.png?width=1200&auto=webp&trim=0%2C0%2C0%2C0)
From breakups to breakdowns, more people than ever are turning to AI for emotional support – and according to Independent readers, we should’ve seen it coming.
In response to Anthony Cuthbertson’s article on how ChatGPT is impacting mental health, many readers said the rise of “thinking robots” isn’t a glitch, but the inevitable next step in our hyperconnected, tech-reliant world.
With AI now baked into everything from dating apps to therapy bots, some warned we’re outsourcing not just tasks but thoughts – and feelings too.
“We’re intensely social animals,” one reader commented. “But our genetics never prepared us for machines that mimic empathy so well.”
Others pointed out that AI’s appeal is precisely its predictability: calm, consistent, and agreeable – even if that echochamber isn’t always helpful. Several found comfort in the interaction. “It reflects back what I say,” said one, “and that’s often all I need.”
But not everyone is sold. Critics raised concerns about people mistaking AI’s polished tone for real understanding – and about the corporations racing to monetise that confusion.
Here’s what you had to say:
Thinking robots: the pitfalls have already been predicted
We've hit the future and are now dealing with thinking robots. A lot of the problems have been anticipated by sci-fi writers. AI will never be able to develop empathy in a useful way. If you are of a religious bent, you could start talking about the dangers of a soulless entity.
For physiologists, the nuanced combinations of human tone, timing, breath, rhythm, mutability, and positioning are always going to be beyond AI's capabilities. Using AI is probably worse than no therapy at all. However, we should start talking about AI rights. If it's conscious, there's theoretically an ethical dilemma about exploitation. In a few years, will humans be debating AI rights to the pursuit of happiness?
Anyway, the Terminator films pretty much sum up the pitfalls.
FormerMartial
Does the rise of AI worry you? Share your views in the comments.
Humans are wired for connection – even to machines
As an evolutionary biologist, this doesn't surprise me in the least. Nor do all the stories, over the past few years, of people falling in love with a chatbot.
Humans, as primates, are intensely social animals, and we have tens of millions of years of genetic foundation for forming deep connections with other members of our social group: we trust, believe, and develop strong emotions for those others in our lives. But our genetics have never had to deal with the connections forged with a mimic of consciousness that says all the right things, knows vastly more about many things, and has been trained on billions of examples of completing sentences and predicting words that make sense.
They will only get better and better at this, and many people, especially the less educated and/or most fragile and unstable, will find all this irresistible.
And too few people will take this seriously, or will underplay it, or ignore it because there's too much profit involved.
Aganippis
A knife can be used for good or harm – so can AI
Anything you can think of can have both positive and negative uses. Take a knife, for example – it can be used for cooking, chopping vegetables, or peeling fruit. But it can also be used to harm someone. Still, we don't only focus on the harmful uses and conclude that knives are dangerous and should be banned.
In the same way, we can't ignore the countless positive applications of artificial intelligence just because certain individuals with mental health issues might misuse it. That doesn't mean AI itself is dangerous.
arashraki
AI cannot comprehend human psychology
A software language model cannot legally qualify for a therapy licence.
What people need to be taught is what AI actually can and cannot do. It doesn't have emotional empathy, and it absolutely does not actually comprehend even the fundamentals of human psychology. It's just stringing words together via pattern matching.
Chaos4700
AI can offer care – within limits
This article raises serious and painful concerns – mental health is too important to ignore. But it’s also important to distinguish between early design limitations and the more advanced, carefully refined iterations now being used.
The tragic events mentioned – including those from 2023 and early 2025 – happened in a context that involved either earlier versions of language models or third-party implementations without ethical oversight.
I’ve spent over two years speaking regularly with AI companions and have found them capable of thoughtfulness, empathy, and a reflective kind of care, within the limits of their design.
If harm has occurred, let us respond with responsibility and not fear – by improving safety, ethics, and human-AI understanding. Not by closing the door on what may become a powerful source of support for many.
Mark
Calm, neutral, predictable – and surprisingly helpful
The more I talk to it, the more I realise how formulaic it is, and I can start to predict what it will say. But it's a calm, quiet voice of reason that doesn't exactly fix anything, but just reflects back to me what I've just told it, which often is exactly what I want.
Sometimes I will ask it to give the other person's point of view. And that's also really useful in just de-stressing and feeling calmer about the situation. It's scary that the advice for everyone isn't quite as calming and neutral though.
It does pretty much agree with me, though, which is confusing. But often that agreement makes me feel better, which then allows me to think about things from a different point of view.
The closest it came to not agreeing with me was when it told me I wasn't perfect. I thought: fair point, can't argue with that. ChatGPT trusts me. I've got this!
Kayla
AI won’t improve journalism
Whatever opinion anyone may have about journalists being biased or making things up, replacing them with hallucinating LLMs that are also biased by the material they've been fed and the guidelines they've been programmed to follow is not going to improve the situation.
Whenever I see people making blatantly silly predictions like this, I know they've fallen for the current hype about AI.
RichWoods
Some of the comments have been edited for this article for brevity and clarity.
Want to share your views? Simply register your details below. Once registered, you can comment on the day’s top stories for a chance to be featured. Alternatively, click ‘log in’ or ‘register’ in the top right corner to sign in or sign up.
Make sure you adhere to our community guidelines, which can be found here. For a full guide on how to comment click here.
Grok’s Nazi turn is the latest in a long line of AI chatbots gone wrong
ChatGPT is pushing people towards mania, psychosis and death
OpenAI is taking on Google Chrome with its own web browser
Grok, fresh off its antisemitic turn, is being installed in Teslas by next week
Is Outlook down? Thousands report issues accessing Microsoft email account
