Human beings have been obsessed with the concept of intelligence and have developed various instruments to understand and measure it. Intelligence is essentially the use of the brain to understand complex and diverse phenomena. There are two types of intelligence and these are individual and group intelligence. Individual intelligence is the type that allows a single individual to perform complex diverse tasks. Group intelligence is the type that allows a population of individuals to perform complex diverse joint tasks.
An example of group intelligence in nature is the colony of ants which individually are not very intelligent but as a group are so intelligence that they build complex ant hills that are efficiently constructed with sophisticated air conditioning. There has been many ways of quantifying intelligence such as intelligent quotient (IQ) and emotional quotient (EQ). Intelligence is a utility that people value and take rather personally. Games such as chess and Go remain popular because they are deemed to improve children’s intelligence. We regard it as a personal insult if our children are not considered ‘smart’. There is even a whole emergent academic discipline of a ‘growth mind-set’ that trains students that intelligence is not fixed and that it does not determine academic performance.
Understandably, as people we are proud of our intelligence. Yuval Harari in his book Sapiens states that what has made us the most intelligent species is the discovery of fire. With fire we were able to burn forests and harvest burnt meat. This instantly put us at the top of the food chain. The burnt meat meant that food was now partially digested outside the human body thus leaving us with excess energy that we previously used for digestion and because of these reasons our brains and heads started to grow big. Because of this growth in brain size, child bearing became very dangerous and only children who were born prematurely were the only ones that survived. To this day our children cannot fend for themselves for a longer period than other mammals. Because of this reason we became social animals that could dream, plan and organise ourselves in large numbers.
This is what has made us the most successful intelligent species on this planet and at the top of the hierarchy. Yet, we don’t exactly understand intelligence fully especially its mechanisms, essence, matrices and how to improve it. The brain is the primary driver of intelligence. As humans we have five ways of getting information from our brains and these are through our five senses which are touching, smelling, seeing, hearing and tasting. Can we design technology which will access information directly from the brain? If we can access information from the brain, can we create a feedback mechanism that will change our brains directly to improve our intelligence?
Several disciplines, such as psychology, philosophy, engineering and computer science, study intelligence within their disciplinary boundaries. However, the multi-disciplinary study of intelligence remains elusive. An intelligent agent operates in an environment by activating activities to reach its goal. When the environment is predictable, the activities that need to be executed to reach a goal are straightforward whereas when the environment is uncertain the activities become complex.
Physicist Alex Wissener-Gross defined intelligence as a way to maximise future freedom of actions. In general, intelligence summarises aspects associated with humans such as flexibility, adaptability, critical thinking, originality and creativity. Scientists have developed ways of creating an artificial intelligence (AI). It is artificial precisely because it is man-made. AI systems such as machine learning are black boxes because we normally give them information and they process the information and give us the output. They are called black boxes because we do not know what is happening inside them.
Deep learning is a machine learning algorithm that is complex, handles large number of variables and its structure has many layers. These machines like children learn by repeated exposure to examples, feedback and modification often with implicit or imperfect rules. Behavioural psychology has observed that human brain has limited capacity for comprehending and processing information. Furthermore, decision making through prediction does not always require meaning and understanding.
Cognitive psychology has demonstrated examples of human beings fabricating reasons to explain their behaviours and random events and are good at story-telling. For AI to become truly human it will have to master skills such as explaining, story-telling and bluffing. Marwala and Hurwitz in their work demonstrated that AI machines are able to learn how to bluff. Despite all these achievements, AI machines remain black boxes and are not easily explainable.
The European Union has adopted a new law called the General Data Protection Regulation (GDPR). This law effectively states that when a decision is made by a machine such as neural network, the people affected by its outcome have the right to inquire why it made such a decision. This effectively implies that the black-box nature of AI can be legally challenged. In the current state of the art in AI, decisions cannot be explained explicitly and this is similar to experts’ intuitive minds.
There is this idea that AI machines will surpass human intelligence and this is what is called singularity. It should be noted here the current state of AI is a specialised AI and here machines are trained to do a specific task. In this regard, Google Deepmind AlphaGo was fed typical Chinese chess Go strategies by watching human experts play and mastering the game though trial and error and was able to defeat a human professional player. Its successor AlphaGo Zero taught itself from scratch, derived the rules by itself and it beat its predecessor AlphaGo.
The big question of our time is whether AlphaGo Zero can be able to learn how to drive, make tea and perform other human activities on its own. Some of the suggestions that have been suggested is to capacitate these machines with capabilities akin to biological evolution. If this is successful, then another type of AI which is called general AI where these machines are able to do diverse tasks can emerge. In this way a single machine can be able to play chess, make you a cup of tea and take your children from school.
As AI advances rapidly on this specialised trajectory, human’s comparative advantage might need to be re-examined. Our current education system is based on specialisation. A popular joke within the university corridors is that a Doctorate which is the highest academic qualification trains graduates to know everything about nothing. In many ways, this increasing specialisation is needed and the golden rule is that anyone needs 10,000 solid hours to become an expert.
In this era where AI is surpassing human beings at handling highly specialised work, how should human beings capacitate themselves? We are living in an era where AI machines play better chess than humans, play better Chinese game Go better than humans and interpret medical images better than radiologists what is to happen to humans? As AI becomes very good in highly specialised areas, how shall we compete with our limited brain capacity and limited physical hours.
The first thing we will need to take advantage of is our ability to understand concepts and to multi-task. The second advantage that we should exploit is our ability to cooperate. The third is our ability to show kindness and empathy. It is clear that the era of domination by specialists who require individual intelligence is coming to an end and it is being replaced by generalists who require group intelligence. This era will require that we deliberately teach at universities across and not in disciplines to develop both group and individual intelligence.