Hello World, I’m Aline!

This is my output from a noisy input.

As an engineer who grew up taking toys’ gears apart just to see how they worked, who has spent 4+ years in industry building applied AI systems, and who holds a Master’s in Deep Reinforcement Learning, it’s genuinely uncomfortable (and honestly a little embarrassing) that I still can’t clearly say how AI systems actually work, or claim that I truly understand what’s going on inside them.

When family and friends ask, “Oh, you work with AI! Tell me how it works,” it always hits me hard. They look at me expecting the gears to fall neatly into place. Instead, while giving the usual cliché answer, I catch myself wondering whether the matrix multiplication thing has even convinced me.

Over time, I realized this isn’t just a personal frustration. It’s something the field itself is starting to confront more openly, now framed as questions of AI safety and AI misalignment. I suspect the gap between performance and understanding is one of the risks we’re still underestimating.

I’m worried about what this gap could mean at a global scale with potentially catastrophic implications, but I’m also genuinely excited to take part in working toward better answers.

This site is my attempt to gather my ideas, learnings, and experiments as I explore these questions. You can navigate it on the left and, if you’d like, join me along the way.