The majority of UK consumers are unable to distinguish an AI-generated image of food from one taken by a professional photographer, a small study has found.
On average, across the 10 different food images shown to participants, 60% couldn’t tell if it was a genuine photograph or one created from a prompt to OpenAI’s DALL-E image generator.
The finding is significant for online grocers, DTC operations and hospitality businesses, said DTC tech company Slerp – which commissioned the study – given its previous research had found half of consumers were more likely to order food online ‘when they can see what it looks like’.
“We know the public generally likes seeing food before they order online, so if a restaurant has the time and budget for a professional photo shoot that can truly reflect their menu, then we would always say to go for it,” said JP Then, founder of Slerp.
“But this experiment shows that for most people, when looking at pictures online, the AI version of the pizza is as appealing as the real thing.”
The study saw UK food photographer David Robson take photos of prepared dishes at Eataly London, and AI-generated images of the same dishes created with DALL-E, before they were shown to 100 consumers.
Nearly three-quarters (73%) could not recognise an AI version of a margherita pizza and two-thirds failed to spot an AI croissant or bowl of cacio e pepe pasta (66% and 69% respectively).
“For those brands and companies constrained by resources, timings or budgets, we would say as long as the images still look good enough to eat and reflect the real item being offered, then why not try it?” Then added.
Can you tell the AI tiramisu from the real one?
Generative AI – a type of artificial intelligence that generates new and original content, using techniques like neural networks and deep learning – is already being experimented with in the grocery sector.
The technology can output conversational text, images, video, computer code, music and more.
Earlier this year, Coca-Cola invited digital artists to generate DALL-E AI artworks based on its brand assets, the best of which will be displayed on digital billboards at London’s Piccadilly Circus.
Carrefour meanwhile is trialling a computer-generated human avatar which responds to customer questions such as how they can purchase healthier, cheaper food via the grocer’s website with ChatGPT-generated answers.
Deliveroo earlier this month ran a generative AI-focused hackathon, which saw its developers present use cases for the technology. Among them was a chatbot which can respond to employee queries.
Speaking to The Grocer earlier this month, Deliveroo CEO Will Shu described the technology as “absolutely mind-blowing”.
“The amount of things we can apply it to what we do today is already huge,” he said.
“It’s not one of those things where it’s a theory – you think about crypto, blockchain, all of that stuff where you’re like ‘OK, what is this really?’ This is live today. I feel as strongly about this as I do the iPhone,” Shu added.
The image of the tiramisu on the left was generated by AI, the one on the right is a photograph
No comments yet