Let’s discuss the meaning of life. Yes, in a simple blog posting. Let’s just end all discussion, for once and for all. Hmmm… unlikely this will happen. But there is new relevance to the question of the meaning of life, as artificial intelligence (AI) develops. AI-enabled systems show autonomy to a certain extent. Systems learn and improve based on interacting with their environemt. Systems start to show early signs of ‘agency’, the capacity to act meaningfully in a certain context. With a little bit of fantasy, with AI we are creating, albeit simple, artificial life forms.
How do we want those systems to behave? How do we define what act meaningfully really means? Our discussion about the meaning of life has focused so far on trying to figure out what it means to us, human beings. But now there is an additional interesting angle: giving meaning to the artificial life we create.
In other words, the meaning of life as a set of design principles. How should an AI function (bet the best possible version of itself), and how should it relate to the rest of the world (how do we define what constitutes a positive contribution).
The problem is that 4,000 years of discussion hasn’t really helped finding definitive answers. In discussing this with regards to AI this won’t be different. From a philosophical angle, a number of perspectives present themselves.
The maker’s perspective would argue that the goal of AI is to function on a as human possible level. Think of the Google Duplex experiment, where an algorithm create a fully natural human voice in a phone conversation. Antropomorphism, the tendency to ascribe human characteristics to things, is not something to be frowned upon here, but actually a goal.
The humanistic perspective teaches us the meaning of life is to develop yourself as much as possible. Translated to AI it means machine learning never stops. AI should strive to function in an as broad as possible environment. Think of autonomous cars being able to deal with every traffic situation, but also other functions like car sharing, being part of overally traffic safety, solving traffic jams and so on. The end goal is to create an AGI (artificial general intelligence).
Custodianship is another perspective. The goal of life is to preserve our environment. In terms of AI, it should really focus on larger themes, such as climate change, or supporting other societal goals. Or, as a variation on the maker’s perspective, serving humanity, instead of the individual human being.
The meaning of life from a biological perspective is to procreate and to sustain. There are already examples of AI building AI. And some types of machine learning are based on darwinian evolution and natural selection. Checking which algorithms work best in practice and propagate those.
Lastly, I’d like to highlight the practical perspective. The meaning of life isn’t really relevant from this point of view. Technology simply has a certain task to perform and if AI helps doing a better job, that is wonderful. But nothing more.
What do you believe in? And how would that translate to your AI design principles?
One more thought. I discussed the meaning of life from a different angle: how we define meaning for the life we create. But does the meaning of life also change for humanity and human beings as we become creators of life ourselves? What does that mean for humanity and our own responsibilities?
Frank Buytendijk (@FrankBuytendijk) is a Research Fellow at Gartner Research & Advisory. He pioneers in the field of digital philosophy and ethics, the #digitalsociety and futurism.