|
I’ve long had an interest in the nature of consciousness—to the point where, a few years ago, I published a novel about the idea of its emerging spontaneously in the silicon-based primordial soup that is the world’s computer network. That was fiction. But, as artificial intelligence (AI) leaps over more of the conceptual hurdles erected by philosophers to distinguish it from the natural, human sort, the question of what exactly consciousness is becomes increasingly pertinent.
As I write this, I am in New York, having attended the 26th meeting of the Association for the Scientific Study of Consciousness. This was the occasion for the announcement of the results of a competition that had bravely set out to test against each other two of the leading theories about how consciousness is generated in the brain, in what was deemed an “adversarial collaboration” between proponents of both. The result was not as clear-cut as the organisers had hoped, but it did lead to the resolution of a bet made 25 years ago by two people who were then rising whippersnappers in the field, and are now established panjandrums.
That aside, some of the most interesting sessions of the conference were those asking questions about
which animals besides human beings are conscious,
and how (and how easily) AI might be given the extra dimension of consciousness.
Such musings raise important questions. That many, and perhaps all, mammals and birds have consciousness is a widely held view, and colours the way people treat them. Though these animals are routinely exploited and killed for human benefit, behaviour towards them that is perceived as being needlessly cruel is widely frowned upon.
Fewer people, though, would consider the matter of cruelty when it came to fish, and fewer still when it came to spiders. Yet evidence presented to the conference suggests that the former (probably) and the latter (possibly) enjoy some sort of conscious experience. As to octopuses, they are rapidly rising up the ladder of perceived consciousness, to the point where there is talk in some places of giving them legal protections against cruel treatment. That certainly creates a dilemma for those such as me who regard octopus flesh as a delicacy.
It is in the area of AI, though, that things are moving fastest. Few experts in the field suggest that systems like ChatGPT have anything resembling consciousness. But many, including Yoshua Bengio, who helped invent the deep-learning approach which led to ChatGPT, and who gave a talk at the meeting, think not only that some future software will be conscious, but also that, by following one of the theories which was tested in the adversarial collaboration, engineers have been offered a way to achieve this. That would certainly result in better and more useful AI. But Dr Bengio and others think it could also lead to something quite dangerous, if such sentient software had a self-preservation instinct programmed into it, and decided, using this, that humans were a threat to its
well-being.
Elsewhere in The Economist:
Thank you for reading this edition of Simply Science. If you have any thoughts or feedback about this edition of the newsletter, please get in touch at
[email protected].
|
Email 
