Artificial Intelligence & Theory of Mind
Students:
Class: Α3 - Grade: A3 - School: 1st Arsakeio Senior High School of Psychico
Course: IT Applications
School year 2022-2023
OBJECTIVE
This website is created to inform the readers about the connection between AI and Theory of Mind.
human-like-robot-and-artificial-intelligence-2022-01-06-00-25-53-utc.jpg
WHAT IS ARTIFICIAL INTELLIGENCE
Artificial intelligence is a field of science that occupies with building computers and machines that can reason, learn, and act in a way that would normally require human intelligence or that involves data whose scale exceeds what humans can analyze.
THEORY OF MIND
ToM is the ability to infer and understand another's emotions (the beliefs, thoughts, intentions and feelings of another), and use this information to explain and predict its next move.
Why is it called a "theory' though?
Experts and psychologies refer to it as such because our beliefs about what might be going on in another person's head are just that—theories. While we can make predictions, we have no direct way of knowing exactly what a person might think.
In psychology, theory of mind is an important social-cognitive skill that gives the ability to think about mental states, both your own and those of others. It encompasses the ability to understand in a deeper level mental states, including emotions , desires, beliefs, and knowledge, and recognize that other people's thoughts and beliefs may differ from yours.
ARTIFICIAL INTELLIGENCE & THEORY OF MIND
589937_gallery_5d6fdd3e97acc_jpg_fa_rszd
Although the Theory of Mind AI is still under research and development, we already know what will be different from the other branches of AI we have seen.
It is well known that a computer equipped with Theory of Mind AI will better understand the entities it interacts with and that is because it is created for that particular job.
It will use meta-learning to build models about the agents it comes across. This neural network will observe the behaviors of these entities, therefore, it will acquire a prior model of these behavior patterns.
0_Xuwt8ha7aSJvRG-K.jpg
What triggered the research and development into the Theory of Mind AI?
AI and ML have given us plenty, e.g., chatbots, image recognition capabilities, natural language processing capabilities.
Take for example the case with image recognition capability. Law enforcement and security agencies can keep a country safe with the help of facial recognition software. Such recognition software uses the image recognition capability of AI.
While these are impressive, society needs more to make a quantum jump in its productivity. Any organization has workers, and workers bring their beliefs, emotions, dreams, and apprehensions to the workplace. In other words, they bring their mind to the workplace!
The mind of a worker significantly influences his or her effectiveness at work, and an organization needs to harness its employee’s emotions positively to deliver superior performance. AI needs to create such “co-workers” for organizations, and that’s much more than the chatbots!
AI can achieve this only when the artificially-intelligent “co-workers” can understand the emotions, beliefs, aspirations, etc. of their human co-workers. This is possible only when the AI-powered “co-workers” have their theory of mind.
This is the key requirement that Theory of Mind AI seeks to fulfill.
What characteristics Theory of Mind AI will have?
Since Theory of Mind AI is still under research & development, its characteristics will likely evolve. However, based on the available information , we can anticipate that it will have the following characteristics:
Computers equipped with Theory of Mind AI will infer the objectives of entities around them from visible cues.
A system powered by Theory of Mind AI will answer simple “what if” questions about potential actions that entities around them might undertake.
Theory of Mind AI-powered systems will be able to simulate the consequences of their actions.
A robot or a system powered by Theory of Mind AI will be more communicative with human beings. E.g., such robots/systems will be able to explain their actions, and this is different from the current generation of AI.
Theory of Mind AI will consist of ML systems that can explain their decisions in languages that human beings understand.
A robot/system equipped by Theory of Mind AI should be able to understand the intent of another similar robot/system.
How to implement the Theory of Mind AI?
a-human-head-with-a-color-overlay-of-the-earth-and-a-series-of-1s-and-0s.webp
As research and development proceeds on Theory of Mind AI, examples of implementing it are emerging. Let’s review one such example.
This project in this example is one conceived by Neil Rabinowitz, a research scientist at DeepMind in London, and his team. This team has created a Theory of Mind AI-powered system named “ToMnet”, i.e., “Theory of Mind net”.
ToMnet observes other AI systems and learns about how they work, and it consists of 3 artificial neural networks (ANNs). An artificial neural network is a kind of an ML system, and it’s modeled after the human brain and how it learns. ANNs use algorithms that enable computers to learn by incorporating new data, and you can read more about ANNs in “What is a neural network?”.
Each of the 3 ANNs in ToMnet has computer programs, e.g., algorithms, moreover, they have connections that learn from experience. One of these ANNs learns the tendencies of the AIs under observation, based on what they did earlier.
The 2nd ANN builds an understanding of the “beliefs” of the said AIs.
Taking inputs from the outputs of the 1st and 2nd ANNs, the 3rd ANN predicts the next moves of the AIs under observation.