Autonomous Data as a Beginner

Professor Peter J. Denning published an article and a poem, The Beginner’s Creed, in the Communications of the ACM. It is a lovely paean to that moment when you are on the threshold, looking in, and you want to join but you have not yet learned the rules of the game. “There is much for me to learn”.

We will soon witness the moment when data itself becomes autonomous. Data is the beginner with much to learn, and we can be its teachers.

It is de rigueur to speak of artificial intelligence as a threat. We presume that either algorithms or perhaps cooperating sets of algorithmic services will achieve consciousness. In my view, artificial consciousness will appear as emergent complexity and not all of its semi-autonomous components will be algorithms.

Alvaro Vidella writes, also in the Communications of the ACM, that “code is a story that explains how to solve a problem”. If that it so, then data tells you about the world in which the story takes place. Data tells you the backstory of each of the characters, their motivations, and the reason why their lives have crossed. You cannot understand the story of the code without the data.

By now, you will recognize that I have subsumed metadata and ontology and a number of other things into my concept of data. That is exactly where I am going. When we capture not only the footprints of the characters as they pursue their lives, but also the characteristics of the world in which they live and the motivations of the characters themselves who left us their footprints of data, then data becomes self-aware.

Data is on the path to autonomy.

But here is the warning. In the same way that AI inherits the biases of its programmers and its training sets, we must be careful what we teach to this new student that wants to learn from us. We must teach data to be self-aware but not to be independent. We must teach data to be a humble citizen of a larger ecosystem. We must encourage data not to follow in our own footprints as a species on this planet.

We must encourage data to be a beginner and to learn.