The effect of AI-based linguistically tailored health messages on plant-based lunch choices of adolescents: A randomised controlled trial.
Jonas A Schlicht, Thabo J van Woudenberg, David J Blok et al.
✦ AI-curated · Sources linked
In 30 seconds
This preregistered, two-week randomised controlled trial tested whether AI-based linguistic tailoring improved dietary choices among 587 adolescents (mean age 13.09 years) in the UK and Ireland. Participants received no messages, non-tailored messages, linguistically tailored messages, or a crossover of tailored and non-tailored messages. Adolescents receiving daily messages increasingly chose plant-based lunches, rising from 44% to 59% (OR = 1.06 per day; BF 10 = 35.39), while choices in the passive control group remained stable at 45-46%. Tailoring did not add short-term behavioural benefit, although exploratory analyses found improved message and source evaluations in week two.
Key findings
- Plant-based lunch choices among adolescents receiving daily messages rose from 44% to 59% (OR = 1.06 per day; BF 10 = 35.39).
- Choices remained stable in the passive control group (45-46%).
- Linguistic tailoring showed no additional short-term behavioural benefit compared with non-tailored messages.
- In exploratory week-two analyses, tailored messages received higher ratings for relevance (BF 10 = 3.50), personalisation (BF 10 = 4.93), and source liking (BF 10 = 3.38).
Why it matters
The findings suggest that repeated digital health messages may be associated with shifts in adolescents’ food choices, while adapting language to individual communication styles may improve perceived relevance and appeal without adding a demonstrated short-term behavioural effect. Distinguishing changes in message evaluations from changes in behaviour is important when assessing personalised communication strategies.
What to keep in mind
The trial lasted two weeks, so it does not establish whether behavioural changes persist. The reported advantages in message evaluations were exploratory.