Unlike narrow or specialized AI systems designed for specific tasks, Artificial General Intelligence (AGI)  can perform a big selection of functions that aim to copy human intelligence’s broad cognitive abilities and flexibility. AGI can function autonomously by making decisions and taking actions independently. AGI may also comprehend ambiguous or incomplete information. 

Achieving AGI is a fancy and difficult endeavor, because it requires solving quite a few difficult problems in machine learning, natural language processing, robotics, and other AI-related fields.

Researchers on the National Key Laboratory of General Artificial Intelligence propose a brand new way of evaluating AGI by introducing the Tong Test. “Tong” corresponds to the Chinese character of general in AGI. 

They propose that AGI evaluation needs to be rooted in scenarios with the complex environments of DEPSI. They are saying that only through evaluations inside DEPSI can the human-like abilities of AGI, similar to commonsense reasoning, intention inference of social interactions, trust, and self-awareness, be promptly assessed. The Tong test offers a brand new perspective on AGI evaluation by emphasizing the importance of DEPSI as ability, value-oriented somewhat than a task-oriented evaluation. 

The Tong test is a benchmark and evaluation system specializing in essential features similar to infinite tasks, self-driven task generation, value alignment, and causal understanding. Their proposed virtual platform could also support embodied AI in training and testing. Embodied AI agents acquire information inside this platform and proceed to learn and finetune their values and talents interactively. 

To support infinite tasks, they follow a compositional graphical model as a basic form of data representation that parses any given scene’s spatial, temporal, and causal relations. They define a fluent space for the time-varying variables; these represent all possible scene configurations that could be represented inside a continuous DEPSI environment space. 

The Tong test spans two domains called the U–V dual system. The U-system describes the agent’s understanding of extrinsic physical or social rules. In contrast, the V-system comprises the agent’s intrinsic values, defined as a set of value functions upon which the self-driven behaviors of the agent are built. The Tong test platform has modules for intermediate data visualization and a panel that displays the model’s performance, indicating how well the tested model performed.

Thus, the proposed  Tong test based on DEPSI  defines the five multidimensional levels of values and talents and provides a practical pathway for constructing theoretical guidance for developing AI algorithms. 

Take a look at the Paper. All Credit For This Research Goes To the Researchers on This Project. Also, don’t forget to affix our 30k+ ML SubReddit, 40k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the newest AI research news, cool AI projects, and more.

Should you like our work, you’ll love our newsletter..