When Canva rebuilt its platform around AI back in 2024, it ran into a major problem. Most of Canva’s users are on its free plan. Before AI, serving these free users was relatively cheap, but serving AI features to millions of free users became extremely expensive.

But the company also realised that its internal structure has to change. So, it’s merging its product, growth and platform teams, and they are reducing management layers. They are moving towards smaller teams where employees can take full ownership using AI.

Now there is a very important emerging pattern here. Jack Dorsey’s Block, Atlassian and Meta, all of them laid off people, and they are offering a lot of benefits as well. And Microsoft just offered voluntary buyouts to 60,000 employees.

Looking at the pattern, it is quite clear that teams are getting leaner. The people who stay have two things: business context and AI skills. Anyone whose job was executing instructions is being shown the door.

Another issue is that the more people you keep in your company, the more you now have to invest in their upskilling and their token usage, because now the cost of an employee is not just the salary but also the tokens they use and their upcoming education. From what it looks like, this trend is only going to get worse.

Nobody is ever okay losing a job, but at least Canva has given them enough resources to stay stable for six months and they are also giving them money to start their own business. Because even Canva recognises that when there is a mass layoff, you can’t expect people to just learn a new skill and get a new job; the easier route is to start doing individual services. So, there is a very high chance that there is going to be an upsurge of one-person companies, and it’s happening because AI knows everything and can do everything. What do we do?

I’m not going to pretend I have the answer. I’m 24. I use these tools every day. I’m scared too. But I’ve been watching what still requires human touch when everything else gets automated. And I’ve gathered these three rules for staying relevant.

Rule one:

Find out who you were before the GPT era. Maybe you were very curious, very petty, obsessive, funny. Maybe you cheated in exams. Maybe you stole other people’s lunches. I don’t care whether you are immoral, evil, good or bad. Do not let society limit your self-discovery. But how do you figure that out? Well, that brings me to rule number two.

Rule two:

Go and ask your favourite LLM any question such as this one. Now which one do you choose? Why that one? What do you hate about this one? Why is this cringe? Why does this design feel so soulless? Why is this joke not so funny? This thing that you are doing, that is called taste; you must develop it.

Rule three:

Go and touch grass. No, scientifically, go and touch grass. There is an old problem in cognitive science called the symbol grounding problem, which asks how words and symbols acquire meaning without referring to more words and symbols. An LLM can describe falling in love. It can describe breaking your leg. It can write 4,000 words describing what the rain outside feels like. But the description is not experienced.

So, build the thing you keep postponing. You should play a sport, grow a plant, and talk to a stranger. Pitch your billion-taka idea to a pigeon at Dhanmondi Lake because maybe the scary thing about AI isn’t that machines are becoming more human. Maybe it is discovering how much of being human we were already missing out on.

Our generation is the first in the history of mankind in which we can clearly separate what looks like intelligent linguistic behaviour from a real human mind. That is a privilege. So, take away my writing, take away my memory, take away my degree, what is left? That is the experiment that I am interested in.