We postulate Smartphones are in cybernetic relationships with humans and are analogous to the mind of robots. Right now, we have the beginning stages of technology like Neurallink to connect a computer to your brain, further pushing the relationship of a cyborg. In Humans, the Theory of Mind is the ability to understand other people and their behavior by attributing mental states to them, which is a future goal of AI.
Theory of Mind (ToM) is possible in AI. Smartphones and wearables are taking over the endless list of daily cognitive memory tasks like calendar reminders, dates, times, addresses, and phone numbers. Facial Recognition, Behavior Trackers, and the push for Emotional Intelligence Apps and Games will continue to be a catalyst for Deep Learning and ToM.
The current AI might not be powerful and capable enough to carry on the different cognitive and neurological processes that contribute to understanding the Theory of Mind. Still, it will likely, later on, be capable of understanding its behavior and that of people. We may understand more about the ToM by studying how it works with AI in smartphones and other machines. But first, let us try to talk more about the Theory of Mind and how it relates to AI in our smartphones or in general.
What Is The Theory Of Mind?
In psychology and, to a certain extent, philosophy, the Theory of Mind refers to the cognitive and neurological processes that allow human beings to attribute certain mental states to other people. It is a process where the person understands his own mental state and becomes empathic to the point that he could also attribute certain mental conditions to others.
Because the Theory of Mind explains and theorizes how we, as humans, can attribute certain mental states to the people around us depending on certain factors, it is also theorized that the same processes allow us to predict other’s actions and that of our own. We can predict how we or others might behave when placed in certain situations because we are putting ourselves in “one’s shoes” in such a situation.
Artificial Intelligence or AI has gone a long way regarding how advanced it has become, especially in our smartphones. While we may know that AI today is still far from what we often see in movies and television, it is certainly on track towards becoming as mentally capable as humans in many areas. But is the AI in our smartphones truly capable of empathy and understanding the mental state of others? This concept is where the Theory of Mind (ToM) comes in.
Theory of Mind in Artificial Intelligence
The reproduction of human cognition is one of the goals of artificial intelligence. As discussed in our article “Smartphone AI: Helpful Artificial Intelligence For The Beginner,” there are many aspects to AI.
There are also different types of AI classification depending on who you talk to, and there may be more. For this article, we are going to suggest that there are seven types of AI or AI-based systems:
- Reactive Machines
- Limited Memory Machines
- Theory Of Mind
- Self-Aware AI
- Artificial Narrow Intelligence (ANI)
- Artificial General Intelligence (AGI)
- Artificial Superintelligence (ASI)
Theory of Mind in AI may become relevant with advanced cognitive characteristics utilized in ever more complex neural networks, machine learning algorithms, and future breakthroughs that far surpass anything of today. The Computational Theory of Mind (CTM) also states that the mind itself is a computational system.
How the Theory of Mind Works
In the most basic application of how the Theory of Mind works, think of it this way to understand what this theory is all about. Let’s say you are working in a shoe store, and you saw Margaret taking her time looking at the different shoes on the rack. The Theory of Mind theorizes that you, being empathic and understanding her situation without even communicating with her, knowing that she is having difficulty picking out which shoe to buy. As such, you offer your professional assistance to Margaret in such a situation.
Given that example, you are figuratively putting yourself in Margaret’s place as you empathize with her situation. It’s similar to when you see someone looking in a kitchen pantry or cabinet, and you deduce that they are hungry or probably bored.
This ability has allowed humans an evolutionary advantage in interacting with the people and environments around us. We attempt to predict the mental state and actions of those around us based on certain situations and information presented to us.
AI Training to the Theory of Mind Understanding
Understanding how the human mind works is an ongoing work in progress. There are many approaches to the Theory of Mind. One is the Theory-Theory (TT) approach, which explains that, as children, humans develop many different ideas about how people around them behave. Meanwhile, the Simulation-Theory (ST) approach explains that we can essentially predict actions and attribute certain mental states to others by taking their perspectives as our own and putting ourselves through simulations while using other people’s perspectives to predict how they would react.
We also have companies like Deepmind’s Theory of Mind using a neural network, ToMnet, which comprises three neural networks and uses meta-learning to build models of the agents it encounters from observations of their behavior alone. There have also been multi-faceted experiments in robotics mimicking the Theory of Mind, and the list goes on, as does the progress.
From his Hypothesis and Theory article in Frontiers in Robotics and AI “Experiments in Artificial Theory of Mind: From Safety to Story-Telling” Alan F. T. Winfield discusses in depth the proposal Theory of Mind. As he says in his paper:
“In summary, we contend that the experimental work outlined in this paper does demonstrate a number of components of theory of mind and can reasonably be described as “experiments in artificial theory of mind.” The main hypothesis of this paper, that simulation-based internal modeling can form the basis for artificial theory of mind has, we argue, been demonstrated. Whilst far from a complete solution, we propose simulation-based internal modeling as a powerful and interesting starting point in the development of artificial theory of mind.”
Frontiers in Robotics and AI “Experiments in Artificial Theory of Mind: From Safety to Story-Telling” Alan F. T. Winfield
However, even though we may have gone far in our studies regarding the Theory of Mind and its different approaches, there are still a lot of questions. Questions regarding the various cognitive processes that gave rise to such a theory because not even our minds are always understanding why we are capable of essentially putting ourselves in the shoes of others.
Is the AI In Our Smartphones Capable Of Understanding?
You now understand more about what the Theory of Mind represents and how it theorizes the way humans can attribute mental states and possible actions to others. If the same theory applies to machines as described above in robotics, is the AI in our smartphones and gadgets, in general, ever capable of Emotional AI, understanding, and empathizing?
