Thanks to the fact that languages and compilers make it possible to more easily represent the complex reality of companies, order their information and process their transactions, we have advanced products and services in real time in many sectors of the economy. Banks, telecommunications, travel or video games would not be possible if their processes had to be written with zeros and ones.
The compilers or interpreters are based on the “Turing machine”, the automata theory developed in the 40’s by the English mathematician. The first computer language was “Fortran” in the 70’s, which allowed to structure a set of commands in a friendlier language than zeros and ones. Then came others such as Cobol, Pascal, Prolog or Lisp, the CASE (Computer Aided Software Engineering) tools and new object-oriented languages (OOP) such as C++ and Python, with increasing capacities to receive complex instructions in a more symbolic way.
In my opinion, ChatGPT, DALL.E and some other not so well known, belong to the next generation of compilers, which are now capable of interpreting natural language and converting it to zeros and ones that look up data or calculate things on a large base of stored information. But that’s not intelligence, it’s automation.
Intelligence is something else. Intelligence is the ability to project and imagine. It’s understanding a friend’s wink, laughing at a bad joke, knowing that someone is lying to you just by looking into their eyes. Intelligence is when your grandfather says that it will rain tomorrow, even though you don’t see a single cloud in the sky, and that it does indeed happen.
People are amazed with these applications and call it “Artificial Intelligence (AI)” because in addition to being entertained by the nonsense that appears on social networks, they were already used to the speed and results of Internet search engines. Few expected the technology of languages and compilers to take this leap forward, understanding what we type and responding with an aura of intelligence to general knowledge questions, summarizing articles, combining advanced photography and graphics, translating sentences into other languages, or encoding instructions in others programming languages.
That is what ChatGPT does, interpreting natural language and translating it into search orders and processing millions of texts and content stored until the year 2021 to give an answer. But it does not have the capacity to solve new problems, those that are not previously documented – perhaps for a mathematical problem it will answer that “a polynomial approximation would be necessary” (a calculation procedure developed by Newton and Lagrange). But there is no intuition behind. There is no new contribution.