Giul-IA

thoughts are text

What it does

giulIA was born from a reflection made almost forty years ago. Will machines ever be able to think and make autonomous decisions? But... what is thought? If I think or dream, I do it in Italian. An Englishman, however, thinks in English. What if language was the mother of thought? At that time I thought that the key to everything was to be able to make a machine generate contextualized text. If the machine generates text, could it think? I wondered... Maybe if the machine could convert images into text descriptions. Could it also think based on what it sees? Dragon Dictate could be used to convert words into text and give it as input. Could it therefore feel? But the time was not ripe. For almost forty years I have carried these questions with me, cataloging them as pure science fiction. But here come the LLM models. They are exactly what I needed to test my ideas.GiulIA is a prototype of artificial intelligence.It generates text but processes it as if it were its own thought stimulated by a particular input, a text, the value of a sensor or even just the passage of time.This thought leads giulIA to decisions (also influenced by its current emotional state) on how to respond or what to do based on its own action list. If something is considered interesting, Giul-IA will not hesitate to store it in its long-term memory.

Built with

  • None

Team

By

Francesco Bozzotta

From

Italy