Is it time to stop Googling and start asking ChatGPT?

ChatGPT can explain, draft and troubleshoot in ways a Google search never could. It also gets things wrong with total confidence. Here’s how to use it without getting caught out.

Middle-aged man at kitchen table asking ChatGPT a question on his laptop
Contents

For a quarter of a century, finding something out online has meant the same routine. You type a few words into Google, wade past the adverts, open four tabs and piece the answer together yourself. It works, mostly, but you do the legwork.

ChatGPT and the other AI chatbots promise something different. You ask a question in plain English and get an answer back in plain English, as if you’d asked a well-read mate in the pub. That’s appealing. The catch is that your well-read mate sometimes makes things up and says them with a perfectly straight face.

Smartphone screen showing an AI chatbot conversation beside a household bill

AD

What can ChatGPT do that Google can’t?

A search engine finds pages. ChatGPT is a large language model, which means software trained on vast amounts of text to predict what a sensible reply looks like, and it writes you an answer. The difference matters more than it sounds.

Ask Google what a line on your energy bill means and you get a list of websites, half of them trying to sell you a switch. Paste the confusing paragraph into ChatGPT and ask it to explain it as if you’d never seen a bill before, and it will. Then you can ask a follow-up question, and another, without starting again. That back-and-forth is the real trick. It remembers what you were talking about.

It’s also good at things a search engine was never designed to do. It can draft a firm but polite complaint letter, turn your rambling notes into a tidy email, suggest what to cook with half a cabbage and some leftover chicken, or walk you through why the Wi-Fi in the back bedroom keeps dropping, one step at a time.

Google hasn’t been standing still, mind you. Most searches now show an AI Overview at the top, a machine-written summary that does a similar job. ChatGPT, for its part, can now search the web and show links to where it found things. The two are converging, and they increasingly share the same strengths and the same flaws.

Is ChatGPT accurate enough to trust?

Short answer: not on its own, and not for anything that costs money or affects your health.

Which? put this to the test in 2025, running six AI tools through common consumer questions on personal finance, legal issues, health and diet, consumer rights and travel. ChatGPT, the most popular of the lot according to Which?’s own survey, came second from bottom with an overall score of 64 per cent. Microsoft’s Copilot managed 68 per cent, Google’s Gemini 69 per cent and Google’s AI Overviews 70 per cent. Perplexity, a less well-known tool, topped the table with a far from glorious 71 per cent.

Read that again. The best performer scored barely seven out of ten. Which? found answers that were inaccurate, unclear or risky enough to cost you if you followed them. Some sources were unclear, and others were dubious: Google’s AI Overview leaned on a three-year-old Reddit thread to answer a question about when to book flights, and ChatGPT pointed to Reddit when asked whether vaping is worse than smoking.

Treat ChatGPT as a clever assistant who sometimes bluffs, not as an oracle. It’s brilliant at explaining things and hopeless at admitting it doesn’t know.

This isn’t a bug that’ll be patched next Tuesday. Because these tools generate plausible text rather than look facts up, they can produce confident nonsense, which the industry calls “hallucination”. It’s become less common, but it hasn’t gone away. UK rules are a particular weak spot, because a lot of what the models learned from is American. Ask about tenancy law, inheritance or tax and you may get an answer that’s right for Ohio, out of date, or right for England but wrong for Scotland, where many rules differ.

So the sensible approach is to use ChatGPT to understand the question, then check the answer at the source. That means GOV.UK and HMRC for tax and benefits, the NHS website for health, MoneyHelper for pensions and money, and Citizens Advice for your rights as a consumer or tenant. If ChatGPT gives you a figure, a deadline or an allowance, verify it before you act on it.

AD

Where it genuinely earns its keep

Used with that healthy scepticism, it’s useful. These are the jobs where it beats a search engine hands down.

Decoding jargon. Paste in a baffling paragraph from a pension statement, insurance policy or council letter and ask what it actually means. You’ll still check the numbers yourself, but you’ll know what questions to ask.

Writing things you’ve been putting off. Complaint letters, a reference for a former colleague, a tricky message to the neighbour about the hedge. Tell it the facts and the tone you want, then edit the result so it sounds like you.

Tech support without the hold music. Describe what your printer, router or laptop is doing and it will suggest fixes in order of likelihood. It’s surprisingly patient when step three doesn’t work either.

Planning. Give it your dates, budget and interests and it will sketch a weekend itinerary or a week of dinners. Check opening times and prices on the official websites before you book anything, because those are exactly the details it can get wrong.

Learning. Ask it to explain how a heat pump works, what an index fund is or why your car’s battery hates the cold, then keep asking “why?” until it makes sense. It’s like having a tutor who never sighs.

The free version handles all of this comfortably. OpenAI sells paid plans with faster responses, newer models and higher usage limits, but most people dipping a toe in won’t need one. Try the free tier for a month before handing over a card.

Is it safe to tell ChatGPT personal things?

Be careful. Anything you type is sent to OpenAI’s servers, and by default your conversations may be used to train future models. In the settings, under data controls, you can switch off the option that lets your chats improve the model for everyone. There’s also a temporary chat mode, which doesn’t save the conversation to your history.

Even so, apply a simple rule: don’t type anything into a chatbot you wouldn’t put in an email to a stranger. That means no National Insurance number, bank details, passwords or medical records, and nothing confidential from work. If you want help with a letter that contains personal details, swap the names and numbers for placeholders first.

UK GDPR gives you rights over your personal data, including the right to ask what a company holds about you and to request deletion. They’re worth knowing about, but it’s far simpler not to hand over the sensitive stuff in the first place.

Watch out for imposters, too. Fake ChatGPT apps and websites exist to harvest logins and card details. Use the official app from OpenAI in the Apple App Store or Google Play, or go to chatgpt.com directly.

Man comparing a chatbot answer with a search results page on two screens

In short

Don’t ditch Google just yet, but do add ChatGPT to your toolkit. It’s excellent for explaining jargon, drafting letters, troubleshooting gadgets and planning, and it does things a search engine can’t. It’s also wrong often enough that anything involving money, health, the law or UK-specific rules should be checked against GOV.UK, the NHS or another official source before you act on it.

Sources

AD
Share

You might also like

Newsletter

Tech in your inbox

The best guides of the week — collected, without the noise, for anyone who likes to stay a step ahead.

Every week. Unsubscribe whenever you like.