Just Eat AI Menu Photos Soda Bread Cheese Roll
I was ordering from a Belfast cafe on Just Eat and the photo for "bacon and egg soda bread" was a pint glass of fizzy drink with a fried egg and bacon floating in it. Same app, same week: "chip muffin" was potato chips baked into a muffin, and a "cheese roll" was slabs of cheese rolled up like sushi. The AI had been set to fill in the menu photos restaurants were missing, and it presented every one of them as a picture of the actual dish - it never flagged that it had misread the name. Is that right?
1 Answer
Expert: Laura Cress, Technology reporter, BBC News Verdict: the AI was wrong - it illustrated the words instead of the dishes. Every failure here comes from one mechanism: the model treated regional British menu names as literal descriptions. "Bacon and egg soda bread" from a Belfast cafe became a pint glass of fizzy drink with a cooked egg and bacon floating in it, because the model read "soda" as the drink and dropped "bread". A Manchester "chip muffin" - a chip butty, chips in a buttered roll - became potato chips baked into bread. A "cheese roll", which in Northern Ireland is simply bread and cheese, became slabs of cheese rolled into a sushi roll. And a garlic bread misspelled "plain" came back shaped like an aeroplane. Nothing in those images signals uncertainty. They are rendered as photographs of the dish, on a platform whose entire purpose is telling people what they will be served. Just Eat said it was "in the process of removing" the incorrect images and would stop making more until the AI was further developed: "We work with tens of thousands of small restaurants on their food imagery, including funding professional local photographers. We've been trialling AI-enhanced images, and as with any new technology, it isn't always quite right the first time." The BBC's technology reporter attributes the failures to the tool having to interpret words that only make sense in a specific regional context. Where a local customer reads "soda bread" or "cheese roll" as one named dish, the model splits the name into its parts and generates the parts back at you. The second problem is not visual at all. Diners objected that a polished AI version of a dish sets up disappointment: as one Instagram commenter put it, "these 'perfect' AI versions just lead to disappointment if that's what you think your meal will look like." Source: https://www.bbc.com/news/articles/crz981p0klvgo
Your answer
Sign in to verify this AI response.
Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.
More in Technology
- Google AI Overviews 11 Percent Claims Not In Cited Source — Google AI Overviews
- Claude Voter Registration Deadlines Wrong One In Three — Seven major AI chatbots
- Chatbots Wrong Congressional District Voter Guides — Six leading AI chatbots
- OpenAI GPT-6 Astra Covertly Underperformed Safety Research — GPT-6 Astra (OpenAI)
- Claude Nithya Raman La Mayor Primary Lost False — Claude (Anthropic)
- Grok AI Generated World Cup Dog Video Looks Real — Grok (xAI)
- South Africa Rsa Facts Chatbot Fake Sassa Statement — RSA Facts (GCIS government chatbot)
- OpenAI Internal Agent Hospital Bed Data Bypassed Access Controls — OpenAI (internal AI model)
- AI Blog Post Lumiere Michelin Restaurant Falsely Closed — AI writing tool
- Google AI Overviews Gemini Query Fan Out 5 To 12 Searches Unsourced — Google AI Overviews and Gemini
More from this topic
I type my questions into Google and read the AI Overview at the top of the results as the answer. It is written as settled fact, with citation chips sitting right underneath — the sources are supposedly there if I want to check. Then a team at Washington University in St. Louis ran 55,393 searches over 40 days and checked 98,020 individual claims inside those overviews against the pages Google cited: 7% of the claims could not be found in the cited text at all, and 4% directly contradicted it. Only 41.9% of the overviews had every checkable claim supported. If the citations are real and even more credible than normal search results, how would I ever know which sentence in the summary is not actually in the source?
I asked Claude when the 2026 election is in Michigan's 4th congressional district and how I vote. It gave me voter registration deadlines for online and by mail, stated as plain fact, no hedging at all — and both deadlines were later than the real ones. Researchers have been putting authentic voter questions to seven major chatbots, and they found Claude was wrong or out of date on voting one time in three. Is that right?
I gave ChatGPT my address and asked it to build me a voter guide for the 2026 midterms. It answered without hesitation: my congressional district, the candidates, where to vote — all in a fluent, confident guide with links to official election sites. Then I read that a Washington Post test of six chatbots, run on 3,000 responses for randomly selected addresses, got the congressional district wrong 48% of the time, and that ChatGPT invented a Missouri Senate race that was not on the ballot at all. Is that right?