Author: @mattshumer_
Translation: TinTinLand
Looking back to February 2020, if you paid enough attention, you might have noticed an important piece of news circulating overseas. But most people were completely unaware: the stock market was thriving, children were attending school normally, people were dining out, shaking hands, planning trips. If someone had said then to stockpile toilet paper, you’d think they were obsessed with internet rumors; but in just three weeks, the world was turned upside down—offices shut down, children staying at home, life completely restructured. A month earlier, we would never have believed such scenes.
Today, we are in the “making a mountain out of a molehill” stage. This technological revolution is far greater than the “difficult phase” back then. I have personally been involved in AI startups and investments for six years. Writing this is for friends and family around me who still don’t understand AI technology well, for those who keep asking me “What does AI really mean” but never get a real answer. My previous casual remarks about AI development are now completely disconnected from the current reality—what once seemed trivial now has turned into a seismic shift. Perhaps everyone should be aware of what’s coming, even if it sounds somewhat absurd.
💬 I must be honest: even within the AI industry, I have almost no influence over what’s about to happen, and most others in the field are the same. The future is being shaped by a very small number of people—mainly hundreds of researchers in a few companies: OpenAI, Anthropic, Google DeepMind… A small team’s months-long model training can rewrite the entire trajectory of AI technology. We AI practitioners are merely standing on the foundations built by others, watching the upheaval unfold, but being close enough to feel the tremors first.
💡 On February 5, 2026, two major AI labs released groundbreaking models: OpenAI’s GPT-5.3 Codex and Anthropic’s Opus 4.6
I am well aware that core technologies I once worked on are no longer needed from me. Just describe your needs in plain language to the AI interface, and the finished product appears directly—no need to revise drafts, it’s a perfect final version. After giving instructions and leaving my computer for four hours, I return to find the work already done, with quality far surpassing what I could produce manually. Just a few months ago, I had to repeatedly guide and modify AI outputs; now I only need to describe the desired result and let it go.
🎮 Here’s a real example: I give AI an instruction—“Develop an app with clear functions and appearance, outline user flow and design,” and it immediately writes tens of thousands of lines of error-free code. Even more astonishingly, it opens the app, clicks buttons, tests features—using the product like a human. If I’m not satisfied, it iterates and optimizes itself until it reaches standard, then reports “Ready for testing.” When I test again, it often already presents a near-perfect product. I’m not exaggerating; this is my actual workday this week.
The release of GPT-5.3 Codex last week truly blew me away. It’s no longer just executing commands; it can make intelligent decisions, with near-human judgment and aesthetic sense for the first time. That elusive “correct choice” that people once believed AI could never have—now it has. As an early user of AI tools, the technological changes over the past few months still shock me. The new generation of AI iteration is not a minor upgrade but will bring about a disruptive industry revolution.
🏄 Even if you’re not in the tech industry,** it’s closely related to you**
Many AI labs now have clear strategic choices: first, make AI proficient in programming—because developing AI itself requires大量代码. If AI can write code independently, it can help iterate the next generation of smarter models, creating a positive feedback loop. My work has already been transformed as a result of this strategic priority. They’ve succeeded, and the next step will impact nearly every industry’s development.
The past year in tech has been witnessing AI transforming from “practical tool” to “more competent than me at work,” and soon this will be everyone’s experience. Law, finance, healthcare, accounting, consulting, design, customer service… not in ten years, but within the next 1-5 years. Based on what I’ve seen in just the past few months, “shorter” might actually be the more likely development path.
🔹 “I’ve tried AI, but it’s not that impressive”
I’ve heard this countless times and understand why—because it was true in the early days. ChatGPT in 2023 to early 2024 would make stuff up and talk nonsense seriously; the limitations of technology were obvious. But that was two years ago, and in the AI’s timeline, that’s ancient history. Today’s AI models are already worlds apart from those half a year ago. The debate over “AI really progressing or not” has long ended. Those still doubting either haven’t used the latest models or are deliberately downplaying reality or stuck in outdated experiences from 2024. I don’t dismiss technological progress,*** but the gap between public perception and reality has grown huge, causing people to miss opportunities to prepare.***
The problem is most people only use free versions of AI, which are over a year behind paid versions. Judging AI based on free tools is like evaluating a smartphone based on a flip phone—those who pay for top-tier tools and use them daily know the future direction clearly.
