Introduction

There is a mooc in the University of Helsinki about the ethics of artificial intelligence. I just started it and it’s great.

The triumph of technicism

Us engineers tend to think that there is no problem (technical or else) that cannot be resolved by technological means. That’s called technicism.

Usually that’s right for small-scale problems or systems, but in the case of artificial intelligence, or being more precise machine learning that’s not true.

The machine is always right?

A machine learning algorithm does not act in a certain way because of how it’s been developed, but it acts in a certain way because of the data it’s been given. If the algorithm receives biased data the algorithm will be surely biased. Of course we can argue that there are some algorithms that are better than others but the main issue is to have a big-enough big-quality dataset.

Computers rule the world

In a world were corporations can access to/obtain big datasets, and are using massively machine learning to take decisions, we have to make sure there is no harm done to the society. As AI is more ubiquitous we have to make sure a minimal rules are laid out in the field to avoid discrimination, to keep users' privacy, and other issues that are arising from time to time.

Conclusion: but is not my fault!

Again, us as engineers and scientist cannot defend ourselves with the argument of “blame the doer, not the creator”. A new power, as powerful as the atom bomb is now among us, and we have to protect the human race. New laws should be made, but we cannot rely only on them and must made sure our creations make no harm by action or omission.