Most people still don't know what AI can already do
These days I spend my time talking with businesses about how they use AI and how they could automate the work inside their own business with it. Some of what I hear shows me how early we still are in the spread of AI, and that even highly educated people who are very good with computers don’t realize what it can do. Some of the questions I’ve heard:
- Can AI read what’s in an image and write it into an Excel sheet in the format I want?
- Can it enter the information into the portal I use in my browser?
- Can it really take the files that come in over WhatsApp and put them in the folder on the server?
- Can it do all of these things I just listed?
I have to say these are already very simple jobs for frontier models. I’ve seen managers and business owners who still do all their work by talking to ChatGPT, and who run their work inside chat sessions.
For them the process is just starting. As technical people, we’re at the front of it, and it feels like we’re leading in an age of transformation. But businesses don’t seem to have noticed the potential yet. That also shows how much potential the people who specialized in AI early still have.
AI is already taking jobs
As anyone could guess, AI keeps putting people out of work. It has already shown its effect at companies in the software industry first. Now that effect looks like it’s turning into a fire that spreads fast. I may share the reasons in a separate post. For now, let’s stay on the topic of this one.
300,000 receipts
Let me give an example from something I’m working on. I noticed that accounting offices that read physical invoices and receipts spend thousands of hours of dozens of employees on this work. I tested how high a success rate an AI-assisted automation system would reach if it did this whole process instead of an employee.
You can find 300,000 invoices and receipts online for free. I downloaded all of them and ran some quality checks and validation steps with decision models like Jev and the OpenAI Decisions API. These models produce structured output and cost very little for hundreds of thousands of operations. Then, with general AI models, I tested a workflow that extracts the information from all of these invoices, turns it into digital formats, and runs general accounting operations on all of them.
As you can guess, this took a while, about an hour. But in the end I had a reading workflow that could digitize 300,000 invoices with over 95% success. I’m sure this flow, which I prepared in 1-2 hours as a demo for a business, covers most of the work tens of thousands of accounting people still do again and again every day.
In the accounting sector I haven’t come across a proper solution that runs this workflow. There are companies that already sell a certain part of it as a service, but if you set up the whole workflow in one go, more than 90% of the work can become automatic. Until I did it, though, no business thought it could be done. And for me it wasn’t even a complicated solution.
This alone shows that businesses don’t realize how good AI is, and that they don’t realize solutions that turn their whole daily workflows into automation, not just ChatGPT-level tasks, can be delivered in a day. Who knows how many more people in how many other sectors, working as managers or just doing their daily jobs, aren’t aware of this power.
”It’s still a few years away”
The strange part is that these people think these solutions will only be possible in a few years, so they believe their jobs aren’t in danger today and that they’ll be affected sometime in the near future. In a time when even the software industry, which needs a lot of talent and skill, has had its share of AI, it isn’t hard to predict that most white-collar workers will be hit hard by companies’ AI transformations. In fact, we’re already inside that transformation. Startups that see the power of AI have already moved.
Not much noise is being made about it, but I think a quiet transformation is happening very fast in the background. In the daily AI news, the only things that seem to get attention are layoffs at giant companies and debates about whether AI can ever be as good as humans at art. Meanwhile millions of people think that having AI do all of the work millions of white-collar people do at their computers still takes very high technical skill, and that it will only become possible with experience. I think the exact opposite. Even today, AI works well enough to put millions out of work. To make it possible, all it takes is a process as easy as someone stepping into the work, solving the workflows and adapting them to AI.
Why businesses wait
Businesses want to be at the front of the AI transformation and benefit from it early. From what I’ve heard directly, every business manager and owner I’ve talked to is sure the future is AI. But they may be far from seeing that they’re already inside that future, because they think it’s something that will happen later and that they aren’t ready for it yet. I can see this first-hand, because they’re well informed about it and I can tell they’re very curious.
Still, they always prefer not to be the one who takes the first step, and to avoid risk. It’s an unknown situation, and they want to go ahead by watching competitors or people they know try it and seeing their results. Not putting their business at risk is a sensible approach. But being sensible isn’t the key to success. I think going through this transformation without taking risks will make businesses late for the opportunities the first experiences bring, and leave them behind. Because I notice they don’t know where the safe point is, or exactly when something has to be done. The fact that they don’t realize the future they’re waiting for is happening today is the biggest example of that.
AI in everyday life
For a normal person, you’ve probably heard that using AI in daily life is still about as limited as ChatGPT, and that very few even pay for a subscription. You can see it in the surprise on people’s faces when I tell them about the great vacations and the tax refunds I planned using AI. Beyond helping with what to eat that day or what to watch on TV that evening, they can’t see many uses for it. That goes along with their daily work and how much they deal with technology. It isn’t a mistake on their part. It shows they don’t realize the opportunities and what they might miss.
The other day a colleague at work told me a normal person doesn’t need AI. I said: “The progress the people who use it make in their lives will soon make normal people feel left behind, so they’ll have to learn it better and fit it into their lives. Otherwise they’ll be bursting with envy.”
People who have fit AI into their daily lives will get more access to opportunities. It will probably give them a better diet, chasing discounted food at the supermarket, catching their dream vacations at cheap prices, catching education opportunities, and much more. These are things you can do without AI too. But the gap will get serious between someone whose AI assistant sends a notification to their phone at lunch saying which supermarket has a discount on a food they like that day, and someone who still reads the supermarket flyers one by one. With every bit of extra time lost, the people using AI assistants will reach more resources, and do it in the most effective way.
As AI-powered tools spread, they’ll also start reaching ordinary people alongside this integration. Cost may be a problem right now, but AI gets cheaper every day, and it looks like this development will continue until it becomes a live system we’re always connected to and that’s with us all the time.
Two different speeds
Businesses and people seem to be integrating AI at different speeds. What businesses want is to calculate how AI automations that take over their workers’ jobs will multiply their income in a short time. A normal person, on the other hand, doesn’t have enough money right now for AI’s direct benefits, or doesn’t understand the technology. I think businesses will adapt to AI very fast and layoffs will increase, and the people who are laid off will see the AI that caused their layoffs as a tool, not an enemy, and hold on to it to get back into the system.
It isn’t hard to predict that the near future will bring very big changes to the economic system, or a completely redesigned economy. I don’t think the transition will be easy, and hard times are waiting for us. But once full integration is reached, I believe AI will bring a very big benefit to everyone. Let’s see what the future shows us.