The weekend of 9-11 June sees the return of the Zurich Art Weekend. This year the festival looks at how artists are responding to the technological revolution of AI. Yesterday I went to Swiss Federal Institute of Technology ETH's Collegium Helveticum to see "Data Alchemy: Observing Patterns From Galileo to Artificial Intelligence".
I enjoyed the Synthetic Strokes exhibition which raises questions about how the development of AI and robotics is impacting on contemporary artistics practices posing the question, how does the practice and perception in viusal artistic creation change when given over to deterministic processes and mechanical action? What kinds of new coupling in the act of artistic creation are possible between a human and a robotic system? The visit inspired this blog on personality, AI and machine learning. How does the personality influence the development of AI?
Personality
Our ability as humans to problem solve and to find solutions to complexity is a unique characteristic of the human brain. Learning is a process in which we deal with the complexity of the world by acquiring knowledge or skills through study, experience or being taught.
There are many ways in which humans learn and gain insights. A #personality is one of the ways in which we perceive the world and also evaluate, think about, and act upon the world. Personality is an evolved solution to the problem of an overly complex world. Each human being has a unique personality structure that has developed through interactions with the world over time.
The major contribution of analytical psychology to our understanding of the personality is the discovery of the #unconscious, the ideas that as human beings we are part conscious and part unconscious and the two states are in constant dynamic relationship with each other. A personality structure is also shaped by the dynamic relationship of the conscious and unconscious minds and in interactions with the world. Some features of the personality structure are common or universal to all human beings. For example, there are four basic psychological functions of the personality: #thinking, #feeling, #sensation, #intuition.
Thinking is a process of cognitive thought, sensation is a process of perception by means of the physical sense organs or five senses, feeling is a process subjective judgement or valuation of something, and intuition is a process of perception by accessing one's unconscious or receptivity to the activity of the unconscious. In other words, the sensation function establishes that something exists, thinking tells us what it is, feeling tells us what it's worth, and intuition gives a sense of what can be done with something, its potential or possibilities.
The ways in which the four functions learn vary considerably. The thinking function accomplishes learning by regulating, planning, enforcing, naming, defining and understanding. The sensation function accomplishes learning by engaging, experiencing, enjoying, implementing, verifying and accounting. The feeling function accomplishes learning by validating, affirming, relating, judging, appraising and establishing the value. The intuition function accomplishes learning by imagining, knowing, divining, entertaining, envisioning and enabling (Beebe, J. (2017). Energies and Patterns in Psychological Type: The Reservoir of Consciousness).
Any of the four functions by itself is not sufficient for understanding the complexity of the world, all four are required for a comprehensive understanding. A fully functioning personality is able to have conscious access to the function or functions required or appropriate for a given circumstance but in practice the four functions are not equally available at one's conscious disposal, that is they are not uniformly or consistently developed in any individual. Invariably one or the other is more developed while the other is literally unconscious i.e. undeveloped, repressed, or even withheld from conscious effort.
#Consciousness brings a personality structure to life, to one's self awareness and to the awareness of others. Consciousness can be defined as 'the distinguishing feature of mental life, variously characterised as the (a) state of awareness as well as the content of the mind, that is, the ever changing stream of immediate experience, comprising perceptions, feelings, sensations, images, and ideas (b) central effect of neural reception (c) capacity of having experience (d) subjective aspect of brain activity (e) relation of self to the environment and (f) totality of an individual's experience at any given moment' (Corsini, R. J. (2002). The Dictionary of Psychology).
Personality functions by influencing our perceptions, motivations, emotions and actions such that people of different personalities actually experience the world differently. When working together to solve problems, groups of humans that include many different personalities are more adept for the ever changing environment. Personalities are worth learning about because they help us to relate to other people and to ourselves more earnestly, productively and with greater understanding. The differences in personality are necessary for human development, it therefore worth putting up with the idiosyncracies of everyone else for the benefit of complex problem solving. There are many ways that we, as humans work to solve complex problems, including the development of technology and social cooperation. Artificial Intelligence is a great example of humans working to solve complex problems through the development of technology.
Artificial Intelligence and Machine Learning
Machine learning is about teaching Artificial Intelligence (AI) machines how to learn from data. Just the like the human psyche, there are different machine learning approaches used for different types of data similar to the different ways in how the personality learns.
