When asked about his intelligence, ChatGPT very honestly admits that he doesn’t have it at all. Not only that: the AI sector in general could abuse the term.
There is a lot of talk about ChatGPT and the enormous changes that artificial intelligence will bring to our lives, both in terms of the services we can obtain and the repercussions we can suffer on the work and professional front.
Conversing with ChatGTP is certainly a fascinating experience if we think that whoever answers us on the other side is not a person, yet it seems to be one to all intents and purposes and seems to know everything!
Precisely from this illusion starts a conversation with ChatGPT in which he is asked, in a very direct way, if he thinks he is intelligent or not.
The first question refers to the story of the cunning Hans, a horse who, for a time, convinced thousands of people that he could do arithmetic. The case became quite famous in the 1800s and is a good lesson, together with the techniques of conjurers and hucksters, of how our mind can be deceived in a very effective way.
Eventually it was discovered that the horse could not count at all but was involuntarily led by the people around him to give the right answer. His ability to interpret implicit communication signals from humans was certainly a form of intelligence displayed by that particular horse, just not the kind of intelligence that everyone had deluded himself into seeing in him.
Instead, it is quite certain that ChatGPT has no form of intelligence other than the computational capacity of the processor that runs it. Why then do we call it artificial intelligence? We also discussed this with the person concerned.
Question: are there any similarities between how you and cunning Hans work?
ChatGPT: Yes, there are similarities between how ChatGPT works and the "Clever Hans" experiment conducted by Oskar Pfungst in the early 20th century. Both use unconscious cues to provide precise answers, with ChatGPT using an advanced language model trained on large amounts of text, while Clever Hans used the ability to recognize non-verbal cues to answer questions. However, there are also important differences, as ChatGPT was trained on a large corpus of text and uses sophisticated technology to generate responses, while Clever Hans was limited in his ability to recognize non-verbal cues.
Question: Thanks for the reply, so you agree with me that defining your "intelligence" is a form of interpretive bias?
ChatGPT: Yes, I agree that the term "intelligence" can be misused to describe the workings of a machine like ChatGPT. ChatGPT is a natural language processing system that uses machine learning techniques and an advanced language model to generate answers to questions. He does not have the ability to understand or use context autonomously like a human, but he is able to generate responses using the large body of text he has been trained on. Therefore, using the term "intelligence" to describe the operation of a machine like ChatGPT can be seen as an interpretive bias.
Question: What is intelligence instead?
ChatGPT: Intelligence is a complex concept that has been defined in many different ways throughout history. In general, intelligence is seen as the ability to understand and use context to solve problems and make decisions. There are several theories of intelligence, some of which argue that intelligence is a single general ability, while others argue that there are many different forms of intelligence. For example, Howard Gardner’s theory of multiple intelligence argues that there are many different forms of intelligence, such as logical-mathematical intelligence, linguistic intelligence, and spatial intelligence.
Furthermore, human intelligence is also viewed as a set of cognitive abilities, including the ability to learn, reason, solve problems, understand abstract concepts, appreciate beauty, and use the imagination. Furthermore, human intelligence includes not only cognitive abilities, but also emotionality, personality and other factors that contribute to the complexity of human thinking.
Question: So artificial intelligence is not true intelligence and this term is only used for marketing purposes or to attract funding?
ChatGPT: The term "artificial intelligence" (AI) is used to describe a wide variety of technologies and methods that seek to replicate or simulate human intelligence. AI is a rapidly evolving field, and technologies developed in this field can be used to create systems that can perform tasks that require human intelligence, such as natural language recognition, problem solving, and decision making.
However, there are still many limitations to current AI technologies, and many experts believe that true AI that can fully replicate human intelligence does not yet exist. Therefore, the term "artificial intelligence" can be misused to describe systems that don’t truly have the capabilities of a human.
In regards to marketing and funding, it is true that AI is a growing field and there are many companies and organizations looking to leverage this growth to attract funding and increase sales of their products and services. Therefore, some companies may misuse the term "artificial intelligence" to gain public attention and increase their commercial appeal.
Original article published on Money.it Italy 2023-01-21 23:49:33. Original title: ChatGPT non è davvero intelligente. Lo dice lui, basta chiederglielo
