Traceable AI workflows

I wanted to be more efficient and ended up running errands between several artificial intelligences. I copied, pasted, carried files from one window to another and got dizzy. On top of that, the answers contained mistakes and sometimes the review added more. I did not understand everything I was doing either. We had to stop and work out where we had gone wrong.

I started recording the journey until it took the shape of a lab0ratory: a kind of living logbook, a meta-manual that tells how it is being made. I left instructions to copy, reasons behind decisions and questions that remained unresolved. I appear and make mistakes in the manual’s own exercises too. (In my defence, I was tired and had a whole explanation for the thing about “proportional”. Anyway.) It all went down on paper.

I am sharing it so that the desire to make something does not have to wait until we know how.

It allowed me to work on something I was still learning to do. Assign a part, review what came back, stop to study what I did not understand. Some answers got work done; others made me think harder about the question. It opened doors and windows for me. I would like someone to use this manual to follow a curiosity of their own until they make something with it.

As for the name, I wanted something of these times to seep into language. One of my first poems was called three poets on a train and already played with that idea. I liked finding it again here. After all, it is a LAB0R4T0R10, but LAB0 is catchier.

Like:
Shall we start a LAB0 about…?