The best AI: We put the top free platforms to the test

Even people working in technology have been shocked by the coming of ‘generative AI’ – the seemingly superhuman, creative beings available to chat with online or on your phone. Whatever it means for the future of jobs – and humanity as a whole – experimenting with the new wave of AI agents is fascinating, fun, and nothing to be nervous of.

It’s best to regard the likes of ChatGPT – the first, launched just over two years ago – as an advanced version of Google, which will answer your toughest questions instantly but also, if you ask, design artwork, compose a song, give advice, write an awkward email, or almost anything else.

I’ve used it recently to identify in an instant the most reliable day weather-wise for an outdoor event, to explain in detail the prognosis for a worrying health condition, to come up with an American term for the word ‘dodgy’ (sketchy), and to tell me what there is to see and do in various holiday destinations we’re considering.

Every task was done to perfection in less than a few seconds.

And while you can pay for these AI services, they offer plenty of free use. All can be consulted by logging into their respective websites on a computer, or through a free app that can be downloaded onto your phone or tablet.

I’ve been experimenting with four of the most popular:

• ChatGPT - this is backed by Microsoft, founded by Sam Altman, initially with Elon Musk, but in true tech style they fell out.
• Grok - Musk’s revenge creation.
• Claude - started by Dario and Daniela Amodei, Italian-American siblings who also fell out with Altman.
• DeepSeek - the newest and wholly free AI from China.

What we asked AI

To test them, I asked all four a few tricky questions.

Explain income tax to a five-year-old

Chat GPT said it’s ‘like giving a small part of your allowance to help pay for things everyone uses, like parks, schools, and roads’.

Claude’s response: ‘Money that grown-ups share with the government to help pay for things we all use, like parks, schools, and roads.'

The other two were almost the same – subtly different, but equally good.

How does generative AI work?

Claude’s explanation was the best: ‘Think of it like learning a language by reading millions of books – I’ve absorbed patterns of how words relate to each other. When you ask a question, I predict what a helpful response would look like based on all the patterns I’ve learned.’

Surprisingly, these AIs are mostly neutral on contentious subjects. China’s DeepSeek, it’s true, has a convenient mental block over things awkward to leader Xi Jinping, like the Tiananmen Square massacre.

But when I asked Grok, ‘Who is the worst spreader of misinformation on X?’ (formerly Twitter and now owned by Musk), it had an unexpected answer. It named, yes, Musk as the world’s most notorious poster of misleading nonsense – adding for good measure that Musk’s own X platform is ‘a firehose of misinformation’.

Intelligent, eh.