Humans and AI can work together – but it can’t and shouldn’t replace accountants & bookkeepers.
1. AI can monitor and learn from you – but this has drawbacks
One of the ways we’re most familiar with AI is that it watches what you do, and then predicts what you’ll do next. And sometimes it gets used in ways that are considered bad, such as Facebook or Google monitoring you and turning that data into advertising.
But for getting your work done, if software knows what to expect, it can be really useful. At Apple decades ago we had a machine that would guide you through the day by giving suggestions based on what you would normally do at that time.. For example, if you usually go to a bagel store on Tuesday, it’ll say, ‘Oh, you probably want to go to the bagel store today’ – those sorts of things actually help us.
The best example of this, of course, is typing. Many of us have forgotten how to spell words now, because of auto correction. You stop worrying about mistyping. So the danger is we don’t want to go too far as the more we automate, the less we will be able to do ourselves. It’s hugely important that any tech with AI involved should work for the human being rather than replace them.
2. AI is not truly intelligent
AI products including things like voice recognition are helpful, but they aren’t intelligent – they’re nowhere near the capabilities of the human brain. You have to be very specific with your questions. For example, you can’t say, ‘Who scored the winning goal in last night’s hockey game?’ because AI doesn’t really know the definition of a winning goal. Computers can recognise certain keywords, and piece them together by mathematical formulas to try to figure out what an answer might be. But they don’t know what the words really mean. So for the hockey score, you’d need to ask for the score in a specific match which the AI can search for and find.
You can show 80,000 pictures of dogs to Google, and Google can now recognise a dog faster than any human. That’s impressive, but it can’t tell if it’s a real dog or a picture on a wall. And a one year old child knows that difference. Google doesn’t know what a dog is. So we sometimes have to be careful with the limitations of technology, how far we’re going to rely on it, we have to use our own trust. Nothing equals the human brain yet. We don’t know how the brain is wired, or we could make one.
3. AI will never care about your clients
I don’t think technology will ever get to the point that it says, ‘What should I do today that is good?’ A human being can sit down and say, ‘Here’s how I can help my clients, here’s something I can do, a step I can take and help the clients and this is what I should do’, and maybe do something new and creative – and innovative even – but we’ve never even spoken about machines ever having that ability to come up with an idea or a thought.
I can only surmise, not being a bookkeeper after all, but I believe bookkeepers need to get all their data into a certain format. You know, standards are great, but there’s probably not really one standard, and the rules change. And for businesses, knowing what sort of deductions you can get, and how to manage this and what sort of funding and if something has a category, and being able to go back and find records needs huge human oversight.
We’ve got to emphasise the importance of the human. The human is like the driver. Tesla promotes autopilot, which really isn’t, but says you, the human, has to be judged responsible. You know, humans get lazy sometimes when they’re allowed to. So it’s almost good to keep them on their feet.