🖥 Machine Psychology—an Alternative Way to AGI
🖥 Machine Psychology—an Alternative Way to AGI
Sam Altman, CEO of OpenAI, thinks that this year we might see strong AI (AGI)—a system equal in capabilities to the human mind. Robert Johansson from Linköping University in Sweden, believes that AGI can be achieved within five years, but such level of AI should not come from machine learning like ChatGPT and other popular generative AI tools.
In his opinion, the Non-axiomatic Reasoning System (NARS) developing "machine psychology" could be an alternative. This is exactly the focus of Johansson's recently published study.
How does NARS work?
➡️ The emphasis is not on pre-learning but on operating with limited data and learning about the world in real-time. The main feature of NARS systems is an ability to reflect and constantly revise knowledge adapting to new experiences just like humans do.
➡️ NARS thinks flexibly, uses deduction, analogy and other logical techniques, rather than relying on pre-trained algorithms.
➡️ The system's memory has a solid structure that changes over time. A standard generative AI model, on the other hand, is much more of a "black box."
🖥 NARS-systems are more "human-like"
This approach makes the system more flexible, capable of functioning in complex and unpredictable situations and learning from environmental interactions.
NARS-based AI systems will help us reach AGI—an intellectual equal to humans, Robert Johansson insists. And the emergence of such AI agents might also test society's moral stances.
"We live in a society with laws and rules and moral premises. It's necessary to take a stance on how to view the rights and obligations of such agents. Maybe AGI will just be a program you run in your browser, but I think that if it has a consciousness it's still some kind of life," Johansson says.
More on the topic:
👁 Yann LeCun on Creating AI Models with Common Sense
👁 Can AI Understand the Real World? The "Chinese Room" Experiment
