Become a TechRadar Insider
- 39% of people change how they write so it doesn’t sound like AI
- Around one-third say they’d stop supporting colleagues or creators due to undisclosed AI use
- People still think it’s acceptable to use AI in brainstorming, research and editing
A major study across the US, UK, EU and Latin America has revealed the impact AI is having on consumers both in terms of how they produce content and how they judge authenticity – something sounding like AI has now become a social stigma, a new report from Use.AI reveals.
Although AI promises to enhance productivity by automating some inefficient administrative workloads, nearly three-fifths (58%) of the more than 12,600 people surveyed said they’ve seen someone criticized online or in the workplace for using AI.
It’s reached the point that nearly half (46%) now worry their own writing could be mistaken for AI-generated content, with 39% changing how they write specifically not to sound like AI.
How to spot AI-generated content
Perfection to the point of not sounding human is generally how people might characterize AI-generated content – this includes excellent grammar, predictable transitions and text that’s emotionally neutral.
But workers realize that, because many now edit AI’s output to make it sound more human by cutting sentences shorter, adding small imperfections and removing the long dashes that AI tools still seem obsessed with using.
“Creative workers face the sharpest version of the problem,” Use.AI warns, noting that their flawless work can actually be seen as a negative, because of its connotations with being AI-generated even when it’s not.
And it doesn’t stop at judgement – around a third say they’d think less of a colleague, creator or classmate if AI had been used without full disclosure (35%) and that they’d be less likely to support a creator as a result (34%).
“AI can be used to flood platforms with cheap content,” the report argues, highlighting the importance of authenticity, education and disclosure in good journalism.
The report also points to social media platforms like LinkedIn. The platform’s existing style of sharp opening lines, short paragraphs, neat career lessons, humble authority and controlled vulnerability often gets confused with AI-generated posts.
Where can AI be used?
While generating complete work is mostly frowned upon, consumers are still supportive of AI in the earlier stages of work. For example, three in five (62%) say using AI for editing, brainstorming and research should simply be viewed as part of modern digital literacy.
However, this newly revealed, widespread distrust and dissatisfaction with AI comes at a cost:
“Use the tool, but leave no fingerprints. Be efficient, but not suspiciously efficient. Write clearly, but not too cleanly. Know things, but not in a way that sounds assembled.”
Use.AI isn’t worried about AI making weak work look competent so much as it’s worried truly skilled individuals will start reducing the quality of their work to appear ‘less AI’.
But more importantly, this study broadly aligns with a growing number of others despite reaching a similar conclusion from a different starting point. AI delivers on its promise to boost efficiency by getting creators access to information or providing them with polished content in as little as seconds, but then much of that time gets negated as the people involved go behind it to refine output – in this case, to make it more human-sounding.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
