I was thinking about how quickly our relationship with AI has changed.
When tools like ChatGPT first appeared, you probably treated every answer with a bit of suspicion. You’d ask it something, read the response and check whether it had got things right.
A couple of years later, how many of us are still checking quite so carefully?
That’s understandable when you’ve used AI hundreds of times and it’s usually been helpful. You start trusting it a little more. Then you give it more important jobs because it did such a good job with the easy ones.
That’s where I’d be a little more careful.
Let’s say someone in your finance team gives an AI tool a spreadsheet and asks it to analyze the figures.
It comes back with a neat summary and points out something interesting. The person reading it thinks, “That makes sense”, and uses the information in a report.
But how did AI reach that conclusion?
That’s the bit you really should understand.
AI is already being used for some fairly important work. People are asking it to help with financial information, expenses, payments and business reporting. It can be useful in those areas because it can get through a lot of information very quickly.
You need to remember that it can get things wrong though.
And it can be convincing when it does.
A bad answer doesn’t always look like a bad answer. It can arrive beautifully written, neatly presented and sounding completely confident.
If nobody checks it, that answer can find its way into a report or influence a decision.
There’s another question you may want to ask your team, too: What are you putting into these tools?
AI usually starts with harmless jobs. Someone asks for help writing an email or summarizing a document.
But once they get comfortable with it, they might start uploading more. Perhaps it’s a spreadsheet, a customer document or some financial information they want help analyzing.
At that point, you should know which AI tool they’re using and what happens to the information they give it.
You need a sensible idea of where you’re happy for AI to help and where somebody needs to check the work properly.
If AI drafts an internal email and somebody tweaks it before sending, there’s probably little to worry about.
If it’s helping with something that could affect your finances, your customers or an important business decision, a knowledgeable person should look at the result before anyone acts on it.
Let it save you time. Let it take some of the repetitive work away. Let your team experiment with useful ways of working.
But don’t get so comfortable with it that nobody asks, “Is this right?”
If you’re wondering how much AI is already being trusted across your business, that’s something I can help you review. Get in touch.