I was speaking with a colleague the other day, and we were all doom-and-gloom about how more and more people are trusting the black box of AI to tell them what is true and what is not. How chat that answers your questions based on the internet information and post-training done by god-knows-who is the culmination of social media marketing’s decades-long project of directing and engineering topics, opinions and debates. This black box now controls how we think, what we think, what arguments we bring and how we bring them.
He said we, as a society, a community, have to pay more attention to the ethical questions. That the tech needs more regulation. More ethical boards. More attention to the topic.
But I think it’s too late for that. Too late not in a sense that technology has gone too far, and there’s no stopping the popularity of it, or the capitalism for what it’s worth (all true points by the way).
My point is that “ethical” attitude should have started earlier, and been directed towards people, not technology. Tech is something we create, we as human beings. And one way or another, we encode the little part of who we are into it. Be that a machine that saves lives, because its maker lost one, a to-do app, because we felt disorganized or an answer box, because chatting our way to the truth sounded fun.
One could argue that even the most ethical engineers have created unethical tech. True. I think it’s similar to how complex systems operate - the failures come from the interaction between components, more than from individual components not working. However, step 1 is for each component to work. For each person inside the development of the tech to have ethics and moral. And for their each decision to be in line with ethics. I won’t try to define here what ethics exactly means; my claim is about where ethics must live - in the judgment of makers, not the paperwork around products.
Because no checklist covers for thousand micro-decisions - so you need a formed judgement where the checklist runs out. This is why compliance regimes and ethics boards feel so hollow - they’re rule-matching trying to do a job only judgement can.
And so if the person created the tech, has the formed morality and ethics that will at least have a chance to create an equally ethical technology…Wouldn’t they pay attention to every step in the process, at those little ethics-induced steps would eventually compound into one ethical technology? Component integrity is necessary, not sufficient.
The entire responsible-tech apparatus runs on two ethical traditions: rules (codes, audits, regulation) and outcomes (risk-benefit assessments). Character - the third and oldest tradition - is simply missing from the debate.
How we get there - how we raise ethical technologists - is a whole other problem, and not mine to design. That blueprint belongs to educators, and I hope they’re already sketching. But my own argument tells me three things about it for sure. It starts early, in school and all the way through university, long before anyone writes their first line of code - because ethics is not innate, it’s acquired the way crafts are. As Aristotle put it: we become just by doing just acts, brave by doing brave acts. The word is habituation. It’s practice, not lectures - you can’t teach the judgment for a thousand micro-decisions from a slide deck, for the same reason no checklist covers them. And it’s communal - because if the failures live in the interactions between components, the ethics can’t live in one lonely engineer.
Next time my colleague and I go doom-and-gloom about the black box, I’ll tell him he’s right: we should pay more attention to the ethical questions. I’ll just add that we’re pointing that attention at the wrong end of the pipeline. Not at the technology. At the people - starting with the ten-year-olds who will one day build it.