Google DeepMind: AI Agents Can Copy Humans For Getting Closer To Artificial General Intelligence

Google DeepMind: AI Agents Can Copy Humans For Getting Closer To Artificial General Intelligence
SHARE

It’s been revealed that Google’s AI offshoot found copy-cat robots that are able to copy living mentors. Check out the latest reports about the matter below.

AI agents can copy humans

A group of researchers specializing in machine learning from Google’s DeepMind have claimed to have demonstrated that artificial intelligence can acquire skills in a manner similar to social learning in humans and other animals. Social learning is a process where one individual acquires new skills and knowledge by copying another.

This process is critical to the development of humans and many other animals. The DeepMind team claims to be the first to demonstrate this process in artificial intelligence.

The team, led by Edward Hughes, a staff research engineer at Google DeepMind, aimed to address some of the limitations in AI agents acquiring new skills.

Until now, teaching AI agents new capabilities from human data has relied on supervised learning from large numbers of first-person human demonstrations, which can be time-consuming and expensive.

Researchers have found a way for AI agents to learn from both human and AI experts with human-like efficiency.

They conducted experiments in a physical simulated task space called GoalCycle3D, which is similar to a computer-animated playground with obstacles and footpaths.

The AI agents were able to learn how to navigate through various problems, even though they had no prior knowledge of humans.

The researchers used reinforcement learning to train the AI agent to identify new experts, mimic their behavior, and retain the knowledge learned in a few minutes. The findings of this study were published in the peer-reviewed open-access journal Nature Communications.

“Our agents succeed at real-time imitation of a human in novel contexts without using any pre-collected human data. We identify a surprisingly simple set of ingredients sufficient for generating cultural transmission and develop an evaluation methodology for rigorously assessing it. This paves the way for cultural evolution to play an algorithmic role in the development of artificial general intelligence,” the study said.

The researchers are eager to see how others in the field of AI will apply these findings more broadly to demonstrate the ways in which cultural evolution can be developed in AI. This involves the development of skills across generations in a community.


SHARE
Rada Mateescu

Passionate about freedom, truth, humanity, and subjects from the science and health-related areas, Rada has been blogging for about ten years, and at Health Thoroughfare, she's covering the latest news on these niches.

Post Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.