There is no consensus on the definition of what AI, and we have been developing it since the 1970s.
Even less consensus there is on the Artificial General Intelligence (AGI). Bless this debate.
AGI is a theoretical form of AI where a machine would have an intelligence equal to humans (the singularity); Yes, there is a certain consensus that the AGI should know about almost everything, be able to learn autonomously (without being trained by humans) and self-correct.
And since the debate is open, here is my proposal:
This graph, like all simplifications of reality, is good just to understand a concept: evolution.
On one axis, an Specialist AI is the one that knows a lot about just one topic. The Generalist one, just as ChatGPT, knows a lot about almost any topic. ChatGPT hit 77% in Geography but 50% in logic. On the othe axis, the single intelligence is that it does only one thing well. For example, Midjourney is amazing at creating images from text. But he only does that. He does it very well, and touches all the sticks within the image (realistic, manga, abstract,…) but within this scope. Midjourney touchen any topic, but has only one kind og intelligence. It does not make video, 3D models or avatars. Chat GPT knows about almost any topic, and has various intelligences, but we could say is especialized in the linguistic intelligence. We are creating greate AI, but with a single kind of intelligence.
For multiple intelligences we can take Gardner’s theory of multiple intelligences, for example, and add some others such as common sense, sense of humor,…
And that’s the beauty, little by little we are going to combine these specialized and simple AIs that are going to bring us closer, little by little, and in an evolutionary way to the AGI
Last week Open AI launched a very interesting manifesto laying the foundations of the AGI. There are many interesting things on it. I only disagree on one, and it is important.
This revolution will not be led by large corporations (although they are now leading the change). The IAG is considered by many governments a possible weapon with greater destructive power than a nuclear weapon.
A world where an AGI is capable of transforming our entire way of work and socialize can only be governed from a Web3, from a decentralized structure where our digital identity is protected and guaranteed, where data cannot be corrupted and manipulated.
Blockchain was born to solve the problem of money. But perhaps it will go down in history for solving the greatest challenge in the history of humanity: how do we humans achieve an AGI without destroying ourselves.
Yours in crypto and AI.
And next week GPT4 may be realised.