As an international professional all-domain digital marketing operation platform, Whale pays close attention to key technological innovations such as artificial intelligence (AI), large-scale Internet of Things (IoT) and data models (Data).
To this end, Whale attaches great importance to helping employees obtain more world-class information, perspectives and diverse learning opportunities in cutting-edge technology fields. We believe that a better Whaler is a better Whale, which can create greater value for customers.
On October 27, 2022, a sharing conference themed on "AI's Next Generation Narrative-Mind Al" will be held as scheduled during Whale Technology Week .
This time Whale Technology Week - Master Sharing Day, we are very pleased to invite the founder and CEO of Mindverse.ai: Dr. Tao Fangbo to share Mind AI technology with us.
01
About Dr. Tao Fangbo
Dr. Tao Fangbo:
Founder and CEO of Mindverse.ai.
Undergraduate of Tsinghua University, Ph.D. of UIUC in the United States, successively engaged in research work in Microsoft Research, NASA, and the US Army Laboratory.
Former Facebook senior research scientist, led the team from 0 to 1 to create an AI base for information flow content, serving 200 million+ users.
A former scientist at Alibaba Dharma Institute, he once led the most cutting-edge interdisciplinary AI research team within Ali.
Next, let us follow Dr. Tao Fangbo and explore the mysterious world of Mind AI together!
02
AI Chronicles
If you want to learn something new, you must first understand its history. Today, Dr. Tao Fangbo starts with the development history of AI.
In 1956, the Dartmouth Conference proposed the concept of "AI". Since then, AI has entered a new historical stage. In the following 70 years, the AI subfield "Skill AI" flourished.
From 2012 to 2022, the development of AI has changed from a single-loop Skill AI to a multi-loop Multi-head Skill AI with scalable training, and to a multi-loop Prompt-based Skill AI that requires no training for the entire cortex!
Dr. Tao mentioned: " In 2022, Skill AI will begin to make a preliminary transformation to Mind AI . "
In terms of technical architecture, Mind AI has added the ability to seek knowledge from the outside (external information brain area), the ability to store and read human memory (memory brain area), the ability to set roles (personality brain area + external tools), Set role + professional knowledge ability (personality brain area + professional knowledge brain area), digital executive ability + (digital) motor brain area.
The development of AI moves from the micro, to the meso, and finally to the macro. Mind AI Transforming the AI World? Mind AI transforms the world! And, this revolution has only just begun...
03
What exactly is Mind AI?
After understanding the development history of AI, Dr. Tao started the "highlight" of this lecture, what is Mind AI?
The essence of the Mind AI technological revolution is to create macro-level "conscious behaviors" in stages with the help of micro- and meso-mature technologies, so that it can finally be perfected and move towards Human-level AI.
Principles of mental intelligence:
Dr. Tao said that the bottleneck of deep learning is already obvious . Although it is very suitable for solving orientation problems with the addition of big data training, it is seriously lacking in generalization, explainability, personality attributes, and rapid learning ability. The entire AI industry desperately needs a new paradigm.
So in this context, a new paradigm of AGI was born: UMM whole brain mental framework. Its underlying technology path is UMM, with the goal of building a virtual mind, based on the whole brain model, and centered on the theory of digital consciousness, to construct a new paradigm with six major innovations.
Traditional AI Paradigms:
Does not have a real thinking mechanism; cannot carry out personality design; cannot solve strong logic dialogues and tasks; cannot introduce industry knowledge; does not have common sense/dialogue ability and logical cognition ability; intensive learning paradigm, difficult training, scene adaptation; cost High; unable to abstract unstructured information, unable to perceive the physical world; poor flexibility, high maintenance and construction costs; no personality and thinking...
New paradigm:
Use a super-large-scale pre-training model to make AI have common sense and dialogue capabilities; use multi-agent reinforcement learning technology to imitate the behavior of agents; construct industry knowledge maps, and use biased rules to solve orientation problems.
Dr. Tao introduced us to the three stages of Mind AI: information-based Mind, thinking-based Mind, and self-driven Mind.
Phase 1 of Mind AI: Informative
Carrier technology: Light-UMM
Key design: In-depth integration of memory information, character design and motivation.
Value embodiment: Let the information-based/human-based needs experience ++ (office/business/education), transforming the relationship between people and information.
At this stage, the information-based Mind is the "medium".
The second stage of Mind AI: thinking type
Bearer technology: UMM
Key design: thought flow + inner world system.
Value manifestation: real "sense of connection", companionship/separation/assistant needs, transforming the relationship between human and digital life.
At this stage, it is "companion relationship".
The third stage of Mind AI: Self-driving
Carrier Technology: Life-UMM
Key design: the value delivery architecture centered on homeostasis/FET, the "engine of life".
Value embodiment: Let Mind really become an independent life (scientist/artist/SBT), and digital life can let itself go.
At this stage, a "symbiotic relationship" is achieved.
04
The future of AI
Dr. Tao explained to us his predictions for the next 5-10 years of the AI industry :
First, the AI industry will be quickly reshuffled, and the previous AI scale will only be less than 1/10 of the future.
Second, just as you don’t need to look at the AI technology before 2017 (pre-Transformer), you don’t need to look at the framework and AI model (pre-Mind) before 2022 in the future. AI changes have always been non-linear.
Third, Mind AI is not only a technological trend, but also makes it possible for AI to form a platform for the first time, so that the AI industry can truly solve the problem of ROI and project system, and greatly accelerate the penetration of AI into the world.
"In 10 years, Mind AI will lead AI to an output value of more than 10 trillion US dollars, and account for more than 80% of the application scenarios." Dr. Tao said. So, what is the application scenario of Mind AI?
Mind OS : A mind operating system that can set personality. Create a unique virtual human by setting the MBTI of the character, opening the visual ability, endowing life memory, learning the common sense of the world, setting the language style, and deeply customizing skills.
Mind Commerce : Subvert traditional marketing, allowing products to speak for themselves, communicating with users or AI. Everything except merchandise can talk and have randomly different and interesting personalities.
epilogue
I am very grateful to Dr. Tao Fangbo for bringing us such a wonderful lecture, and all the students of Whale have benefited a lot.
As Dr. Tao envisioned, in the future all human beings will be able to explore the true definition of life in the digital world, and civilization will enter a new stage, where everyone can create and trust each other, and AI will be the true civilized partner of mankind.
We will continue to invite industry and technology experts to bring more high-level dialogues to Whaler, so stay tuned!