Note — Jan 30, 2022

Morals in the Machine

For some obscure reason, I’d never read anything by jenny (phire) zhang before, and that’s a real shame. Fantastic piece, very clearly written. She starts from advice given to her as a manager; to delegate things she’s good at and work on the things she needs to practice. zhang then wonders why we would try to build moral machines, considering that ‘we’ are prone to very imperfect morals. It’s a well made point worth reading–integrating ideas from Safiya Noble, Virginia Eubanks, Ursula Franklin, and Ruha Benjamin in the process–and also a sharp look at AI in general, including great and very understandable definitions of machine learning and “trustlessness.” I wanted to write more but really, she’s got everything in the last paragraph, do click through.

There’s no such thing as a universal code of ethics; the concept of justice cannot be extricated from the contextual value system it’s applied in. It seems unlikely that machine learning engineers will be able to reconcile what moral philosophers have been yelling at each other about for thousands of years, with no consensus in sight. […]

The mere fact of technical systems seems to make us want to trust in their virtue, maybe even at the cost of our own judgment. It might be learned helplessness engendered by the complexity of computers, or it might simply be bias laundering: evading culpability for harms caused by machines by pretending the outcomes are neutral simply because they are digital, instead of taking accountability for the tools we build. […]

Our tendency to trust the decisions of computer systems as perfectly rational despite ample evidence to the contrary feels like an abdication of our responsibility to reciprocate. If our fear of the chaos raging around us leads us to put our trust in machines so that we don’t have to trust one another, we relinquish the reciprocity that lets us advance our shared humanity, and those bleak assumptions about humans being immutably selfish turn into self-fulfilling prophecy. […]

We are so excited by the idea of machines that can write, and create art, and compose music, with seemingly little regard for how many wells of creativity sit untapped because many of us spend the best hours of our days toiling away, and even more can barely fulfill basic needs for food, shelter, and water.