Back in 2018, Kai-Fu Lee — the former president of Google China and one of the most influential voices in global AI, best known for his book AI Superpowers: China, Silicon Valley, and the New World Order — made a prediction that sounded almost like science fiction at the time: within 15 years, AI would be capable of replacing somewhere between 40% and 50% of all human jobs.
In 2024, well into the era of ChatGPT, Midjourney, and AI agents doing everything from writing legal briefs to running customer service desks, someone asked him whether that old prediction still held up. His answer was blunt: it’s “eerily accurate.”
That’s a sobering statement. But the more useful question isn’t whether the prediction is correct — it’s this: which side of that prediction are you currently standing on? Are you one of the people AI is quietly preparing to replace, or one of the people using AI to pull ahead? The uncomfortable truth is that the tool itself doesn’t decide that for you. Two people can be using the exact same AI subscription, paying the exact same monthly fee, and end up on completely opposite sides of that divide.
Two People, One AI, Two Very Different Outcomes
Picture two freelancers, both paying for the same AI chatbot subscription. Both are intelligent, hardworking, and genuinely trying to use AI to get ahead.
The first person treats AI like a smarter version of a search engine. They open it up, type a question, get an answer, copy it into their work, and close the tab. It’s useful — genuinely useful — but it’s essentially the same workflow they had before, just a little faster.
The second person treats AI completely differently. To them, AI isn’t a search box; it’s staff. They don’t ask it one question and walk away — they give it standing instructions, chain multiple AI tools together into a workflow, and let that workflow run with minimal supervision, producing something they can actually sell at the end of it: a finished product, a completed service, a piece of content ready to publish.
Same technology. Same monthly bill. Radically different results. That gap is not about intelligence, and it’s not about who paid for the more expensive AI plan — every serious AI product today is remarkably capable in the hands of someone who knows how to direct it. The gap is about what each person believes AI is for.
The Trap: Why Getting Better at Prompting Doesn’t Make You Richer
Here’s the uncomfortable part for a lot of people who’ve spent the last two years getting genuinely good at “prompt engineering” — learning how to phrase requests, how to ask follow-up questions, how to get better outputs out of a chatbot. It feels like progress. It is progress, in a narrow sense.
But asking better questions and getting better answers is fundamentally still a question-and-answer loop. It saves time on the task you were already doing. It doesn’t fundamentally change your relationship to that task. You’re still the one sitting down to do the work — you’re just a little less exhausted by the time you finish.
And time saved on a task you were already doing at the same rate doesn’t translate into more income. If you’re a freelance writer who now finishes an article in three hours instead of five, you’ve bought yourself two hours back — but unless you use those two hours to build something new, your income stays exactly where it was. Efficiency is not the same thing as an asset. An asset is something that keeps generating value for you even when you’re not actively working on it. A faster Q&A session with a chatbot isn’t one of those. It’s a productivity boost with a ceiling.
The Shift: From “Do This For Me” to “Run This For Me”
The people who are actually making meaningful income from AI made one specific mental shift, and it’s simpler than it sounds. They stopped asking AI, “Can you help me finish this one task?” and started asking, “Can you help me design a system where several AI tools handle an entire process, start to finish, without me being the bottleneck in the middle?”
แนะนำ
That’s the difference between using AI as a tool and using AI to build a business.
In practice, this usually looks like breaking a business process into stages and assigning an AI “role” to each one:
- An idea and strategy AI — researching topics, spotting trends, planning content or product angles.
- A production AI — actually writing the content, generating the designs, or building the product itself.
- A sales and customer service AI — answering customer questions and closing sales around the clock, with no lunch breaks and no time zones to worry about.
What used to require hiring four or five separate employees — a researcher, a writer, a designer, a customer service rep — can now be triggered by one person who has simply designed the workflow correctly and hits “go.” That’s not a hypothetical. It’s already how a growing number of solo entrepreneurs and small digital agencies operate, and it’s precisely the kind of leverage Kai-Fu Lee was warning about back in 2018: the jobs most at risk aren’t the ones requiring creativity or judgment, they’re the ones that are purely repetitive execution — exactly the kind of execution an AI pipeline can now absorb.
What an AI “Workforce” Actually Looks Like in Practice
This isn’t as exotic or technical as it might sound. A solo content creator, for example, might set up one AI to scan trending topics in their niche every morning, a second AI to turn the strongest of those topics into a full draft, and a third to repackage that draft into social media captions, thumbnails, and email newsletters — all before the creator has finished their morning coffee. A small e-commerce seller might have one AI drafting product descriptions, another handling customer messages on chat apps, and a third analyzing which products are trending so restocking decisions practically make themselves.
None of this requires the person to know how to code. It requires understanding how to break a goal into steps, and how to connect existing AI tools — many of which now offer simple automation features — so those steps happen in sequence rather than requiring a human to manually kick off each one.
The View From Southeast Asia: A Region Racing to Figure This Out
This shift isn’t unique to any one country — it’s playing out across Southeast Asia’s fast-growing digital economies in remarkably similar ways, even as the local details differ.
In Vietnam, a booming freelance and e-commerce scene has seen small sellers on platforms like Shopee and TikTok Shop increasingly lean on AI for product photography, listing copy, and automated customer replies, letting a handful of people run storefronts that would once have needed a full team. In Indonesia, with one of the largest gig-economy and social-commerce populations in the world, AI-driven content and chatbot automation have become a competitive necessity for small businesses trying to stand out in an incredibly crowded digital marketplace.
