Can we trust the intuition of AI?
Intuition is the link between perception and reality; the grey area where logic is incomplete without emotions. The possibilities of Artificial Intelligence developing logical element of intuition are green because of its detail-oriented observation of all inputs and the result of their implications. The question is, can AI intuition be trusted with decisions that can affect human lives? Can AI intuition evolve to entirely replicate human intuition?
Dreyfus insisted that “human beings have an intuitive intelligence that ‘reasoning’ machines simply cannot match.”
US courtrooms are using risk-assessment algorithms to decide the release or arrest of defendants on bail. The AI scores them on different aspects like the possibility of them committing another crime or that of skipping court, enabling Judges to decide accordingly. However, in 2016, news broke that the AIs are biased in their judgement as observed by their leniency towards Caucasians and assignment of more jail time for African-American in the backdrop of same crime.
The algorithms developed by the risk-assessment AIs were an outcome of their observation of human behavior and humans have biases. AI is a technology in its infancy and it is learning from its parents-Humans. The flaws of humans will be observed by AIs, but unlike humans, they won’t employ their own values and emotions to distinguish between right and wrong. Hence, their decisions will be reflective of these flaws.
Intuition is a cognitive function that is governed by non-logical reasoning, whereas, AI is governed by logical reasoning. Their intuition can be developed enough to fill in the gaps between information to take correct decisions based on logic. However, deep logical reasoning is not enough for some decisions where a touch of emotions is required.
Our mistrust on its intuition is also because of the inexplicable nature of AI actions. Even the developers of AI can’t reason the decisions of Artificial intelligence. To be able to trust this technology with the power of decision making in crucial sectors like Government, Health, Judiciary or Military, we need a truly interpretable AI.
Human intuition supersedes AI’s abilities because of another reason as well. IBM’s Watson super computer employs the Deep QA/UMA algorithms to provide solutions to an enquiry. The whole process of analyzing and calculating the correct answer takes less than 3 seconds, which is still slower than an estimated 20 millisecond human intuition reaction.
AI’s logic-based processes need time and large search spaces to come up with a calculative decision. Human intuition is quick in real life scenarios where crucial decisions are needed to be made in milliseconds.
In a world where Artificial Intelligence will soon be replacing humans in different sectors, the decision-making power will be concentrated in human hands only where the decisions will shape the future of mankind.