How I understood the importance of the alignment problem
The artificial intelligence research I'm deep diving..
On my holiday last December 2022, I decided to delve into why people are so alarmed about the potential disruption caused by artificial intelligence in the near future. Being a researcher on conciousness and morality, I thought venturing out further into why aligning AGI to human intentions has not taken off yet would make for an interesting project. During my scouring, I was able to find some enlightening material which made me lighten up on some of the heavier aspects of AI - such as why machines do not understand nor care if they do good or bad until it is being cued by its human controllers.
With an ever-increasing flow of AI alignment material being churned out every week, it can be an overwhelming task to take in all of it. That's why I like to approach this area of research from a first principles thinking perspective, focusing on foundational materials and only looking at those which fit an archetypal / jungian model on intelligence. Though admittedly this view isn't always 100% accurate - after all, artificial intelligence is still in its infancy stages and any theory utilized here must be able to hold up to intense scrutiny.
As many of us know, AI alignment matters - a lot. Focusing on the archetypes that make up human perception is key to achieving AI alignment with our moral principles. This month and the next, I'm going to dive deeper into this in my research paper, and discuss why archetypes are important for tackling goal misgeneralisation, reward hacking or power-seeking behaviour which continue to confront us in this technological advancement. I am on a mission to bridge the divide between AI and humans - and it's no easy task! Paying closer attention to archetypes and how they shape human perception, I'm confident we can finally align human intentions with AI and enter a technology-driven future that benefits us all.