Machine learning is about teaching machines how to learn from data. By using machine learning algorithms to learn from data sets, machines can match or surpass human performance in many tasks. There are three learning approaches: supervised, unsupervised, and reinforcement learning as well as the learning tasks associated with each.
All these approaches to machine learning
follow the same fundamental workflow, which consists of four main stages:
Manage data: data is collected, prepared, and split for training and testing.
Train model: the task, features, and algorithms are chosen, and the model is trained.
Evaluate model: the trained model is assessed and improved.
Deploy model: the trained model is deployed for prediction on new data, and the model’s performance is monitored and eventually re-retrained.
Supervised learning refers to a situation where a task has an input variable and an output variable, and an algorithm learns to map the input to the output based on examples. (Learned-Miller, 2014:2). This is a process similar to the thinking function or cognitive thinking that is to say the machine's cognitive ability is supervised by a human's coginitive ability or thinking function.
Unsupervised learning is when a machine “learns patterns in the input even though no explicit feedback is supplied” (Russell & Norvig, 2016:694). This approach to machine learning is based on input variables only because there are no output variables in training the AI algorithm. This is a process similar to the intuition function. The machine is given free rein to discover patterns in data supplied by a human and from its own evaluation of the data.
Reinforcement learning is when a machine is not told how to process the data, or which actions to take, but instead learns by examining the outcomes that follow each behaviour (Sutton & Barto, 2018:1). This type of machine learning is based on learning-by-doing. This is a classic example of the sensation function which prefers to learn by doing. It requires no previous knowledge or experience, the machine will learn through practical experience.
The personality type of an AI data scientist can influence which of the three approaches are adopted for machine learning based on their own preferences for learning in terms of the four functions. There are of course other factors which also contribute to the machine learning approach adopted by an AI data scientist.
Each of the machine learning approaches has algorithms that are typically associated with it. Data and the learning approach are critical to machine learning and by using algorithms to learn from data, AI can match or surpass human performance in many tasks. Does this mean AI learns like humans? No. AI does not involve consciousness like human learning does. Instead machine learning is an approach that seeks to find patterns in data using statistics, mathematics, technology and trial and error (Hao, K. (2018). What is Machine Learning?)
The three learning approaches coalesce around thinking, sensation and intuition. The feeling function is notably missing from the approaches. The AI data scientist is responsible for ensuring the insights gained from AI is fair, legal and ethical. The feeling function is the process of the valuation of others. An AI data scientist whose feeling function is unconscious, repressed or withheld from others is likely to contribute to AI technology that is harmful to others. AI can perform highly complex problem-solving (such as unravelling intricate cancer diagnoses), but it can also suffer major setbacks (such as the potential for racial discrimination). The impact of an unconscious feeling function in the development of an AI algorithm can have unlawful and catastrophic impact on others.
AI shapes us and we also shape AI through our personality type. It is very much a collaboration between human and machine, similar to the dynamics of conscious and unconscious states. Whilst AI is adept at processing data in speeds well beyond human capabilities, humans remain more unique and intelligent. AI is a data processing or computational system, it does not have a body of human flesh like humans. The absence of the body in AI means humans remain psychologically more intelligent because of the body mind connection. AI does not understand our bodies, or the history of our personality. There are greater sensory modalities and perception in us which exceed AI capabilities.
The choice of learning approach to run the algorithm is a key part of machine learning. It is not always clear which learning approach is the most appropriate, it can partially depend on the personality type of the AI data scientist carrying out the learning task. Just like our own personalities have to choose which of the four functions is appropriate to use in a given situation. Given the valid concerns of AI, the feeling function strikes me as one of the more urgent functions that an AI data scientist should be consciously aware of in the development of algorithms to avoid creating harm to others.
#JungianBitsofInformation offers a new service to organisations. An Artificial Intelligence service to identify opportunities for AI in your organisation and guidance on the ethical considerations to address the common pitfalls of AI with a unique perspective from #analyticalpsychology. Contact me for more information.
In the meantime please enjoy some robotics created paintings from the ETH's Collegium Helveticum's AI exhibition "Data Alchemy: Observing Patterns From Galileo to Artificial Intelligence".
