AI brought me back to social media

After roughly a decade away, I’m back on social media. For good.

While “how I’m using social media” is probably an interesting post on its own, today I’ll cover “why I’m using social media.”

And it has everything to do with the AI revolution.

AI irreversibly changed how to learn

We’re here. We were all dreaming “Ugh I can’t wait until I can just get a computer to do this for me!” .. and now we can.

While obvious in hindsight, it’s hard to foresee the real consequences of computers doing things for us: I mean, if computers are doing stuff for us, then what are WE doing??

I noticed this clear theme in my conversations about AI with peers, friends and family: they know they need to “learn AI,” but they don’t quite know how to go about it. Lots of “how do you do this, Dui?” and “Oh, I should look into that” — whatever that is.

Well, that’s not gonna cut it.

The rules of learning have now completely changed.

• We need to learn extremely quickly
• There’s no curriculum — the information is scattered across the internet
• We have no idea what good looks like
• You’ll be outcompeted if you’re not one of the best

There are 2 important reasons related to AI that I’m back on social media, and the #1 has to do with learning. If I had to summarize my thesis on how learning in 2026 is completely different from 2025, it’d be:

“Books and practice were enough to become an expert on virtually any intellectual field, but AI moves so quickly that books are no longer sufficient to convey the culture and information needed for AI expertise.”

The immense RoI of books depends heavily on how quickly the field progresses. If a book takes 1–2 years to be written, you get an excellent deal: 100–300hrs of the author's work, consumed in 5–10hrs, at the expense of learning outdated content by 1–2 years.

In most fields, learning content that’s outdated by 1 year is practically negligible. In AI, it would mean learning AI before Claude Code existed.

But can’t you just learn on your own? Sure — the way you can solve a Rubik’s cube on your own. You’ll be much slower, and possibly not succeed. And in the AI world, they’re one and the same.

There’s a reason why humans socially evolved to build networks of information: it outcompetes information silos. Building knowledge on your own may be more fun (also an evolutionary response), but it will be much slower, and the world won’t wait.

And what got you here won’t get you there.

Can’t rest on your laurels anymore

A decade ago, I joined a company as a Sr. Engineer. I had 12 years of software development experience then, but I remember seeing a prolific Jr. Engineer submitting a PR on a new technology I was unfamiliar with and thinking “well shoot, I don’t know if my 12 years of experience make much of a difference for this new tech.”

That’s not to say that all my 12 years of experience back then were garbage. Just that as new technologies come out, your prior experience depreciates depending on how different the new state of the art is from your prior expertise.

AI has shifted this dynamic by an order of magnitude. It’s still valuable that you have expertise in your prior areas, but it depreciates very quickly, and your prior expertise won’t make up for your AI gaps.

In short, it is a bit like one big leveling of the field: the most valuable skills for us to learn need to be learned from scratch now, and your prior core competencies are now “just” valuable differentiators.

This is already true today: In software engineering, if you don’t know how to use AI, the professional market for you is virtually non-existent: a few niche roles, exclusively if you’re really prolific and well-known, but almost no company will hire a regular software engineer that won’t use AI in their day-to-day work.

Soon, and I mean very soon (months, not years), AI being a required competence will be true of virtually every role in the knowledge work sector: leaders, managers, PMs, TPMs, operators, analysts, recruiters, accountants.. everyone.

Importantly: if you are really competent at using AI, your prior expertise in programming, product management, management, etc will still be valuable and an important differentiator in your role. But just having the competence that made you excellent in your role in 2025 suddenly is completely irrelevant to even being qualified for it in 2026!

The overhaul in needed skills has an important implication to your current professional experience: it no longer says you’re competent to perform in your role.

Credibility in the AI world

The programming interview is broken and we don’t quite know how to fix it. Companies are still debating whether, and how much, to let people use AI to solve problems, and how to evaluate them when they do.

And that’s saying nothing of the application process itself. AI is submitting hundreds of job applications for candidates, recruiters are receiving thousands of applications per week, and AI is being used, with dubious effectiveness, to help people sort through this noise.

In short: AI broke the cycle we’ve run for the past 30 years — programmer applies, recruiter reads resume, hiring manager screens, team does a tech interview.

The “Marketing” and “Sales” process for software engineers is being overhauled by AI, and this will be true of every other role in months.

There's another shift happening: people used to be “roughly” interchangeable. Back when we were writing code by hand, you had 2x engineers, sure, but very few 10x engineers, and one 100x engineer in a blue moon. On average, a good hire was about as good as another good hire.

AI blows that up. The ceiling on individual value creation is now orders of magnitude higher, and the variance between people is exploding. We're going from a bell curve where most people were around the middle to a long tail where most people cluster at the left, and the few who break out can be 10x, 100x, or even 1000x more productive than the average engineer.

In this new AI world, the “personal brand” nice-to-have becomes essential. No one has been using agentic AI for several years. No one has leveraged multi-agents on several production products over the span of their career. What you’ve done before is a differentiator, but it doesn’t tell the main story of what you can do.

And that’s reason #2 that I’m back on social media. Not only you must be really good at what you do, you must show exactly how good you are to your target market, just like companies do. Resumes can’t keep up with a field that significantly changes every few months — credibility no longer depends so heavily on tenure.

What I’m doing about it

I spent a decade off social media and I don’t regret it — it was the right call for that version of the world. But the world changed.

As the famous adage goes: “When facts change, I change my mind. What do you do?”

So I’m back. Following the people who are pushing AI forward, learning in public, and sharing what I learn with others along the way.

This is both the beginning and a sign of things to come: In a world that’s rapidly changing, you must either rapidly change or be left behind.

I chose change. I hope you do too.