There are two reasons why I’m not worried about ChatGPT and its by-products.
First, it is far from the kind of artificial superintelligence that could pose a threat to humanity. The models that underpin it are slow to learn and require vast amounts of data to build something similar to the versatile concepts that humans can build from a handful of examples. I wouldn’t call it “smart” in that sense.
Second, many of the more catastrophic artificial general intelligence scenarios rely on assumptions that I find hard to believe. For example, there seems to be a common (implicit) assumption that sufficient intelligence equates to infinite power in the real world. If this were true, more scientists would become millionaires.
In humans, perception is part of the physical environment (including our bodies), and this environment imposes limits. The concept of AI as a “software mind” unconstrained by hardware is more closely related to 17th-century dualism (the idea that the mind and body are separable) than modern theories about the mind existing as part of the physical world. have a lot in common with.
Why are you suddenly worried?
Still, predicting fate is outdated, and the events of the last few years probably didn’t help. But there may be more to this story than meets the eye.
Many high-profile people calling for AI regulation work for or are associated with existing AI companies. The technology is convenient, but money and power are at stake, and fear-mongering is an opportunity.
Almost all the information involved in building ChatGPT has been published as publicly accessible research results. OpenAI’s competitors can (and have) replicated this process, so it won’t be long before free and open-source alternatives flood the market.
