AI skills are like any other, it takes time to get results. I’ve been snowboarding for about 18 years. When I started, I could have said, “NO! You can’t snowboard!” For an Estonian snowboarder who goes to a big mountain a few times a year, this means 3 or 4 years. In real time it took me 8-10 one week trips to slopes over that period to feel comfortable on the board.
How is this related to AI? I know plenty of people who say, “I tried snowboarding once. Not for me!” Be honest, you’re not willing to put in the pain and effort it takes. You would rather do something else. We all stay in our comfort zones, most times.
Don’t look for ways you can’t do something!
It’s a waste of time. There’s no point in focusing on what AI can’t do. We’re already in that situation. This applies to everything else you need to do to get results. If you’re brainstorming with colleagues about creating a new service to bring in revenue and clients, those who only discuss why it won’t work should be kicked out of the room.
Put your time into figuring out how to make your daily grind easier. Focus on what you actually need and keep experimenting. That’s how you find out what doesn’t work.
Then there are the AI pseudo-critics who feed the Millennium Prize Problems into ChatGPT and complain it can’t solve them. One skill that’s becoming increasingly important is understanding what AI can help you with and what it can’t. This isn’t new. If you have a hammer, it’s useful to know you can’t use it to screw in bolts.
Don’t look for problems! They’ll find you.
Look for solutions!
At least once a day, I get so frustrated with AI that I feel like steam is coming out of my ears. Still, I manage to do my work with AI in a way that meets or exceeds expectations. Of course, the result is a hybrid, part me, part machine. Sometimes more me, sometimes more machine.
The AI keeps getting better.
When ChatGPT came out on November 30, 2022, it was almost useless in my native Estonian. Now we have AI that can produce almost copy-paste material but occasionally throws in absurd mistakes just to keep things interesting. And then there are the people who say you can’t use AI because it makes mistakes. I am happy for those people who don’t make mistakes, a bit envious even. I make 3 mistakes in one sentence… sometimes.
You can already get results with AI!
AI won’t give you results right now with the press of a button! Getting results takes effort. For example, I worked on a tedious task for my company that was absolutely necessary. Prompting took me an hour and a half! Only then did I get what I needed. The result was 30 pages of text. If I had done it manually, it would have taken me 10-12 hours, and because it’s such a mind-numbing task required by some bureaucrat, it wouldn’t have fit into a single day. I would have finished it a week later. AI can help in so many fields.
No point doing work you don’t have to do!
In this case, I got my work done 6 to 8 TIMES faster. You’ll find that AI can’t handle most tasks. I couldn’t just press a button and get it done. I had to “convince” the machine for over an hour to get the result I wanted. And I could have ended up with nothing. In most cases AI can’t do what you want.
It’s extremely funny to watch programmers who think the machine won’t replace their jobs. They get bad code, the same way I get bad text. They wrestle with it a bit and then do the rest themselves. I feel a bit sorry for them because, in a few years, they’ll have to take on new challenges and learn something more exciting. Then there are some programmers I know, who say that they have improved their productivity 2 to 3 times with AI. Really good developers too.
AI isn’t going anywhere!
Find the opportunities where AI can help you now. Even a 5-10% productivity boost is worth it. Stay up to date with AI developments in your field.
Always try things that didn’t work before when new models come out. Step by step, year by year, all kinds of office work will be handled by AI. If AI can do parts of your job now and you’re not keeping up, it might lead to problems.
Agents are coming
Agents are AI tools that act independently. Right now, you can’t tell ChatGPT to scrape all the financial data from Wikipedia, analyze different demographic indicators by per-capita GDP, and find correlations between traffic deaths and petty crime. That’s a random, simple example you could give an intern or assistant, but AI won’t do it… yet.
Today, there’s no “do my job” button. But there’s insanely high demand for “do my job” button. All AI developers are working on creating agents. At first, agents will be clumsy and fail at most tasks.
But even here, as millions of people use the agents, AI developers experiment, they train the machines, and AI keeps getting better.
Reskilling?
I don’t get people who talk about reskilling workers to get them new jobs to replace jobs lost to AI. If we need to retrain workers being replaced by AI, what exactly are we training them for? What skills should we teach digital marketers, paralegals, financial analysts, and customer service reps to keep them busy? And we have to assume they’ll even want to learn these jobs. Strawberry pickers? Estonians don’t want to pick strawberries.
Right now, you have the chance to keep up with AI development and become an “AI leader.” Eventually, AI agents will take over that work too. Every task that can be written down and is repetitive will eventually be done by machines.
If you keep up with AI, you’ll have more choices in the future.
Oh, and the cover image is the experiences you get when you start snowboarding.
