Baru Obračajová

In this post, I would like to outline my key principles of how I utilize GenAI tools – primarily chatbots, although these can be extrapolated to vibecoding tools as well.

I know that if I am not going to be very mindful and intentional about this, I could get completely overwhelmed.

Therefore, I am writing these down so that I can hold myself accountable and refine them as the tech landscape changes.

I decide where GenAI belongs in my process

It is not a default; it is an extra that I can mindfully engage, conscious of its strengths and limitations.

I never use GenAI write the first draft

The act of writing is the biggest vessel of sensemaking in my life and a cornerstone of my humanity.

Yes, I still write even small emails by myself.

GenAI is my second pair of eyes

If there is a specific assignment, I do use GenAI to check if I have hit all the checkboxes.

I will ask GenAI to poke holes in my arguments and point out the parts where my writing lacks clarity or gets too long-winded.

That being said, if I want to check the fluidity of my sentences or typos, I only ask for a list of corrections, not a full rewrite of the text.

I only ask for variations once I have produced some on my own

Ideation and iteration are in my own personal opinion the most powerful use cases.

But I don’t want my brain to get lazy and sacrifice my creativity – hence I only ask for additional ideas and iterations once I have done my own thinking.

For example, when formulating design challenges. I have to produce quite a few variations to really, really reflect on a problem. Only then I am willing to use GenAI as a sparing partner on top of my input.

I do not use GenAI to make my decisions, but I do use it to facilitate the decision-making

I find it interesting to explore pros and cons of decisions with GenAI. It helps me overcome my own biases and explore decision-making strategies.

The final decision, its impacts and implications and how I act on the context – that is on me.

I do my translations myself, but ask for tweaks

I exist fully bilingually between Czech and English, with an occasional dash of other languages, so my life is a lot about switching.

I find it that no matter how I try to prime the models, I just do not like the translations. I see my writing style differ across the languages, and translating is not just a mechanical process.

I will, however, ask for help when translating concepts or buzzwords (like when I was translating Three Futures Frames into Czech and wanted to maintain the semantics and the aliteraiton).

If I am exporing, GenAI is one of the steps

It is only one of the steps, I do not come in for final answers.

A generated interface is not my destination.

I will gladly have papers recommended, but will follow the rabbitholes myself.

What did I forget?

My relationship with AI is something that I constantly navigate and tweak.

Therefore, I will probably amend these principles over time.

But I want to do so explicitly and intentionally, as I design my design work.

What would you add to the list?