of those artists to myself any longer).
AI is a wonderful tool, use it for what it’s best at. Assisting with pattern recognition, not leading insight generation. Allow us humans to bring the cultural, emotional, and contextual interpretation, and what you will end up with is efficiency and scale, but depth, relevance, and meaning as well.
The music industry didn’t die, it transformed itself by recognizing what technology could and couldn’t do. And we’re ‘a quarter to midnight’ in the qualitative analysis world, the time is now to recognize the mistakes of the past and level with the fact that more data can’t substitute for human understanding.
Vive la AI revolution, but only if it’s a human leading the charge.