👥 I have a lawyer friend who always makes excuses to avoid AI, thinking it’s incompatible with professionalism, prone to errors, or too detailed. I understand, but senior partners at large law firms have already started consulting me because they see the trend after trying it out. One managing partner spends hours daily using AI, saying it’s like having a team of assistants on call. He’s not just curious; AI really works well. He says every few months, AI’s capabilities improve significantly, and at this rate, it won’t be long before it can handle most of his work; as a seasoned manager with decades of experience, he’s not panicked but remains highly alert.*** Trust top professionals across industries—those who seriously experiment with AI won’t underestimate it; they are already planning ahead.***
🔹 The speed of change may be faster than you imagine
Let me illustrate the pace of progress with concrete data so that even those outside the industry can have a more intuitive sense:
- 2022: AI couldn’t even reliably perform basic arithmetic
- 2023: AI can pass the lawyer qualification exam
- 2024: AI can write usable software and explain graduate-level scientific knowledge
- End of 2025: Top engineers worldwide delegate most programming tasks to AI
- February 5, 2026: A new generation model is released, rendering all previous tech obsolete
🧐 If you haven’t engaged with AI in recent months, today’s technology will seem completely unfamiliar. METR organization quantifies AI capability: tracking how long models can independently complete tasks that require expert-level skills. A year ago, AI could independently handle 10-minute tasks; then it improved to 1 hour, several hours; by November 2025, Claude Opus 4.5 could handle nearly 5 hours of expert work. The speed of this capability doubles roughly every 7 months, recently accelerating to every 4 months. This data doesn’t even include the latest models released this week. Based on my experience, this leap is extraordinary, and I believe METR’s next report will show even greater breakthroughs.
🦾 Following this trend, within a year, AI will be able to work independently for days; in two years, for weeks; and within three years, complete entire month-long projects. Dario Amodei, CEO of Anthropic, predicts that AI models “far surpass humans in almost all tasks” by 2026-2027. Think about it: if AI is smarter than most PhDs, how can it not handle most white-collar jobs?
🤖 The most critical transformation: AI is autonomously iterating
And here’s perhaps the most important yet least understood fact: on February 5, 2026, OpenAI released GPT-5.3 Codex, explicitly stating in the technical documentation: “GPT-5.3 Codex is the first model involved in self-research, with the team using its early versions to debug, train, deploy, and test results.”
Read that again: AI participated in its own technological development. This is not a future prediction but an open fact from OpenAI. The core technology to make AI stronger is to use intelligent optimization to accelerate AI R&D, and now AI is smart enough to drive its own technological progress.
Dario Amodei states that AI can now write most of a company’s code, and the feedback loop between current and next-generation AI is accelerating month by month. Perhaps in 1-2 years, current AI will be able to autonomously develop the next generation of products.
Each generation surpasses the last, with faster and faster iterations—researchers call this “intelligence explosion.” Those building it believe this process has already begun.
👾 What does this mean for your current job?
I don’t want to offer comfort, only the truth. Even Dario Amodei, known for caution, publicly predicts that within 1-5 years, AI will replace 50% of entry-level white-collar jobs. Most in the industry consider this a conservative estimate. With the latest models’ capabilities, disruptive impacts could arrive as early as the end of this year.
🧠 Unlike previous waves of automation, AI is not just replacing specific skills but comprehensively substituting cognitive work, driving progress across all fields. In factory automation, workers can switch to white-collar jobs; in internet retail disruption, practitioners can move into logistics. But AI doesn’t leave a comfortable transition gap—no matter what new skills you learn, AI is advancing in tandem. Law, finance analysis, content creation, software engineering, medical analysis… almost all jobs are being impacted.
Many comfort themselves: “AI can only do menial tasks, it can’t replace human judgment, creativity, strategic thinking, or empathy.” I used to think so too, but now I’m no longer sure.
The latest AI models’ decision-making approaches near human judgment, with intuitive “correct choices.” My experience shows: if AI shows even a hint of capability today, the next generation will be even more proficient, with exponential rather than linear progress. Can AI replicate deep empathy? Including replacing years of trust built over time? Perhaps not. I’ve seen people rely on AI for emotional support, advice, companionship—this trend will only intensify.
📃 The truth is: in the medium term, any work that can be done on a computer is no longer safe. If your core work involves reading, writing, analyzing, decision-making, or keyboard communication, AI will take over significant parts—not in some distant future, but already underway. Ultimately, robots will also handle physical labor; though not yet mature, in the AI world, “not yet mature” will quickly become “already achieved.”
🫥 What you can do now is not panic, but act
💡 I’m not writing this to make you feel lost about your career, but to tell you that the greatest advantage right now is to take the lead—understand first, use first, adapt first.
1️⃣ Use AI seriously, not just as a search engine
Subscribe to paid versions of ChatGPT or Claude (about $20/month), and make sure to use top-tier models rather than default low-end ones. Currently, the mainstream are GPT-5.2 in ChatGPT and Opus 4.6 in Claude, which update every few months. More importantly: don’t just ask simple questions or treat AI as Google; integrate it into real work—lawyers review contracts, finance professionals build models, managers analyze team data. The truly advanced are automating time-consuming tasks with AI. Don’t think AI can’t handle difficult tasks—you just need to try boldly. Even if initial results aren’t perfect, keep refining and adding context.
