This has to be due to changing cultural attitudes of consumers, since the technology itself has improved significantly over the last 2 years
Key point: fewer artists admit to using AI, due to backlash
Classic Gartner hype cycle, though the peak of inflated expectations has been stratospheric. Now that people are using these tools and finding their limitations, they’re abandoning them in favor of proven technologies.
Title is missing a very important word “Say”.
The number is so high it makes me not want to listen to new music at all.
So, um… are they *good* artists? Or like… tons and tons of nobodies who don’t even try, have no talent, and are just trying to make a quick buck? And billionaire’s sons who think they’re rappers/hip-hop artists?
The cynical side of me thinks that AI has improved so much in those 2 years that fewer artists feel the need to admit to using it.
This doesn’t make sense. It’s either saying that their 2023 study was a bad representation of the population, or people are just forgetting they used AI in their music? Assuming the proportion of people sampled in 2025 has a relatively low percentage of new artists, it shouldn’t really be possible for the % who have *ever* used AI to fall so significantly.
The only other possible explanation is that people are now lying more than they were in 2023.
This is a nonsense survey.
1) The backlash against AI in the arts community will almost certainly have affected responses.
2) Using ChatGPT as a rhyming dictionary counts as AI, but in effect is no different to using any other rhyming dictionary, online or otherwise.
3) Two artists use two different plugins to clean up ambient noise on a badly-recorded vocal. One of those plugins uses AI, but probably not in a way we imagine it does. What’s the difference? And how many of us use effects, that have AI under the bonnet (hood) that we don’t know about?
4) A songwriter putting a demo together uses Logic’s session players – that’s using AI, but is it vastly different to using something from the loop library?
This survey proves nothing.
Remember, we’ve been having this debate since the Mellotron came along and string players were freaking out that they were going to lose work. And to be honest, they were right. We’re fucked; most incidental music we hear on TV will be auto-generated within the next five years, for example. We’re already in a situation where making a living as a composer/songwriter is almost exclusively the preserve of people from wealthy backgrounds; AI is just going to compound that. It’s a tool that rich people use to access skills without having to pay for it.
AI is not the problem: capitalism is.
People here are debating about the honesty of the answers of the survey, but every single survey has this problem.
People can lie about income, can lie about sexual orientation, can lie about political opinion, can lie about everything.
That’s why when we do surveys, we compare data in different times, to see the actual change of their responses in time, preserving all the biases and comparing the changes in normalized numbers (percentages, in this case)
The lie factor is accounted in surveys.
Don’t believe this stat. I know many indies and none use AI. Maybe 50% of the avalanche of crap on Spotify over the last year has been AI, but not verifiable human artists in real life.
Artists using DAWs were “using AI” long before this became the loaded discourse it is today.
It’s in fine print, but both survey participants and participant count are wildly different. Not sure why, if the goal is to be statistically accurate. 1299 in 2023 and 2500 in 2025 suggests they used ALL participants in 2023 but may have artificially cut off participants in 2025 (the cutoff on the google doc definitely seems like this is the case). 60% of 1299 is 779 while 48% of 2500 is 1200.
Data Analysis:
South Africa, UK and US had the highest responses with 25.2% (up from 18.2%), 18.6% (down from 20.1%), and 13% (down from 14.6%) respectively. Nigeria and India followed up with 7% (negligible change) and 4% (down from 5.7%).
Doing analysis on the recorded responses from specific countries on T/F AI usage: Quite a few single responses from countries in 2023 are simply not represented in the 2025 data. Notable ones include Cyprus, Iraq, Georgia (not the state), Croatia. Many respondents that responded that they used AI in 2023 didn’t even fill out the survey in 2025 as their countries simply have no data for 2025.
The apparent 12% decrease in AI is explained by the vast majority of countries having decreasing AI usage, but the # of responses collected for many countries is so small it’s not statistically significant. 1 Peruvian response said they used AI in 2023, while 0 did in 2025. But max 3 people responded in both years combined so even though it looks like less people use AI than before it’s not enough data to be an accurate perspective.
**TLDR**: AI usage in US/most of Europe has gone up, while in many African/South American countries as well as Spain, India, and Germany has gone down.
