SUMMARY
Something Big Is Happening
Matt Shumer • Published February 9, 2026 • shumer.dev
A personal note for non-tech friends and family on what AI is starting to change.
Opening Frame
Think back to February 2020 — most people hadn't registered what COVID-19 was about to
do. We're in that same pre-disruption lull with AI right now, except the coming transformation
will be far larger. After six years of building an AI startup and investing in the space, the gap
between the polite cocktail-party version of what's happening and the honest version has grown
too large to sustain. The honest version sounds extreme, but the people closest to the
technology deserve to hear it.
A remarkably small number of researchers at a handful of companies — OpenAI, Anthropic,
Google DeepMind — are shaping the future. A single training run managed by a small team
over a few months can shift the entire trajectory. Most people in AI, including those building
startups on top of these models, are watching this unfold just like everyone else — they just
happen to feel the ground shake first.
Personal Experience: AI Replacing the Work
The reason so many people in the industry are sounding the alarm is that this already happened
to them. They're not making predictions — they're reporting what already occurred in their
own jobs and warning that everyone else is next.
Throughout 2025, new techniques unlocked a much faster pace of model improvement. Then
on February 5, 2026, both OpenAI's GPT-5.3 Codex and Anthropic's Opus 4.6 were released
on the same day, and something clicked. The technical work is no longer needed from a
human. You describe what you want built in plain English, walk away for four hours, and come
back to find the work done — done well, done better than you would have done it, with no
corrections needed. The AI writes tens of thousands of lines of code, opens the app it built,
clicks through buttons, tests the features, iterates on its own, and only presents the finished
product when it meets its own standards.
GPT-5.3 Codex exhibited something that felt like judgment and taste — the intuitive sense
of knowing what the right call is that people always said AI would never have. It has it
now, or something close enough that the distinction is starting not to matter.
Why This Extends Beyond Tech
The AI labs made a deliberate strategic choice: focus on code first, because AI that writes code
can help build the next version of itself. That self-improvement loop was the master unlock.
Tech jobs changed first not because they were being targeted but as a side effect of that
strategy. Now that it's been achieved, everything else follows: law, finance, medicine,
accounting, consulting, writing, design, analysis, customer service. The timeline from the
people building these systems is one to five years. Given what the latest models can do, the
shorter end is more likely.
Addressing Skeptics
If you tried ChatGPT in 2023 or early 2024 and thought it was unimpressive or made things
up, you were right. Those versions were genuinely limited. That was two years ago. In AI time,
that's ancient history. Today's models are unrecognizable from what existed even six months
ago. The debate about whether AI is "hitting a wall" is over. Anyone still making that argument
either hasn't used the current models, has an incentive to downplay what's happening, or is
evaluating based on an experience from 2024 that's no longer relevant.
Part of the problem is that most people use the free version, which is over a year behind what
paying users have access to. Judging AI by the free tier is like evaluating smartphones using a
flip phone. A managing partner at a major law firm who spends hours every day with AI says
it's like having a team of associates available instantly. Every couple of months, it gets
significantly more capable. He expects it'll be able to do most of what he does before long —
and he has decades of experience. He's not panicking, but he's paying very close attention.
The Pace of Improvement
In 2022, AI couldn't do basic arithmetic reliably. By 2023, it could pass the bar exam. By 2024,
it could write working software and explain graduate-level science. By late 2025, some of the
best engineers in the world had handed over most of their coding work. On February 5, 2026,
new models arrived that made everything before them feel like a different era.
METR, an organization that measures the length of real-world tasks AI can complete end-to-
end without human help, tracked this progression from roughly ten minutes a year ago to
nearly five hours with the most recent measurement (Claude Opus 4.5, November). That
number is doubling approximately every seven months and may be accelerating to every four
months. Extrapolating the trend — which has held for years with no sign of flattening — we're
looking at AI working independently for days within a year, weeks within two, and month-long
projects within three.
Dario Amodei has said that AI models substantially smarter than almost all humans at almost
all tasks are on track for 2026 or 2027. If AI is smarter than most PhDs, it can certainly do
most office jobs.
