Is AI a Mirror? If So, What Do You See?
The AI we chat with has learned from huge amounts of content, much of it created by people. And what we ask it brings a little of us into the conversation too. It does make you wonder how much of ourselves we might recognise in its replies.
People have plenty to say about AI. Some love it. Some want nothing to do with it. Others use it quite happily, but would still welcome a day without everyone talking about it.
When I read posts about AI by people I know, I find myself thinking, “Yes, that sounds like you.” Their familiar opinions and way of looking at life come through in what they say about AI. That’s what got me wondering: is AI a mirror?
Someone brings it a head full of ideas and asks for help finding the thread. Someone else wants help with their English because they know what they mean, but the words won’t quite cooperate. Another sees an opportunity to churn out more sales messages. The technology matters, but so does what we want from it.
And if you don’t use it, you’ll have your reasons too. What interests me is how differently we each respond to the same technology.
But if AI has become somewhere you take your business ideas, what are you hoping for when you open that chat? Help putting something into words? A different way of seeing it? Or a reassuring, “Yes, that’s a wonderful idea”?
I can see the appeal of that last one. It’s lovely to hear that your idea is a good one. But has the reply helped you think it through, or simply told you what you were hoping to hear?
A new technology, a very old collection of ideas
Do you remember going to the library to look something up? You’d find a book, turn to the index and hunt for the page you needed. Another book could give you quite a different account of the same subject. Being on the shelf didn’t mean every author agreed, or that everything they wrote was right.
That’s a useful way to think about the material behind AI chat tools. They learn patterns from vast collections of writing, including websites and books. Different AI tools haven’t all learned from the same library.
And some of those shelves reach a long way into the past. One collection of text used to train AI language models includes books published before 1919. This is more than yesterday’s posts and this morning’s news.
Physical books have been part of that collection process too. A 2025 court order describes Anthropic buying millions of print books, often second-hand, having their bindings removed and their pages scanned, then discarding the paper originals. Quite a different fate for a book from being returned to the library trolley.
Those shelves hold human wisdom and human prejudice, acts of kindness and accounts of cruelty. They contain beliefs we still value and others we’ve spent years challenging.
The material AI learns from can contain all of that. So when we ask it to be “more human”, which parts of being human do we mean? The library has plenty of examples of both our care and our capacity to cause harm.
Who decides what the mirror shows?
Before you type anything into an AI chat, other people have helped shape how it will answer. They choose material for it to learn from, guide its responses and set limits on what it should help with.
I recently tried an AI tool that assessed a business idea against seven criteria and gave it a green light. Useful, but did that mean people would buy it? That hadn’t been established.
Someone had chosen those seven criteria. The tool helped me examine the idea through the questions they had selected, but it couldn’t settle every question I had about going ahead.
That brings us back to the mirror. What it shows has already been shaped by someone else’s choices about what matters. Then we bring our own expectations to the answer.
What do you bring to the conversation?
You can ask AI to challenge an idea, help you explain it or find reasons it could work. You’re bringing the same idea, but asking for something different each time. That helps shape what comes back.
Then there’s how you like to be spoken to. You can ask for warmth, plain English, humour or a more direct answer. Over time, with instructions and corrections, the conversation can start to feel familiar. “Yes, that sounds more like it.”
Some people give their AI a name or refer to it as “he” or “she”. Others see it as somewhere to empty their thoughts before deciding what to do. Giving it a name doesn’t tell us who you are, but the role you give it is worth considering. Is it an assistant, a sounding board or someone you’ve started asking for permission?
For a heart-led business owner, an idea can begin as a quiet knowing, long before you have words for it. Taking it to AI can help you explore how to express it. When a suggestion comes back, you still decide whether it belongs with what you wanted to create.
That’s another part of the mirror: what you ask for, what you correct and what you choose to keep.
When the reflection doesn’t quite fit
Sometimes you know what you think, but talking it through helps you explain it. That’s where AI has been useful to me. It can offer another perspective or help put scattered thoughts into an order someone else can follow. I still have to decide whether those words say what I mean.
But when it comes to writing, I’ve noticed something else. I can bring a strong opinion into the chat and get back a softened version. The sentences are tidy, everyone’s been politely considered, and somewhere along the way the point has lost its teeth.
Then I push back. “No, that’s not what I’m saying.” The next version comes closer to what I meant. Sometimes having to explain why the first answer didn’t fit helps me understand my own point better.
But that also makes me pause. When it changes its answer, has it understood my point better, or is it simply agreeing with me because I pushed back? Agreement alone doesn’t settle that.
So the mirror image has its limits. What comes back can help you recognise a thought, but it can also soften it, miss it or give you something you didn’t mean at all. You still get to read it and say, “Hang on. That doesn’t sound like me.”
What do you see?
If AI is a mirror, the reflection is a mixture. There’s the human material it has learned from, the choices of the people building it and what you bring to the conversation. It won’t give you a complete or reliable picture of yourself. But the exchange can give you something worth noticing.
You ask for suggestions and realise none of them fit. What is missing? You feel relieved when it praises your idea. Were you looking for help developing it, or permission to begin? You reject a polished paragraph because it makes a promise you would never make to a client. What you accept or question can tell you something about what matters to you.
For those of us who value inner knowing, that moment matters. You can welcome another perspective and still notice when something doesn’t sit right. You can also feel drawn to an answer and take time to check it. Something sounding right and something being right aren’t always the same.
You don’t have to settle the whole debate about AI to consider your own relationship with it. And if you choose not to use it, your concerns deserve attention without being explained away.
The next time you open a chat, notice what you ask for and what you do with the answer. What do you keep? Where do you push back? What becomes clearer about what matters to you?
The answer on the screen is only part of the conversation. Your response to it belongs there too.
How does your work come across to someone new?
Knowing what you mean is one thing. Putting it into words that someone else understands is another.
Someone finding you through a search or an AI answer may be seeing your work for the first time. They don’t know your background or what you’ve shared before. Do your words explain what you offer and why it could be useful to them?
My free guide, Beyond Your Followers, helps you look at one offer page or post with that new reader in mind. Simple questions help you check whether it explains what you offer, who it’s for and what to do if they’re interested.
You’ll finish with one thing to make clearer, so someone new can see how your work could help them.