For starters, let us talk more about how the AI of today’s world works. As advanced as our AI tech has become, it has always been more concerned with cold cognition in the sense that today’s artificial intelligence focuses more on extracting and applying the information it got from data.
Thanks to the rise of the digital world and the growing deep neural network provided by the internet, today’s AI now has a larger and deeper pool of data to work with to extract the information it needs to perform the tasks it was programmed to do. This technology is generating Smartphone applications in a plethora of areas about cognitive functions.
Please see some of our other interesting articles, such as “Mobile Devices From Smartphones to Robots” and “Why Is AI So Expensive For Business?”
AI Programmed to Play Games
A good example of this is an AI that was programmed to play strategic games. Because AI is faster at processing information and can use a wide pool of data to draw information out of, it is quite easy to beat even the best human players in certain games of strategies. We have seen AI becoming better than actual doctors in specific areas using and processing data to diagnose certain medical conditions.
Current AI technology is progressing in areas to consider how to simulate how a human mind functions. The AI in machines and smartphones is smart enough to extract and process data faster than any human mind. It will use information and experience to learn from them and develop a potential mental state as it evolves.
Going back to the ToM, the important factor in attributing certain mental states to other people is having similar thoughts and emotions. Currently, machines don’t have their own unique mental state because they are only programmed with logic and reasoning processes.
Structured Cognitive Processes
The machines are also not currently structured with the cognitive processes that allow computer minds to develop individual mental states based on their particular experiences and learnings. This ability would then enable them to theorize and simulate how people act or react in certain situations when this does happen.
With that said, the AI in our smartphones and machines are not yet capable of understanding and attributing mental states. But that doesn’t generally mean that it isn’t possible. Artificial General Intelligence (AGI) will likely develop and be capable of social interaction. It will understand the rules of socializing and of moral norms up to such a point that the AI itself can integrate with human society and even with other AI.
For that to happen, our AI must recognize more advanced and complex patterns while also learning how to learn. We see progress in many areas of Machine Learning (ML), Deep Learning (DL), Reinforcement Learning (RL), and Unsupervised Learning (UL). Unsupervised Deep Learning is making incredible progress and doing things never thought possible.
How Can We Make Use Of The Theory Of Mind In AI, Smartphones, and Other Machines?
Self Driving Cars
How would we benefit from such advancement in artificial intelligence if we can use the Theory of Mind in the AI of our smartphones and that of our other machines?
A good example of that is a smartphone app that can integrate with self-driving cars. The biggest challenge for self-driving cars is that it is impossible to completely program them to know all of the possibilities that can happen on the road.
However, by potentially integrating the Theory of Mind with what these machines are capable of, our smartphones and our cars would be able to attribute certain human behaviors to the other people on the road given the data presented to it on the road so that it will be easier for the AI to predict what to do next.
AI and Theory of Mind in Medical Apps
Another instance where we can use the Theory of Mind in smartphone AI is by developing apps in the medical field that can correctly diagnose certain medical conditions that can be difficult to diagnose without the help of a human doctor.
For example, with a combination of AI and programs, Deepminds AlphaFold created a “Highly accurate protein structure prediction.” Work with neurological and mental diseases can be possible for AI to diagnose if it is capable of empathy for humans. The continued progress of AI in conjunction with the Theory of Mind would facilitate this.
It is also possible that by studying how the Theory of Mind works at a more synthetic level in machines, we will also understand how it works in the human brain. Given How would we benefit from such advancement in artificial intelligence if we can extrapolate and factor in the different cognitive processes that the AI goes through when trying to empathize and understand certain mental states.
Emotional Artificial Intelligence (Emotional AI)
Emotional Artificial Intelligence describes technologies that use computing and artificial intelligence to interact, sense, and learn with human beings and their emotional life. Social Media to Hospitals and all aspects of society will be using and contributing to its function.
This technology will soon permeate our lives and will be present in many places such as hospitals, classrooms, restaurants, shopping malls, stores, in your building, and at your job. It will also be in everyday objects like your Smartphone and voice assistants, smartwatches and wearables, toys, games, and cars, just to name a few.
Emotional AI is a probable precursor to the Theory of Mind and, in conjunction with Biometric technologies, will try to read the emotions of everyone it comes into contact with. The Data Sets created by this technology in conjunction with AI and Deep Learning will eventually grow, train, and permeate the fabric of our society.
Biometrics
Like those in your smartphone, biometrics allows objects to be identified and authenticated based on recognizable and verifiable data, unique and specific. Biometrics is an amazing technology and, if combined with AI, is an incredible medium of interaction between machines and humans.
Think of the possibilities with Biometrics integrated with Deep Learning and the Theory of Mind. There are many types of biometrics technology:
- Behavioral Biometrics
- DNA matching
- AI-Powered Gait Detection
- Face Recognition – Facial Biometrics
- Iris Recognition
- Ear Acoustic Authentication
- Voice Recognition
- Palm recognition:
- Fingerprint Recognition
- ‘Composite’ Biometric Authentication
Conclusion:
The possibility of what AI can do with the integration of the Theory of Mind seems endless, given that we are making machines and smartphones think more and more like humans. To some extent, it is both exciting and scary, but such technology will only become more present, as we are only beginning to scratch the surface of what AI can achieve.
References:
Knowing me, knowing you: theory of mind in AI
Experiments in Artificial Theory of Mind: From Safety to Story-Telling
[1802.07740] Machine Theory of Mind
Are ‘you’ just inside your skin, or is your smartphone part of you?
DARPA Machine Common Sense (MCS) program