Sam Altman, the CEO of OpenAI, is reportedly seeking $7 trillion in investment to produce computer chips for artificial intelligence (AI) systems. Altman also believes that the future AI-saturated world will require a significant increase in energy, potentially necessitating a breakthrough like nuclear fusion. However, as an AI researcher, I have doubts about the future of AI.
Current AI systems, including ChatGPT, are not truly intelligent. Without fundamental changes to their functioning, there is no evidence that they can achieve true intelligence. The definition of AI as a computer system that can perform tasks associated with intelligent beings is vague. While spreadsheets and robots can perform tasks once done by humans, they do so by following algorithms.
It is important to note that there is no single AI system that can perform a range of intelligent actions like a human. Instead, there are various AI technologies that can accomplish different tasks. Discriminative AI helps with decision-making, while generative AI systems like ChatGPT generate outputs based on inputs. However, these generative AI systems have limitations and can produce false responses.
The goalposts of AI have constantly shifted over time. In the 1980s, I worked on an expert medical advice system that was considered an AI landmark. However, it was essentially a look-up table matching lab results to diagnostic advice. While it was considered AI at the time, it lacked true intelligence.
Generative AI systems still rely on sets of rules and cannot easily be explained in human terms. They lack insight and judgment, which are key components of real intelligence. While generative AI can produce impressive results, it is not truly intelligent.
Generative AI has its uses and will continue to improve, but it is not magic. It is a clever algorithm created through human ingenuity. As we become more familiar with generative AI systems, we will realize their limitations and understand that they are not truly intelligent.
