2024-09-20 02:04:00
AGI – Artificial intelligence, like humans, is also capable of self-correction and reaching new conclusions by “learning by thinking”. This is demonstrated by research by Tania Lombrozo, professor of psychology and director of the Natural and Artificial Minds initiative at Princeton University, reported in the journal Trends in Cognitive Sciences. “There have been some recent demonstrations of what looks like cognitive learning in AI, especially in large language models,” said Tania Lombrozo.
“Sometimes ChatGPT corrects itself without saying it publicly; this is the kind of thing that happens when people are busy learning about thinking,” Lombrozo continued. The study identified four examples of learning through reasoning in humans and AI: Learners can acquire new information without external input through explanations, simulations, analogies and reasoning. In person, explaining to a child how a microwave works can reveal gaps in understanding.
Arranging furniture in your living room often involves creating a mental image to practice different arrangements before making any physical changes. Downloading pirated software may at first seem morally acceptable, until you compare it to stealing physical goods. If you know that your friend’s birthday is on a leap day and tomorrow is a leap day, you can assume that your friend’s birthday is tomorrow.
Artificial intelligence exhibits similar learning processes. When asked for more information on a complex topic, the AI can modify or adjust its initial response based on the information provided. The gaming industry uses simulation engines to approximate real-world results, and models can use simulation results as input for learning.
Asking a language model to draw comparisons can lead him to answer questions more accurately than he would with simple questions. If you ask the AI to think step by step, you can get answers that you wouldn’t be able to get with a direct question. “This raises the question of why both natural and artificial spirits have these characteristics,” Lombrozo pointed out.
“I argue that philosophy is a kind of ‘learning on demand,'” Lombrozo added. When you learn something new, you never know how useful the information will be in the future. Lombrozo argued that people can leave knowledge for later, until the context makes it important and it is worth using the cognitive effort to think and learn. Lombrozo acknowledges the difficulties in defining the boundaries between thinking, learning, and other higher-level cognitive functions, an area of debate in cognitive science. The review also raises other questions, some of which Lombrozo plans to explore, such as whether AI systems are actually “thinking” or simulating the results of such processes.
“AI has reached a point where it is advanced in some ways, but limited in others, that we are able to study the similarities and differences between human and artificial intelligence,” said Lombrozo. “We can learn important things about human cognition through AI and improve AI by comparing it to human beings,” Lombrozo continued. “It is an important moment where we find ourselves in this new position to ask interesting comparative questions,” concluded Lombrozo. (AGI) Sci/Advertising
#Artificial #intelligence #learn #human