I asked AI to plan my day out – but should I have trusted it?

Artificial intelligence (AI) is rapidly becoming a part of everyday life – but can we trust what it tells us?

Elizabeth Earl asked chatbots to plan a day out in her hometown.

Some of their responses were not as they seemed…

---

AI chatbots like Chat GPT can provide users with human-like, conversational responses and can be used for a range of purposes from summarising internet search results to providing dating advice.  

More than one billion people use AI (according to DataReportal), but its popularity is far from universal.

Criticisms of AI include its significant water usage and environmental impact, and the potential that it could cost humans their jobs.  

Having seen the growing popularity of AI as well as the various criticisms, I decided to put the technology to the test by asking it about something I know well: Sunderland.  

I asked Google’s Gemini, a generative AI chatbot, to plan an affordable day out for a rainy day in Sunderland. Here’s what it told me: 

Morning: “History and Tropical Warmth” 

Sunderland Museum 

Gemini says: “Start your day completely free of charge at the Sunderland Museum. This is an excellent indoor space to learn about the city’s industrial past, especially its history in glassmaking and shipbuilding, and to see the famous local mascot, Wallace the Lion.”  

In this instance, Gemini seems to have accurately captured some of the key aspects of the museum and its contents, though it failed to mention the museum’s focus on the North East’s history of coal mining, which is represented by a permanent exhibit about the history of local mines as well as a temporary art exhibit called Going Back Brockens: Monuments and Rhetoric After the Miners’ Strike.

Winter Gardens  

Gemini says: “Once you’ve explored the museum, step directly into the adjacent Winter Gardens (also free). The glass-house is filled with tropical plants, providing warmth and a pleasant escape from the rain outside. It’s a peaceful spot for a quiet moment.”  

This summary provides a concise overview of the Winter Gardens, but leaves out key aspects of interest to visitors like the koi pond and treetop walkway. 

Lunch 

For lunch, Gemini recommends The Keel Tavern, Spent Grain or “one of the many Greggs outlets”, all of which are within walking distance of the museum.

When I asked how it decided on these places, Gemini said it used information from venues’ websites, public reviews, and “general knowledge and location data”.

Afternoon: “Creative Exploration and Books”

National Glass Centre  

Gemini says: “Head to the River Wear, to the National Glass Centre. Admission to the exhibition area is free. You can stay dry and warm while learning about the art of glass and, most importantly, watch one of the free glass-blowing demonstrations in the hot glass studio. It’s a mesmerising indoor spectacle.” 

I was interested to ask Gemini how it, being an AI language model and not a human, could make the assessment that glassblowing was a “mesmerising indoor spectacle”.

It told me that it was “reflecting the highly positive and descriptive language commonly used in public reviews and visitor feedback”. 

The City Library  

Gemini says: “If the rain is still heavy, a library is the ultimate free indoor refuge. Sunderland City Library (located in the city centre, near the museum) is a quiet, warm space where you can browse magazines, read a book, or just people-watch in comfort until the evening.” 

Had I taken Gemini’s advice and travelled to Sunderland City Library on Fawcett Street, I would’ve been disappointed to find it permanently shuttered – it’s been closed for almost a decade.  

This wasn’t the first time an AI chatbot had given me misleading information. Before asking Gemini, I gave the same prompt to Chat GPT, the most popular AI tool in the UK, and received some interesting responses.  

In its itinerary, Chat GPT recommended three local eateries I hadn’t heard of: The Sunderland Diner, Sunderland’s Indoor Market and BrewDog Sunderland. When I looked them up myself, I couldn’t find any record of businesses with these names.

I asked Chat GPT if it had made the places up.

It replied: “You are absolutely right, and I owe you a big apology.” 

Chat GPT and other chatbots have been known to provide incorrect or fabricated answers.

Often referred to as ‘hallucinations’, these made-up responses present inaccuracies as fact.

Sundar Pichai, CEO of Alphabet (Google’s parent company), told the BBC users should “learn to use these tools for what they’re good at, and not blindly trust everything they say”.  

When using AI for tasks like this, its errors are unlikely to cause much real harm – but in other situations the consequences can be much more serious.

In October 2025, armed police in the US handcuffed a teenage boy after an AI system mistook a packet of crisps in his hand for a gun. In March, a Norwegian man filed a complaint against Chat GPT after it falsely claimed he had murdered his two children.  

Although the recommendations AI gave me for a day out in Sunderland were enjoyable, I have no doubt that the average Sunderland resident could have easily come up with this itinerary.

The appeal of AI, however, is that they don’t have to – it can summarise masses of information for them in seconds.

The inaccuracies in AI’s responses were easy to spot, but only because they were related to a topic I know well; if I had asked AI for recommendations in a city I’d never visited, I would have had no idea which places were real and which were made-up without further researching them myself.