Parts of the Intellect
Over the past year, as OpenAI’s ChatGPT has gone from a specialist tool to a worldwide cultural phenomenon, there has been one anxious question controlling the discussion: is this time different–are computers now really intelligent–and what does this change about the human self-understanding? If human beings are exceptional, then it is in large part because of intelligence.
It didn’t help that a computer was now considerably more likely to pass one of the most clearly defined, functional tests for artificial intelligence, the so-called “Turing test:” give a human being the chance to pass messages back and forth with a partner behind a veil; if the human cannot tell that he or she is conversing with a machine, it passes the test. It is intelligent, practically speaking.
There are a lot of problems with this test. Still, the bar was raised. Furthermore, if the standard for “real” artificial intelligence is a moving target, always a few steps ahead of whatever computers are currently capable doing, then maybe the questions about artificial intelligence are hopelessly philosophical, likely to generate new pathways for analysis but impossible to answer with any closure.
When I wrote above that human intelligence is an essential quality of the human–of human exceptionalism–I meant it in two senses of that word: that intelligence is (1) a distinguishing quality of the human, and that (2) as a quality, it has the special status of an essence. The essential quality and its object are hard to separate. What is intelligence? Look to human beings, see it in action. What are human beings? Homo sapiens, thinking things, subjects with intelligence.
What we may be seeing right now is a shift of intuitions, a breakdown of confidence that intelligence is an essentially human quality. This does not mean that artificial intelligence is like human intelligence, or that computers are (will someday be) more intelligent than humans. But it does suggest that intelligence is increasingly detached from how it was previously defined: through human beings.1
A new situation emerges; imagine pieces, bits, scraps of (general) intelligence circulating throughout the environment. More people may have to make constant judgments about the scope of the intelligence of various things.
I cannot see into these scraps of intelligence, know what they are. Their capabilities and intentions (if they exist) are opaque to me–like those of other human beings. Maybe there is a new standard for artificial intelligence: is it necessarily unknowable? Then it is intelligent.
I am reminded of the debate about viruses and life: are they alive? If so, how? Are viruses alive in the same way that living things (people?) are? Here is another philosophical question that is difficult to operationalize. Viruses interface with life, need life, latch onto life and push it in new directions. One cannot help but ask the question about viruses because they are so strange; maybe what is being sought is another, comparable essence that applies to the virus. The virus, by suggesting a comparison with life, makes life seem less like an essential thing and more a definable set of processes which can be recombined in ways so strange that categorization falls apart.
I wonder if something similar is happening with intelligence right now. What was once a unitary essence that attached to the human is now being decomposed into X number of parts, parts that we can see, uncover, build into new entities that display some of the qualities of intelligence, without the human.
- There are other reasons for this shift in understanding about human privilege related to intelligence, like a better understanding of the capabilities of other animals. ↩