But most of these predictions are coming from people working in commercial interest companies in AI. It was noteworthy that none of the researchers we did to this article are ready to offer AGI’s definition. However, they were ready to explain how the current system is low.
“I think AGI would be something that would be stronger, more stable – usually not more smart, but more integrated into its abilities.” Ariel GoldstineA researcher at the Hebrew University of Jerusalem. “You will expect a system that can be able to do X and Y too Z and T. Somehow, these systems look more scattered. Surprisingly, one thing is surprisingly good and then surprisingly bad on something else that seems to be associated with it.”
“I think it’s a great distinction, the idea of being normal,” the neuro scientist recovered Crista Baker NC State University. “You can learn how to analyze the logic in the same circle, but if you come in a new situation, it is no longer a stupid.”
Mariano ShenA Google Engineer who has cooperated with Goldstine, focused on the capabilities that reduce this general capacity. He mentioned the ability to deploy both the long -term and task -related memory and the skills developed in a job in various contexts. These are unclear in the current AI system.
Beyond these specific boundaries, the baker noted that “it has been a very human concentrated idea of intelligence that only humans are intelligent.” It has fallen within the scientific community because we have studied more about animal behavior. But there is still prejudice to give privilege of human behavior, such as the reaction of human voice created by large language models
There is a fruit bee that baker studies can connect a variety of sensory information, control four sets of organs, navigate the complex environment, meet its energy needs, produce new generations of the brain and much more. And they all do it with a brain that has less than 150,000 neurons, which are far less than the current big language model.