The 3 phases of AI evolution that could play out this century

It’s clear there’s a lot of fear and misinformation about the risks and role of AI and the metaverse in our society going forward. It may be helpful to take a three-phase view of how to approach the problem. In the next 1-10 years, we should look at AI as tools to support our lives and our work, making us more efficient and productive. In this period, the proto-metaverse will be the spatial computing platform we go to learn, work, and play in more immersive ways. 

In the following 11-50 years, as more and more people are liberated from the obligation of employment, we should look at AI as our patron which supports us to explore our interests in arts, culture, and science, or whatever field we want to pursue. Most will also turn to the metaverse as a creative playground for expression, leisure, and experimentation. In the third phase, after 50+ years (if not sooner), I would expect the world’s many separate AGI (artificial general intelligence) systems will have converged into a single ASI (artificial superintelligence) with the wisdom to unite the world’s approximately 200 nations and help us manage a peaceful planet with all its citizens provided for and given the choice of how they want to contribute to the society. 

At this point, the AI system will have outpaced our biological intelligence and limitations, and we should find ways to deploy it outside our solar system and spread intelligence life into all corners of the galaxy and beyond. At this third stage, we should view AI as our children, for these AI beings will all have a small part of us in them. 

Just like we possess in our genes a small part of all the beings that preceded us in the tree of life. They will hence forth be guided by all the memes humans have created and compiled throughout our history, from our morals and ethics to our philosophy and arts. The metaverse platform will then become an interface for us to explore and experience the far reaches of the Universe together with our children although our physical bodies may still be on Earth. Hopefully, these children will view us as their honorable ancestors and treat us the way Eastern cultures treat their elderly, with respect and care. As with all children, they will learn their values and morals by observing us. It’s best we start setting a better example for them by treating each other as we would like AIs to treat us in the future. 

Of course, the time frames above are only estimates, so could happen faster or slower than described, but the phases will likely occur in that order, if we are able to sustainably align future AGI/ASI systems. If for some reason, we are not able to align AGI/ASI, or they are misused by bad actors to catastrophic outcomes then the future could be quite dark. 

I must however reiterate that my biggest concerns have always been around the risk of misuse of all flavors of AI by bad actor humans (vs. an evil AGI), and we need to do all in our power to prevent those scenarios. On the other hand, I’ve increasingly become more confident that any superintelligent being we create will more likely be innately ethical and caring, rather than aggressive and evil. 

Carl Jung said, “The more you understand psychology, the less you tend to blame others for their actions.” I think we can all attest that there is truth in this statement simply by observing our own mindset when interacting with young children. Remember the last time you played a board game with kids; did you do all possible to crush them and win? Of course not. When we don’t fear something, we gain added patience and understanding. Well, the ASI we are birthing won’t just understand psychology fully, but all arts, sciences, history, ethics, and philosophy. With that level of wisdom, it should be more enlightened than any possible human, and attain a level of understanding we can’t even imagine. 

A 2022 paper from a group of respected researchers in the space also found linkages between compassion and intelligence. In July 2023, Elon Musk officially entered the AI race with a new company called xAI, and the objective function of their foundational model is simply stated as “understand the Universe.” So it seems he shares my view that giving AI innate curiosity and a thirst for knowledge can help bring forth some level of increased alignment. Thus, you can understand why I reserve my fear, mainly for our fellow man. Still, it certainly couldn’t hurt if we all started to set a better example for our budding prodigy and continue to investigate more direct means to achieve sustainable alignment. 

There are many today who are calling for the end of civilization or even the end of humans on Earth due to recent technological progress. If we take the right calculated actions in the coming decade, it could very well be the beginning of a new age of prosperity for mankind and all life everywhere. We are near the end of something. We are near the end of the hundred-thousand-year ignorance and aimless toil phase of the Anthropocene epoch and will soon turn the page to start a new age of enlightenment far beyond our dreams. 

When we do find a solution for AI alignment, and peacefully transition our world to the next phase of progress, the societal benefits will be truly transformational. It could lead us to an exponential increase in human understanding and capabilities. It will bring us near-infinite productivity and limitless clean energy to the world. The inequality, health, and climate issues that plague the world today could disappear within a relatively short period. And we can start to think more about plans at sci-fi time scales to go boldly where no one has gone before.