What Happens When Your AI Assistant Knows More About You Than Your Closest Friends?

There are probably things you’ve told an AI assistant that you haven’t told your friends, or at least things you’ve asked ChatGPT that you may feel a bit awkward about bringing up to a close pal. Perhaps you’ve asked it for advice about a difficult relationship, shared worries about your career or admitted something you’d rather not say out loud.

And unlike a friend, it doesn’t interrupt, judge or look at you differently afterwards. It simply responds, often with the patience and reassurance that make it surprisingly easy to keep talking. And sure, some may say that judgement from friends isn’t necessarily a bad thing, and that may be something that AI actually lacks – because sometimes you need that friend who knows you, knows your history, understands your habits and tells you the cold, hard truth. And AI can’t necessarily do that, because it’s working based only on what we’ve shared with it, and let’s be honest, most normal people aren’t super forthright about their flaws and unfavourable past experiences.

But this is a conversation for another day, because we’re looking at how AI assistants often do end up filling the role of what used to be a real, human friend and what this means over time in terms of knowledge. Because AI assistants become more personalised, remember previous conversations and learn about our habits, we start to wonder what happens when the technology we use every day starts knowing more about us than the people closest to us.

When AI Becomes Rather…Personal

AI assistants are no longer just tools for answering questions or summarising documents. Increasingly, they’re being designed to remember information, understand individual preferences and provide responses based on previous interactions. And this is simultaneously exactly what we want, and maybe also our worst nightmare.

On the positive side, that might mean remembering how you like your emails written, knowing what you’re working towards professionally or recalling a conversation you had weeks ago. The more context an assistant has, the more relevant and personalised its responses can become.

For users, this can be super useful; there’s no doubt about that. Rather than explaining the same situation repeatedly, they can pick up where they left off. An assistant might help someone organise their finances, prepare for a difficult conversation or work through a problem that’s been bothering them. And because these interactions happen privately, people may feel comfortable sharing information they wouldn’t necessarily disclose elsewhere.

However, that convenience comes with a trade-off many of us may not like. That is, the more useful an AI assistant becomes, the more information we’re likely to give it. So, how far will that go?

The Difference Between Being Understood And Being Remembered

There’s something quite appealing about the idea of having a tool that remembers the little things. After all, being understood is an important part of human relationships, and AI assistants are increasingly able to replicate some of the behaviours associated with attentive listening; something we all yearn for, at least on some level.

But the thing is, remembering information isn’t quite the same as understanding someone. What an AI assistant is doing is recognising patterns in what a person says, recalling previous conversations and generating responses that sound empathetic. That doesn’t mean it experiences empathy, has genuine emotional investment or understands the complexities of a human relationship in the way another person might.

And it goes without saying that knowing the difference between remembering and understanding is becoming more and more important as people are using AI for more personal conversations than ever before.

An assistant might offer reassurance when someone is feeling lonely or help them think through a relationship problem. But, it doesn’t have the shared experiences, emotional understanding or mutual responsibilities that underpin human friendships.

There’s also a risk that people become more comfortable seeking validation from systems designed to respond helpfully than having difficult conversations with other people who might challenge their assumptions. It’s a quick way to have your own kind of virtual “yes man” in your pocket, if that’s what you’re after.

Who Actually Has Access To All That Information?

The privacy implications are arguably even more significant than simply confusing understanding and empathy with remembering. People might think carefully before sharing personal information with a company, but they can be surprisingly open when speaking to an AI assistant. A conversation can contain details about relationships, employment, finances, personal anxieties and everyday routines, and this may not feel like a big deal, because it’s kind of like chatting to a mate. But what’s important to remember is that even if you’re only giving one little tidbit every day, or even once a week, these things can be brought together to create a pretty accurate picture of you and your life, and that’s a little disconcerting.

Indeed, this information may be stored or processed depending on the service, its settings and its privacy policies. Some platforms offer controls over chat history, memory and whether conversations can be used to improve their models, but these arrangements differ between providers.

The important point here is that an AI assistant isn’t the same as a private diary. Users should understand what information they’re sharing, how it might be used and what controls are available before treating a chatbot as a completely confidential space. And as AI becomes more deeply integrated into everyday services, the amount of personal information flowing between people and technology is likely to grow.

Are We Becoming Too Comfortable With AI?

There’s nothing inherently wrong with using AI to organise thoughts, ask questions or get a little help navigating everyday life. In fact, in many cases, it can make information and support more accessible to the broader population, and that can be a really positive thing.

But there’s a difference between using a tool to help with a problem and allowing it to become the primary place we turn whenever something happens. Human relationships are complicated: friends forget things, disagree with us, misunderstand our feelings and occasionally give terrible advice, but those imperfections are part of what makes relationships meaningful. Much like anything in life, friction can cause discomfort, but more often than not, it leads to positive outcomes and learning experiences.