It Isn’t Just Bias and Threats to Creative Livelihoods. Generative AI Is Built on Nihilism. Its Real Product Is Despair.
Every month it seems we are greeted by news of the launch of yet another artificial intelligence tool that simulates human creativity. These tools are usually lumped together under the heading of Generative AI. The latest is ChatGPT, a text-producing engine that generates remarkably human-sounding English. There are many reasons to be very concerned about Generative AI. But some of the major questions surrounding it need to be asked more publicly. When we look at the justifications developers give for pursuing these projects, we find them embracing a profound nihilism on the one hand, and offering no coherent reason for being on the other.
As they have with other Generative AI programs like DALL-E and Stable Diffusion, creators have raised alarms about the technology. Some of their worries have to do with replacing their jobs — jobs which, it should be noted, are generally among the kinds of work that people enjoy doing and, presuming they are properly compensated, derive significant satisfaction from doing. Some have to do with consumers of the simulated product confusing it with work by human beings. Some have to do with students and others substituting the AI product when human work is required, especially in higher education.
Other objections have been raised by researchers looking into racial and other forms of bias. Many prior text-based Generative AI projects have almost immediately started spewing racist and misogynistic messages, emerging from the vast amounts of web-based bigoted data on which their models have been trained. Researchers quickly uncovered that ChatGPT is using some fairly blunt filters to prevent bias — filters that can be bypassed with “simple tricks” that are “superficially masked,” as computer scientist Steven T. Piantadosi tweeted; he expanded on these concerns in an interview with technology journalist Sam Biddle. A small twitter account soon produced a “racism rap” full of biased statements.
While some bias concerns might be addressed by improving the software, creativity concerns are only likely to be exacerbated as software like ChatGPT gets better. Creators and educators might say that ChatGPT should not exist at all, even if it could be freed from…