Why WuDao could be more useful even than GPT-3 for the development of a future Artificial General Intelligence?

What makes Wu Dao very relevant as an?AI?project is not that it is 10x more powerful than GPT-3 in terms of parameters (1.75 trillion vs 175 billion) but Wu Dao is much more “conversationalist”, something that is absolutely key to emulate an Artificial General Intelligence (#AGI).

Probably language is the best vehicle to model the intelligence or at least to emulate it inside a machine. And it seems we will see it soon, probably in this decade (this goes beyond the Turing test). This is not about to be a strong artificial intelligence but about to seem a strong artificial intelligence in front of humans.

Additionally, WuDao is multimodal and this is really a great advance in front of the OpenAI project: it can learn from text and images and tackle tasks that include both types of data (GPT3, doesn’t). This is really powerful in terms of learning from the external world in terms of data.

We should remember that both?GPT-3?and?Wu Dao?were born with the ambitious mission of moving current technology towards an Artificial General Intelligence scenario, using language and its understanding to achieve it. Now, it’s not only text, it is also images and their understanding. Language is the vehicle to manage the world and the interactions among machines, humans, and the environment. In the case of the Chinese project, the training data is divided in: 1.2TB Chinese text data in Wu Dao Corpora, 2.5TB Chinese graphic data, and finally1.2TB English text data in the Pile dataset.

If we get an?AI?that understands, learns, communicates, and establishes relationships between ideas and meanings, it would be “almost” a true?AGI?in many senses (although maybe not “sensu stricto”). This is why I really think Wu Dao is so important in terms of AI evolution and especially in order to create a new kind of artificial intelligence that could help us to develop the concept of extended human that needs an intelligence machine to really create human and intelligent interactions.

This is only the beginning and Wu Dao could be really the tipping point that we thought OpenAI was.