Will AI Take Our Jobs? The Reality of Freelancing
Recently, I've been hearing from clients more often, 'If you can make it with AI, you can do it for cheaper, right?' Honestly, when I first heard this, I thought, 'It's finally here...', but after actually using AI for work, I've learned a few things.
[I actually tried implementing AI tools]
I felt a sense of crisis myself, so over the past six months, I've invested about 10,000 yen a month in AI tools like ChatGPT Plus, GitHub Copilot, and Figma plugins.
So, how is it after actually using them? Coding has definitely become faster. Especially when creating React components, a task that used to take three hours now takes one. For brainstorming design ideas, what used to take me two days of agonizing now takes half a day when I use AI as a sounding board.
However—and this is important—while 80% of the code AI produces is usable, the remaining 20% is tricky. It will casually write code that is a security nightmare, and it will confidently propose processes without considering performance.
[Cases where I actually lost work]
To be honest, there are cases where I lost work. I had a client for whom I was doing simple LP production, three per month, for a recurring fee of about 150,000 yen. The contract ended last October because they said, 'We can now make them in-house using AI.'
At the time, I was depressed, thinking, 'Are you serious?', but three months later, that client contacted me. They said, 'We want to rely on a pro after all.' When they showed me the site they made in-house, it looked clean at first glance, but the responsive design was a mess, and it was terrible for SEO. In the end, I ended up redoing it and was able to sign a contract at a higher rate than before.
[How AI has changed the way I work]
What has definitely changed thanks to AI is the allocation of my working hours. Previously, it was 70% implementation and 30% proposal/communication, but now it's 50% implementation and 50% proposal/communication.
I've started spending more time in the requirements definition phase, pinning down with the client what features are 'truly necessary.' This is because I need to distinguish between what can be quickly made with AI and what needs to be properly designed by a human.
In a project last month where I was asked to customize an e-commerce site, the client said, 'You can do this with AI, right?', but when I heard the details, it required integration with an inventory management system. Things like this are absolutely impossible with just AI. In the end, it became an 800,000 yen project.
[New types of work that have increased]
Conversely, there is work that has increased thanks to the spread of AI. I now get two or three inquiries a month for things like 'AI implementation consulting.' Presidents of small and medium-sized enterprises ask me, 'We want to use AI to improve our business efficiency, but where should we start?'
Just last week, a production company with 30 employees asked me, 'We want to use ChatGPT for work, but what should we do about security?', and I was paid 300,000 yen for three days of work. I helped them with everything from selecting AI tools to establishing internal rules, and this is work that was born entirely because AI exists.
[So, what is the conclusion?]
Having been a freelancer for six years, I think that talk about jobs disappearing due to technological progress has been around for a long time. When WordPress came out, people said, 'Coders' jobs will disappear,' and the same thing was said when no-code tools became popular.
But in reality, every time a new technology appears, the value of those who can master it increases. It's the same with AI; those who cannot use AI will be weeded out, but those who can master AI and provide added value are actually earning more.
My revenue has also increased by about 1.3 times compared to last year before I introduced AI. Because my work efficiency has improved, I can handle more projects in the same amount of time, and I've also been able to take on new AI-related projects.
However, I have started to decline 'leave it all to AI' type projects. Those projects have low unit prices, and in the end, problems arise later, which is a hassle. I am focusing on work where I think together with the client and create better things while using AI as a tool.
Perhaps the people who will truly lose their jobs to AI are those who were only doing simple tasks that could be replaced by AI. But honestly, wasn't that just a matter of time even before AI came along?