^(Made a whole spreadsheet to analyze this. If enough people want it, I’ll share it publicly)
![[OC] 48% of Artists Use AI to Make Music - Fewer Than in 2023](https://www.europesays.com/wp-content/uploads/2025/05/gk2k3vswvw1f1-1920x1024.jpeg)
13 comments
This has to be due to changing cultural attitudes of consumers, since the technology itself has improved significantly over the last 2 years
Key point: fewer artists admit to using AI, due to backlash
Classic Gartner hype cycle, though the peak of inflated expectations has been stratospheric. Now that people are using these tools and finding their limitations, they’re abandoning them in favor of proven technologies.
Title is missing a very important word “Say”.
The number is so high it makes me not want to listen to new music at all.
So, um… are they *good* artists? Or like… tons and tons of nobodies who don’t even try, have no talent, and are just trying to make a quick buck? And billionaire’s sons who think they’re rappers/hip-hop artists?
The cynical side of me thinks that AI has improved so much in those 2 years that fewer artists feel the need to admit to using it.
This doesn’t make sense. It’s either saying that their 2023 study was a bad representation of the population, or people are just forgetting they used AI in their music? Assuming the proportion of people sampled in 2025 has a relatively low percentage of new artists, it shouldn’t really be possible for the % who have *ever* used AI to fall so significantly.
The only other possible explanation is that people are now lying more than they were in 2023.
This is a nonsense survey.
1) The backlash against AI in the arts community will almost certainly have affected responses.
2) Using ChatGPT as a rhyming dictionary counts as AI, but in effect is no different to using any other rhyming dictionary, online or otherwise.
3) Two artists use two different plugins to clean up ambient noise on a badly-recorded vocal. One of those plugins uses AI, but probably not in a way we imagine it does. What’s the difference? And how many of us use effects, that have AI under the bonnet (hood) that we don’t know about?
4) A songwriter putting a demo together uses Logic’s session players – that’s using AI, but is it vastly different to using something from the loop library?
This survey proves nothing.
Remember, we’ve been having this debate since the Mellotron came along and string players were freaking out that they were going to lose work. And to be honest, they were right. We’re fucked; most incidental music we hear on TV will be auto-generated within the next five years, for example. We’re already in a situation where making a living as a composer/songwriter is almost exclusively the preserve of people from wealthy backgrounds; AI is just going to compound that. It’s a tool that rich people use to access skills without having to pay for it.
AI is not the problem: capitalism is.
People here are debating about the honesty of the answers of the survey, but every single survey has this problem.
People can lie about income, can lie about sexual orientation, can lie about political opinion, can lie about everything.
That’s why when we do surveys, we compare data in different times, to see the actual change of their responses in time, preserving all the biases and comparing the changes in normalized numbers (percentages, in this case)
The lie factor is accounted in surveys.
Don’t believe this stat. I know many indies and none use AI. Maybe 50% of the avalanche of crap on Spotify over the last year has been AI, but not verifiable human artists in real life.
Artists using DAWs were “using AI” long before this became the loaded discourse it is today.
It’s in fine print, but both survey participants and participant count are wildly different. Not sure why, if the goal is to be statistically accurate. 1299 in 2023 and 2500 in 2025 suggests they used ALL participants in 2023 but may have artificially cut off participants in 2025 (the cutoff on the google doc definitely seems like this is the case). 60% of 1299 is 779 while 48% of 2500 is 1200.
Data Analysis:
South Africa, UK and US had the highest responses with 25.2% (up from 18.2%), 18.6% (down from 20.1%), and 13% (down from 14.6%) respectively. Nigeria and India followed up with 7% (negligible change) and 4% (down from 5.7%).
Doing analysis on the recorded responses from specific countries on T/F AI usage: Quite a few single responses from countries in 2023 are simply not represented in the 2025 data. Notable ones include Cyprus, Iraq, Georgia (not the state), Croatia. Many respondents that responded that they used AI in 2023 didn’t even fill out the survey in 2025 as their countries simply have no data for 2025.
The apparent 12% decrease in AI is explained by the vast majority of countries having decreasing AI usage, but the # of responses collected for many countries is so small it’s not statistically significant. 1 Peruvian response said they used AI in 2023, while 0 did in 2025. But max 3 people responded in both years combined so even though it looks like less people use AI than before it’s not enough data to be an accurate perspective.
**TLDR**: AI usage in US/most of Europe has gone up, while in many African/South American countries as well as Spain, India, and Germany has gone down.
^(Made a whole spreadsheet to analyze this. If enough people want it, I’ll share it publicly)
Comments are closed.