Out of alignment

 

Christopher Wallis is a scholar of religion and philosophy who also practices a form of Indian Tantrism. Here, he criticizes the notion that Artificial Intelligence (AI) can become conscious and somehow take over the world, exterminating all humans in the process. Or, in a more benign version, co-exist with us while enjoying all the usual human rights. (I can already see libertarians and cornucopians campaigning for the rights of robots.) Both scenarios are equally absurd, according to Wallis.

For starters, “consciousness” and “intelligence” are not the same thing. Consciousness simply means subjective experience. Your pet has it. Intelligence means problem-solving ability. A machine could have advanced problem-solving abilities (Wallis´ example is a computer that can solve problems associated with the complex strategy game “Go”) without having any consciousness at all. On a deeper level, Wallis´ objection is that materialism (here called physicalism) is simply wrong or at the very least unproven. The idea that AI can become conscious is clearly based on a physicalist metaphysic, in which consciousness arises (perhaps inevitably) once matter is organized in a particularly complex manner. This is what might turn robots or computers into Terminators (or benign right-holders). As a Tantric, Wallis is a metaphysical idealist. Consciousness is the basis of Reality and our “material” world exists *within* this universal consciousness, almost like a dream.

Wallis argues that consciousness is made up of four interrelated powers or “shaktis”: enjoyment, desire/will, cognition and action. There is also a kind of meta-capacity: autonomy or freedom. There is also a force Wallis calls “prana”, usually translated “life force”, but here more algebraically interpreted as “that which biological entities have in common”. AI of course isn´t biological. Only conscious beings develop desires that are in conflict with those of their “creators”. AI can´t do this. There is no “alignment problem” here, except in the sense that AI can be used (by humans) in ways detrimental to society. The game-playing computer mentioned earlier was actually defeated by a “lower-ranking” human player who assumed that the machine doesn´t really know what it´s doing!

Of course, if you believe in some particularly dark-side form of Tantrism, I suppose you could still argue that robots with positronic brains could be possessed by demons, and then it´s off to Terminator land anyway, but this particular scholar seems to be a moderate…