Artificial General Intelligence (AGI) is a type of AI that can perform a wide range of tasks and replicate human intelligence. It has the ability to make decisions independently and understand ambiguous information. However, achieving AGI is a complex task that requires solving problems in machine learning, natural language processing, robotics, and other AI-related fields.
Researchers at the National Key Laboratory of General Artificial Intelligence have introduced a new way of evaluating AGI called the Tong Test. This test focuses on evaluating AGI in complex environments and assesses its human-like abilities such as commonsense reasoning, intention inference, trust, and self-awareness. The Tong test emphasizes the importance of DEPSI (the complex environment) as an ability and value-oriented evaluation rather than a task-oriented evaluation.
The Tong test is a benchmark and evaluation system that focuses on essential features like infinite tasks, self-driven task generation, value alignment, and causal understanding. It also supports embodied AI in training and testing. Embodied AI agents can acquire information within the Tong test platform and continuously learn and improve their values and abilities.
To support infinite tasks, the researchers use a compositional graphical model as a basic form of knowledge representation. This model parses spatial, temporal, and causal relations in a given scene. They also define a fluent space for time-varying variables, representing all possible scene configurations in a continuous DEPSI environment space.
The Tong test spans two domains called the U–V dual system. The U-system focuses on the agent’s understanding of extrinsic physical or social rules, while the V-system comprises the agent’s intrinsic values. The Tong test platform includes modules for data visualization and performance evaluation.
Overall, the Tong test based on DEPSI defines five multidimensional levels of values and abilities and provides a practical pathway for developing AI algorithms.
Check out the research paper for more information. And don’t forget to join our ML subreddit, Facebook community, Discord channel, and subscribe to our email newsletter for the latest AI research news and projects.
If you like our work, you’ll love our newsletter. Sign up now!
[End of article]
SEO Keywords: Artificial General Intelligence, AGI evaluation, Tong test, DEPSI, embodied AI, infinite tasks, compositional graphical model