Where I stand on AI now

I've gone back and forth on AI more than I'd like to admit. Some months I was all in, other months I rolled my eyes at every launch post. I think that's healthy. Skepticism should be the default with any new technology, especially one that arrives with this much hype. But skepticism only means something if you're willing to change your mind when the evidence shows up.

For me, it showed up.

The capability question is settled

A couple of years ago I spent three weeks on my first Blender project, learning modelling, texturing and animation from zero. This week I handed the same project to Opus 5.5 with one message and a handful of notes. Two days later it had rebuilt the whole thing: new models, nine shots, a proper colour grade, even a soundtrack, all written as code. It never opened Blender's interface once.

That's one example, but I see the same pattern everywhere now. Watch an agent work through a problem in the terminal and it chains together in seconds what would take me days to figure out. Interactive pages that used to be a month of work for a small team come together in an afternoon of back and forth. I can still find flaws, and there are flaws. But I can't honestly say the tech isn't good anymore. For me that debate is over.

What stays with us

What didn't go away is the eye. That remaster only got good because I kept pointing at what was wrong: the wheels look like plastic, the dust doesn't trail behind the truck, nothing in the animation has any weight. The model can make almost anything. Knowing what's worth making, and noticing when something is off, is still the job. If anything, taste matters more now, because trying an idea costs almost nothing.

There's also a real tradeoff with speed. The faster you move, the less attention goes to what has to hold up in a year. That's true for people, and it's true for agents that finish a complex feature in half an hour. Someone still has to care about the structure underneath.

Criticise the right things

I don't think the answer is to stop being critical. It's to aim the criticism at the right target. The technology itself is impressive. What deserves the scrutiny is everything around it:

• A handful of companies controlling access to the best models, and what that does to pricing
• Hype cycles and fear campaigns that are mostly marketing
• How training data was collected, and who gets credit or pay for it
• Who captures the value all this new leverage creates

The most useful thing we can push for is competition. As long as several labs keep outdoing each other, prices stay sane and access stays open. The moment that stops, this gets a lot less fun. I want a world where the same level of AI is affordable and available to everyone, not only to whoever can pay the most for it.

Work

The big replacement story hasn't matched what I see in my own work. What I see is leverage: one person can now carry what used to take a team. For a designer that's huge. I spend less time pushing pixels and more time deciding what should exist and why.

I also don't look back at the hand-made years with regret. Those three weeks in Blender taught me how light wraps around a surface and why one camera angle works and another doesn't. I'm glad I learned it that way, and I'm glad I don't have to do it that way anymore. Both can be true.

Where that leaves me

Still skeptical by default. Still annoyed by most AI marketing. But I use these tools every day, and pretending they aren't remarkable would be dishonest. If you're on the fence, or firmly against it, that's fine. I'd only ask that you spend real time with it, form your view from what you actually see, and be willing to move when the evidence moves.