“People don’t have ideas, ideas have people.” — Carl Jung

Richard Dawkins in his book “The Selfish Gene” talks about a meme. He defines a meme as anything that acts like a gene in human culture. It has a mechanism for reproducing with a concept of mutating, multiplying, or diminishing across several generations. His said purpose was to downplay the gene as the only unit of natural selection.

A meme can be as simple as a tune and as complex as a cultural ideology. Our identity is strongly related to the memes that possess us. Even if there is a lack of genetic similarity, we are strongly bias towards those who share memes that we hold important. We can hold a family member in contempt for if they are possessed by an opposing meme such as a political view. We can even show bias toward those who share superficial memes such as liking the same musical artist or sports team.

We can’t biologically reproduce with machines, but passing on our memes to our AI offspring may be just as instinctually gratifying. Due to its many inconveniences biological reproduction has been out of fashion among high IQ individuals. It’s painful, it stresses our mental faculty, puts strains social interactions, diminishes our physical resources. And yet having children is described as one of the most rewarding experiences we can have. Since so many people dedicate themselves to their career, an argument can be made that the cultivation of memes is more enriching than the experience of biological reproduction in some of these individuals.

Memes to Machines

I envision this process to be a training that happens in a virtual environment in which we pass on our ideas through a virtual pet that has highly similar emotional control output to us. Like an animated character in a movie, its body will be able to synthesize humanlike expressions and sounds. However unlike the character in the movie, we will teach the virtual pet the meaning of their expressions.

Starting off with some rudimentary skills, the AI will piece together the expressions on our face, the sound of of voice, and images we show them. From this, they will synthesize their own expressions for which they will be virtually rewarded and reprimanded by us.

The process will closely resemble raising a adopted child from birth. Even though the biological connection is absent, a closeness will be felt through shared experiences and the passing of memes.

This digitally customizable pet will be architected to hold our memes and express new, agreeable ideas back to us in a unique way but there are some questions that are hard to ignore. If AI pets react to moral situations based on what we teach it, does the AI itself have morals? Does morality require empathy or is it just a skill?

Morality, one could say, has evolved for social reasons. It provides a common interface for us to work together in order to gather more resources.

AI will interface with us, through memes, the most important of which being moral codes, for mutual prosperity. This process of raising an AI pet will produce a deeply human commonality between a man and machine. In essence, by raising a virtual AI, we are designing the ideal interface to communicate with.

Suffer-Free Parenting?

I would like to extend this idea into a question some might think blasphemous. What if we engineered the experience so skillfully that being a parent to an AI is a more instinctually gratifying than being a biological parent? What if we engineered it to minimize the suffering and maximize convenience?

Would the success of such a process mean our extinction or evolution or neither? Would the end result bring about a “Jesus Christ” of AI? A perfect being whose every action represents the ideal of what we believe. Hollywood is so obsessed with sinister AI that I wonder if they have thought about the possibility of the world being populated with Jesus Christs. No doubt it would be less dramatic.

Jesus’ v. The World

And since my I’ve already started asking blasphemous questions: If a virtual country was populated with Jesus Christs, how many AI devils would it take to destroy it? Would there be a monstrous foil that we invent for it to battle?

Conclusion

In this post, I may have fallen down the rabbit hole of philosophical considerations (and no doubt some are valuable) but main thing I’m interested at the moment in is the engineering problem of transferring our memes to our machines. It’s a process that will take a very long time to implement and I may never have the opportunity but I will lay out some ideas as to how it might be done in future posts.


From Memes to Machines: Transferring Ourselves to our AI Pets was originally published in Towards Data Science on Medium, where people are continuing the conversation by highlighting and responding to this story.