We all know that AI can write a short story but can it judge a short story competition?
This was what we set out to discover with our 2026 summer competition.
Members were asked to write a short story (between 2,500 and 3,000 words) on the theme of ‘Crossroads’.
The same brief was given to Claude AI.
Judged by ChatGPT on six criteria
All 11 entries were then given to ChatGPT with a brief to judge each entry on 6 specific criteria: originality and imagination, characterisation, plot and pacing, quality of writing, effective use of the theme, and overall impact.
Each story was first assessed on its own merits, given written feedback and a score based on the 6 criteria, then compared before the placings were decided.
Three of the entries were ‘disqualified’ for being under the minimum word count – but still received marks and commentary.
Claude AI’s entry not eligible to win
One of those entries not eligible to win was Claude AI’s entry ‘Four Ways’.
Beth Rogers reading her winning entry, ‘Be Careful What You Wish For’.
Andy Barrett, Beth Rogers, Michael Ranes, Lorraine Forrest-Turner and Sonya Weiss
However, ChatGPT awarded it the highest marks, along with human writer Beth Rogers’ story, ‘Be Careful What You Wish For’ in first place.
Had Claude AI been able to do an accurate word count, Beth and Claude would have tied.
Interestingly, second and third place entries, Andy Barrett’s ‘Crossroads’ and Lorraine Forrest-Turner’s ‘Best Served Cold’, also received equal marks.
ChatGPT awarded second place to ‘Crossroads’ because its double narrative was more completely constructed around the competition theme.
Just two marks behind second and third place was the Highly Commended story, ‘I Did Not See A Butterfly’ by Sonya Weiss.
The members decide
All the entries, results and comments have been sent to all members and next week (9 November) we’ll have a lively discussion about the competition.
Do we agree with ChatGPT’s findings? Did it judge the competition fairly? And does Claude’s ‘Four Ways’ deserve the highest score?