I came into this semester with a question I hadn't quite let myself ask. What kind of designer am I trying to be? I had projects scattered across folders, half-finished concepts, ideas I liked but couldn't explain to anyone. The seminar gave me a structure to stop collecting and start curating.

Roman Sellner Novotný became my mentor and shifted the framing in a way I needed. The deliverable wasn't another isolated project. It was a body of work with a consistent voice, where each piece argued for the same thing in a different way.

That thing, it turned out, was speculative design. Not as a method I tried once, but as a position I was ready to defend.

The semester ran in parallel tracks. On one side, I worked through new speculative projects under Roman's facilitation. On the other side, I went back into older work and asked which pieces still earned their place when the standard was higher.

The most developed project of the semester was FutureSafe AI, a speculative scenario asking what happens if the state conditions social benefits on genetic analysis. The workshop produced three futures, optimistic, reflexive, and dystopian. The dystopian one stuck. We built a fictional police report documenting the case of a woman who refused mandatory genetic screening, and watched the conversation in the room shift from theory to a kind of low-grade panic.

That artefact taught me something I kept applying for the rest of the semester. The point of speculative design isn't to predict. It's to make the wrong future tangible enough to refuse, while there's still time to refuse it.

"The issue isn't the technology. It's the legal context it lands in."

— Workshop participant, FutureSafe AI

Around FutureSafe I built out the rest of the portfolio. Older speculative ideas got reworked. New experiments were tested as small artefacts rather than full projects. Pieces that didn't fit the voice got cut, even ones I was attached to. That cutting was harder than making.

Before this semester, I treated speculative design as one of the things I do. By the end, it was the lens I was looking through. Not because I decided it was. Because that's what the portfolio kept revealing once I forced myself to read it as a whole.

Roman's role was less about teaching me a method and more about giving me permission to take the position seriously. He treated speculative design as legitimate professional work, not a student curiosity, and that framing reset how I read my own output.

A portfolio with a point of view costs more to build than a portfolio of nice projects. You have to cut things. You have to defend choices. You have to live with the discomfort of saying this is what I do instead of I can do anything.

But it's also the only kind of portfolio that gets you the work you actually want.

The full FutureSafe AI case study, with the workshop process and artefacts, is documented in FutureSafe AI: Algorithmic control & ethics →