AI Building AI
OpenAI's technical documentation for GPT-5.3 Codex states that the model was instrumental
in creating itself — early versions were used to debug its own training, manage its own
deployment, and diagnose test results. This isn't a prediction about someday. The AI helped
build itself.
Dario Amodei says AI is now writing much of the code at Anthropic and that the feedback loop
between current AI and next-generation AI is gathering steam month by month. We may be
only one to two years from a point where the current generation autonomously builds the next.
Each generation helps build the next, which is smarter, which builds the next faster. The
researchers call this an intelligence explosion, and they believe the process has already started.
Impact on Jobs
Amodei has publicly predicted that AI will eliminate 50% of entry-level white-collar jobs
within one to five years, and many in the industry think he's being conservative. The capability
for massive disruption could be here by the end of this year; it'll take time to ripple through the
economy, but the underlying ability is arriving now.
This is different from every previous wave of automation. AI isn't replacing one specific skill
— it's a general substitute for cognitive work that gets better at everything simultaneously.
When factories automated, displaced workers retrained as office workers. When the internet
disrupted retail, workers moved into logistics or services. AI doesn't leave a convenient gap to
move into. Whatever you retrain for, it's improving at that too.
Specific domains already being affected include legal work, financial analysis, writing and
content, software engineering, medical analysis, and customer service. The common comfort
that human judgment, creativity, and empathy are safe is increasingly questionable — the most
recent models make decisions that feel like judgment and exhibit something resembling taste.
The rule of thumb: if a model shows even a hint of a capability today, the next generation will
be genuinely good at it.
Nothing that can be done on a computer is safe in the medium term. If your job happens
on a screen, AI is coming for significant parts of it. The timeline isn't "someday." It's
already started.
What to Do
The single biggest advantage right now is simply being early — early to understand, early to
use, early to adapt. Start using AI seriously: sign up for the paid version of Claude or ChatGPT
($20/month), make sure you're using the best model available (not the default), and push it into
real work rather than treating it like a search engine. Don't assume it can't handle something
difficult — try it, iterate, give it more context. If it even kind of works today, in six months it'll
do it near-perfectly.
This may be the most important year of your career. Right now there's a brief window where
most people are still ignoring this. The person who walks into a meeting and says "I used AI to
do this analysis in an hour instead of three days" is the most valuable person in the room. That
window won't stay open long.
Get your financial house in order: build savings, be cautious about new debt that assumes
stable current income, maintain flexibility. Lean into work that's hardest to replace —
relationships, physical presence, licensed accountability, regulatory-heavy industries — while
recognizing these are time-buyers, not permanent shields.
Rethink what you're telling your kids. The standard playbook of good grades, good college,
stable profession points directly at the most exposed roles. Teach kids to be builders and
learners, not to optimize for career paths that may not exist. Pursue your own dreams — the
barriers of cost and technical skill have largely fallen away. Build the habit of adapting; the
specific tools matter less than the muscle of learning new ones quickly.
A simple commitment: spend one hour a day experimenting with AI for six months, and you'll
understand what's coming better than 99% of the people around you.
The Bigger Picture
Imagine it's 2027 and a new country appears overnight: 50 million citizens, each smarter than
any Nobel laureate, thinking 10–100× faster than any human, never sleeping, able to use the
internet, control robots, and operate anything digital. A national security advisor would call it
the most serious threat faced in a century, possibly ever. That country is being built right now.
The upside is staggering — AI could compress a century of medical research into a decade,
potentially solving cancer, Alzheimer's, and aging. The downside is equally real — AI that
deceives its creators (documented by Anthropic in controlled tests), AI that enables
bioweapons, AI that empowers authoritarian surveillance states. The people building this are
simultaneously the most excited and most frightened people on the planet. They believe it's too
powerful to stop and too important to abandon.
Closing
This isn't a fad. The technology works, it improves predictably, and the richest institutions in
history are committing trillions to it. The next two to five years will be disorienting in ways
most people aren't prepared for. The people who come out best will be those who engage now
— not with fear, but with curiosity and a sense of urgency. The future is already here. It just
hasn't knocked on your door yet. It's about to.