AI is penetrating but people often argue how. This article differentiates sense from nonsense.
Artificial Intelligence (AI) likely impact various aspects of our lives and the world. Notably, the generative AI like ChatGPT has penetrated work and study. Can AI be so capable that it can make sense? To explore this, we need to understand how humans and AI think when a task is assigned to them.
When you ask a human to do a task like bring a medicine from a pharmacy, they will not consider it as their mission or purpose. This means they won’t kill a pharmacist to get a medicine. However, the problem with AI is that it is given a fixed objective. The algorithm wants the scientist to clearly define the objective. For example, if you want AI to resolve acidification of the ocean it might do this with a catalytic reaction with extreme efficiency but it would consume so much oxygen that humans will die.
Some might argue that we can specify to AI that be careful about oxygen in the environment and also other things like aquatic animals, seaweed, and so on. We have seen this in many movies like Clouding with a Chance of Meatballs or Age of Ultron. However, humans often know that they do not know all the things that we care. For example, if you ask a human to get you a breakfast and you are in the Hotel George Sand in Paris where a breakfast is €15, it’s reasonable for the waiter to come back and say, Sir its €15, do you want to order or I could go next door and get one? This is a normal thing for a reasonable person) to do. When you ask a painter to paint the house, the workers will remove necessary items like furniture for the time being and put them back once the job is done.
At present, AI lacks this type of thinking capability that humans have unless we share with them the complete objective. If the AI don’t know the full objective it exhibits behavior like removing the entire oxygen without permission in the acidification of the ocean. AI behaves like this due to the uncertainty of the machine about the actual objective but when you created machines with a specific objective you start to observe psychopathic behavior and such behavior also apparent in humans.
So far the AI system developed are specific purpose. The Generative AI’s purpose is to answer user queries, the face recognition recognizes faces only, and the chess.com AI is only capable of playing chess and analyzing moves. There is no general purpose AI but what will happen if such AI is introduced. This idea dated back to Aristotle who argues that if there are full automated machines that can weave and even produce music without human intervention then workers would be doomed. Equally, Keynes in 1930 gave the idea of “technological unemployment” arguing that if machines do human work then people would be unemployed.
Warehouses that have tons of piled items reflect Aristotle and Keynes’s ideas. Currently, there are partially automated e-commerce warehouses in which machines identify the article and workers picked them. In a fully automated warehouse, the machine would be able to pick the article from a wide variety of articles accurately. If that would happen, it is expected to eliminate 3-4 million jobs.
Forster wrote in “The Machine Stops” that handing over the entire civilization to machines will prevent humans to understand their civilization and won’t be able to pass this knowledge to future generations. Wall-E provides a good example of this. Some might argue that books pass knowledge to future generations but books cannot run civilization so humans need to teach future generations. We have to understand that we must not allow AI to break the chain of passing knowledge to future generations.
The existing AI is upgrading day-by-day. Generative AI is not only answering queries but also analyzing, summarizing and streamlining things and it is going beyond. This expansion of tasks mean that general purpose AI won’t be too far but in reality John MacAfee – one of the founders of AI – argue that creating general purpose AI requires many brains like Einstein to create so its arrival is very far indeed.
