Substack survey shows deep creator divides on the subject of generative AI

News flash: Generative AI is a polarizing topic among creators.

That may seem like an obvious point to anyone who keeps tabs on tech industry trends, but new research published by Substack shows just how wide the chasm is. The platform took the temperature of more than 2,000 newsletter publishers and found that cohort to be deeply divided regarding the utility, ethics, and pitfalls of generative AI models.

One of the simplest questions in the Substack survey produced a telling result. 45% of respondents said they’re not using generative AI, while 52% claimed to employ it. (2% of publishers were unsure if they were using it.)

The percentage of respondents who use AI could be higher than initially reported, since some creators rely on systems that quietly employ artificial intelligence. “That ambiguity makes it harder to pin down how publishers think about and engage with AI,” reads a blog post accompanying the survey. “While the public conversation often centers on content generation and its implications for art and authenticity, publishers on Substack are often using AI tools in more varied and nuanced ways.”

If we take the reported percentages at face value, some telling patterns emerge. Publishers over the age of 45 are more likely to use AI than their younger counterparts; 51% of respondents in that older group 

said they use AI, compared to about 38% of the under-45s. The percentage of men who reported using AI was 17% higher than the equivalent figure for women (55% to 38%). And AI usage is far more common in fields like business, tech, and finance than in more artistic genres like music, literature, and visual art.

Substack’s findings are a reminder of genAI’s contentious status in the creator world (just ask MrBeast.) Some digital creatives are inspired by the open-ended possibilities of AI-generated material and willing to share their data as training material in exchange for financial compensation. More pessimistic sorts are troubled by the eerie realism of the latest AI-generated videos and the litany of ownership issues that come with unauthorized AI training practices.

Can those two disparate factions be united? Some pro-AI members of Substack’s survey group argued that large language models can empower people who deal with disabilities like ADHD, dyslexia, and blindness. If generative AI is here to stay — and with the way the tech industry is going, that seems like a likely outcome — then detractors may have no choice but to find use cases that serve their own needs.