Perhaps you are impressed with large language models (LLMs) such as ChatGPT, which can write code, build an app, and pass the bar exam? But LLMs still lack artificial general intelligence, that is, the state of a hypothetical autonomous system capable of performing intellectual tasks performed by humans or animals.
And according to Yann LeCun, Meta’s chief scientist for AI, LLMs aren’t even as smart as dogs. He asserts that LLMs are not truly intelligent because they cannot understand, interact with, or apprehend reality and rely solely on learning language to produce a result.
LeCun asserts that true intelligence goes beyond language and that most human knowledge has little to do with language. LLMs such as ChatGPT are devoid of emotions, creativity, sensitivity, and awareness, which are the cornerstones of human intelligence.
ChatGPT lacks the cognitive abilities to use common sense
ChatGPT can solve a complex mathematical problem and, without its guardrails, can explain how to create harmful substances, according to OpenAI’s GPT-4 white paper.
But, ChatGPT lacks the cognitive abilities to sense, plan, use common sense, or reason based on real-world experiences. However, GPT-4, the most recent version of OpenAI’s language model, has demonstrated human-level performance in math, coding, and law, indicating that the realization of artificial general intelligence may be coming. .
OpenAI continues to train and extend the capabilities of its GPT language models in hopes of one day achieving artificial general intelligence. The company recognizes, however, that the realization of such technology could significantly disrupt society.
Artificial beings will be smarter than humans and will have to be “controllable”
In May, OpenAI CEO Sam Altman testified before the US Senate and said his greatest fear was that his technology would cause “significant harm to the world”.
In a blog post, OpenAI says intelligent beings can generally serve many purposes, but that using technology and conducting research responsibly is paramount.
LeCun asserts that one day, artificial beings will be more intelligent than humans and that at that time they will have to be “controllable and fundamentally submissive to humans”. He claims that people’s fear that artificially intelligent beings want to take over the world is unfounded, because “there is no correlation between being intelligent and wanting to take over”.
OpenAI’s sentiments on creating an AI capable of achieving artificial general intelligence are similar to LeCun’s. The company believes that it is impossible to prevent the creation of artificial beings capable of becoming as intelligent or more intelligent than humans.
But OpenAI’s mission is to ensure the technology is developed with great caution, as it believes the risks of artificial general intelligence could be “existential” if it falls into the wrong hands and is deployed at malicious purposes.
The future of artificial intelligence, once thought to be reserved for science fiction movies, is in our near future. Will we be ready?
Source: “ZDNet.com”