Thailand’s freelance and small-business ecosystem is following the same trajectory. Freelancers here increasingly use platforms like Fastwork — a popular Thai online marketplace where businesses hire freelancers for everything from graphic design to marketing — as a launchpad, then use AI to fulfill orders faster and take on more clients than they physically could alone. Small merchants and service businesses across the country have also started connecting AI chatbots to LINE Official Account (LINE OA) — a business tool built on top of LINE, the messaging app that functions in Thailand roughly the way WhatsApp does in much of the world — so that customer inquiries get answered instantly, at any hour, without a human needing to be on the other end of the chat.
แนะนำ
The pattern is consistent across all three markets: the earlier a solo operator moves from “AI as assistant” to “AI as team,” the bigger their advantage over competitors who are still doing everything by hand — and that advantage compounds the longer it goes uncontested.
You Don’t Need to Code — You Need a Different Question
The genuinely encouraging part of this story is that none of it requires a technical background, a computer science degree, or the ability to write a single line of code. It requires learning to think in systems rather than single tasks — and that is a skill, not an innate talent, which means it can be learned by anyone willing to put in the time.
It also doesn’t require building an elaborate ten-step pipeline on day one. Most people who make this shift successfully start small: three AI tools, each doing one clearly defined job, connected in a simple sequence. From there, the system gets refined and expanded as the person learns what actually works for their specific business or niche.
The Bottom Line
Kai-Fu Lee’s warning about AI displacing 40-50% of jobs within 15 years was never really about the technology being scary. It was about a gap opening up between people who use AI passively and people who use it structurally — and that gap only gets wider the longer someone waits to cross it.
If you’re currently in the first camp — asking AI one question at a time and calling it a day — the good news is that crossing over doesn’t require starting from scratch. It requires asking a different question the next time you open your AI tool: not “can you help me finish this,” but “can you help me build something that keeps working after I close this tab.” Whether you’re freelancing in Bangkok, running an online store in Jakarta, or working independently anywhere else in the world, that single shift in framing is the actual dividing line Kai-Fu Lee was pointing to all along.
Key Takeaways
- Kai-Fu Lee’s 2018 prediction that AI would displace 40-50% of jobs within 15 years is, by his own 2024 assessment, proving “eerily accurate.”
- The real divide among AI users isn’t skill or subscription tier — it’s whether they treat AI as a search tool or as a team they manage.
- Getting better at prompting only makes tasks faster; it doesn’t create the kind of recurring income an asset or system does.
- The income breakthrough comes from assigning AI distinct roles — ideation, production, and 24/7 sales/support — chained into one workflow.
- This shift is already visible across Southeast Asia, from Vietnam’s AI-powered e-commerce sellers to Thai freelancers automating client work via Fastwork and LINE OA.
Frequently Asked Questions
Q: Can you really make money with AI tools in 2026, or is this overhyped?
A: Yes, but the income comes from building AI-driven systems and workflows, not from simply chatting with an AI more often. People treating AI as a one-off Q&A tool see time savings, not income growth.
Q: What did Kai-Fu Lee actually predict about AI and jobs?
A: In 2018, he predicted AI could replace roughly 40-50% of existing jobs within about 15 years. In 2024 he confirmed the prediction was tracking “eerily” close to reality.
Q: Is Kai-Fu Lee’s prediction about AI replacing jobs still considered accurate?
A: According to his own 2024 remarks, yes — he described the forecast as holding up with unsettling precision given how AI adoption has accelerated.
Q: Why doesn’t getting better at using ChatGPT lead to higher income?
A: Improved prompting speeds up tasks you’re already doing, but it doesn’t create a recurring, sellable asset. Time saved without a new revenue system attached simply results in less effort for the same pay.
Q: What’s the difference between using AI as a tool versus using AI as a “team”?
A: Using AI as a tool means asking single questions and manually applying the answers. Using AI as a team means assigning different AI systems specific ongoing roles — research, production, sales — that run with minimal human intervention.
Q: Do I need to know how to code to build an AI-powered income system?
A: No. Building an effective AI workflow is primarily about structuring a process into clear steps, not writing software. Many no-code and low-code AI tools now support connecting multiple functions together.
Q: How many AI tools do I need to start automating my work or business?
A: Most successful setups start with as few as three AI tools, each handling one clear function, before being expanded as the operator learns what works.
Q: How are freelancers in Southeast Asia actually using AI to earn more?
A: Many use AI to speed up content creation, product listings, and customer service, allowing a single freelancer to serve far more clients than would be possible manually — a pattern visible on platforms in Vietnam, Indonesia, and Thailand alike.
Q: What is Fastwork, and how does it relate to AI monetization in Thailand?
A: Fastwork is a Thai online freelance marketplace connecting businesses with independent service providers. Freelancers on the platform increasingly use AI to complete more orders per day, effectively scaling their capacity without hiring help.
Q: What is LINE OA and why does it matter for AI-driven customer service?
A: LINE Official Account (LINE OA) is a business messaging tool built into LINE, the dominant chat app in Thailand. Small businesses connect AI chatbots to LINE OA to answer customer questions instantly, at any hour, without staffing a live agent.
Q: Is this AI income trend unique to Thailand, or is it happening elsewhere too?
A: It’s a regional and global pattern. Similar AI-driven automation is reshaping freelance and small-business work in Vietnam and Indonesia, and comparable shifts are underway among digital workers worldwide.
Q: What’s the first practical step someone should take to start earning with AI systems rather than just chatting with AI?
A: Identify one repetitive part of your current work, assign an AI tool to fully own that step, and connect it to at least one other AI-handled step — turning a single task into the start of a small automated workflow.
