Articles
GPT: ARTIFICIAL INTELLIGENCE OR PARROT INTELLIGENCE?
I was somewhat forced to write this editorial since Artificial Intelligence (AI) is the topic of the day in dentistry. I started by reading some definitions on Wikipedia, that in my opinion is one of the few very useful tools of the web. So I use Wikipedia definition for AI that stands like “intelligence exhibited by machines, particularly computer systems. It is a field of research in computer science that develops and studies methods and software that enable machines to perceive their environment and uses learning and intelligence to take actions that maximize their chances of achieving defined goal”. This technology is widely used in many disciplines but in the context of this editorial we are interested in its applications in dentistry and in particular in assisting dentists to choose the optimal therapy for a given patient.
Then I superficially consulted PubMed which is, by the way, another form of artificial intelligence, to have a generic update. I simply used as keywords: artificial, intelligence and dentistry. PubMed gave me 3004 hits, so I started to scroll the titles. After the first 100 titles, mostly elementary and repetitive narrative reviews in sub-scientific journals, I become so bored that I had to stop. Too many useless articles disguised as “scientific papers” diluting the real knowledge, just with the aim to position some more “academics” in universities around the globe.
So what was the next step? One of the most popular tool at the moment is ChatGPT, also because there is a free access version. GPT stands for “generative pre-trained transformers” that are large language models based on the semantic relationships between words in sentences (natural language processing). Text-based GPT models are pre-trained on a large corpus of text, usually from the internet. The pre-training consists in predicting the next token (a token being usually a word, subword, or punctuation). Throughout this pre-training, GPT models accumulate knowledge about the world, and can generate human-like text by repeatedly predicting the next token. Obviously, current GPT models are prone to
generate “hallucinations”, i.e. falsehood. So with the help of a promising and fresh-minded dental student, we interrogated ChatGTP posing four clinical questions on which I have been conducting clinical research for decades. The answers were well structured logically, well written and easy to understand, were given very fast, but were totally acritical and full of common banalities. It was clear that were summarized from the web which is rich of false beliefs.
In conclusion, my concept of medical intelligence is something else. Intelligence for me is the capacity of critically elaborate information, while the so called GTP artificial intelligence is able, in an incredible short time, I must admit, to make a summary of what is reported in the web, being totally unable to critically elaborate the content. Maybe we should rename it as the “parrot intelligence”. I am not sure that used in this way, it will bring a big benefit to humanity in the medical field. I see it more like the Big Brother of knowledge which will induce atrophy in the brain of the new generations who are already very skilled in using ChatGPT to create essays, reviews and homework, but cannot critically elaborate medical information. Will parrot intelligence be a new revolution? Yes, definitively it will be, actually it is already. However, as too many revolutions in history, it may not necessary bring a better world.
Serene reading
Marco