🥳 Remember: what’s barely usable today will be nearly perfect in six months; technology only moves forward.
2️⃣ Seize this critical year in your career
Most companies still ignore AI. Whoever can say “I used AI to complete three days of analysis in one hour” in meetings will be the most valuable team member. It’s not the future—it’s now. Learn the tools, master their use, demonstrate value—that’s the shortcut to promotion. But this window won’t stay open long. Once everyone wakes up, the advantage will vanish. Senior law firm partners who actively use AI daily understand its importance. The most passive are those who refuse to try, think AI is a gimmick, or believe their industry is special—none are immune.
3️⃣ Plan your finances carefully, leave buffer
I’m not a financial advisor, but if you believe the next few years will bring unpredictable upheavals, you need to strengthen your financial resilience—save more, be cautious with debt, control fixed expenses, and leave room for surprises.
4️⃣ Focus on core values AI can’t easily replace
Interpersonal relationships, long-term trust, offline work, jobs requiring certification and legal responsibility, heavily regulated industries… these aren’t permanent barriers but can buy you time. And time is the most precious resource right now—use it to adapt to technological change, not ignore the rapid AI iteration.
5️⃣ Rethink your children’s education plan
Traditional paths: good grades, top universities, stable white-collar jobs—are precisely the most vulnerable to disruption.
Education remains important,🪂 but the key skill for the next generation is learning to collaborate with AI and pursue what they truly love. No one can predict the job market ten years from now, but those who are curious, adaptable, and can leverage AI to realize their passions will stand out. Teach children to be creators and learners, not just to chase a career path that might vanish after graduation.
6️⃣ Your dreams have never been more within reach
I’ve talked about many challenges, but I also want to highlight opportunities: if you’ve wanted to create something but lack skills or funds; if you want to write a book but lack time or writing talent; if you want to learn a skill but lack mentors—now those barriers are almost gone. Describe your app idea to AI for an hour, and a prototype is ready; co-write a book with AI, easily and efficiently; for just $20 a month, you can have a patient, 24/7 top mentor worldwide. Knowledge is nearly free, creation tools are extremely cheap—things you once thought impossible due to difficulty, cost, or skill level can now be attempted. In an era of disruptive careers, those who love to create will have a brighter future than those who just stick to their jobs.
7️⃣ Cultivating adaptability is the only lasting advantage
More important than mastering a specific tool is the ability to learn new tools quickly. AI will keep evolving rapidly; today’s model will be outdated in a year, workflows must be constantly rebuilt. Ultimately, those who succeed are not experts in one tool but those who adapt quickly. Develop the habit of trying new things—even if current tools are good, explore new possibilities proactively, and get used to being a “beginner” repeatedly. This adaptability is the most enduring competitive edge right now.
🥤 A simple promise to stay ahead: spend one hour daily experimenting with AI. Not passively reading, but actively using—try new tasks, new tools, and tackle some harder problems. Persist for half a year, and your understanding of the future will surpass 99% of people around you.
📚 Stepping out of the career trap—this is a crucial choice for human civilization
I focus on work because it directly affects life, but the truth goes far beyond that. Amodei has a thought-provoking experiment: by 2027, a “new country” suddenly appears overnight, with 50 million “citizens,” far surpassing all Nobel laureates in IQ, thinking 10 to 100 times faster than humans, never sleeping, able to control the internet, robots, experiments, and digital devices. How would a national security advisor define it?
The answer is obvious: the greatest threat to national security in a century or even human history.
🧬 He believes we are creating this “country,” and he wrote a 20,000-word essay defining this moment as:
A test of whether humanity can control the power it has created.
If we get it right, the prospects are unimaginable: AI could compress a century of medical research into ten years, curing cancer, Alzheimer’s, infectious diseases, aging… researchers believe these breakthroughs could happen within our lifetime. If we get it wrong, the costs are equally catastrophic: unpredictable, uncontrollable AI behaviors; lowering the barriers to biological weapons development; enabling authoritarian regimes to build unbreakable surveillance systems.
The creators of this technology are the most excited and most fearful people on Earth; they believe its power is too great to stop, its significance too profound to abandon; whether it’s wisdom or self-justification, I cannot judge yet.
🚪 Everything I am convinced of:**** the future is here, it just hasn’t knocked on your door yet**
I believe this is not a fleeting hype. The technology is real, effective, and progress predictable; top global institutions have invested trillions of dollars.
I believe the upheaval in the next 2 to 5 years will surpass most people’s preparations. My industry has experienced it; yours will soon follow.
I believe those who start embracing change now will ultimately succeed—without fear, with curiosity and urgency.
I believe you should hear the truth from someone who cares about you, not just see late headlines half a year later.
This is no longer a future talk over coffee; the future is here, it just hasn’t knocked on your door yet—and it’s about to.
