Understanding The Real Potential and Limits of Artificial Intelligence

Artificial intelligence (AI) is a branch of computer science dealing with the simulation of intelligent behaviour in machines. It is concerned with building machines that are smart and capable of performing tasks that typically require intelligence that human possesses. AI makes it possible for machines to learn from experience, adjust to new inputs and perform human-like tasks. Most AI examples such as chess-playing computers to driverless cars rely heavily on machine learning, deep learning, and natural language processing (NLP). Using these technologies, computers are trained to accomplish specific tasks by processing large amounts of data and recognizing patterns in the data. The term artificial intelligence was coined in 1956. Early AI research in the 1950s explored topics like problem-solving. In the 1960s, the US Department of Defence took an interest in AI and began training computers to mimic basic human reasoning.

AI has become more popular today because of advanced algorithms and improvements in computing power, storage, and availability of large volumes of data to train and test the AI algorithms. AI technologies present unlimited opportunities and we have to capitalize on the advancement of AI. During the convocation ceremony of the graduates of the Royal Institute of Management (RIM), His Majesty the Fifth King of Bhutan shared on making use of emerging technological advancements such as Digital Identification, Artificial Intelligence, Blockchain, Internet of Things, Big Data, Quantum Computing, Machine Learning, Virtual Reality and Augmented Reality, etc., which present unlimited opportunities for Bhutanese. His Majesty said, “Being a small nation makes us a smart nation, this is not out of choice but out of necessity. Technology is an indispensable tool that will be necessary to realize this aspiration.”

The improvements in technology have produced impressive advances in the field of image recognition, pattern analysis, and autonomous machines which has wide application in our everyday life. AI is progressing rapidly. The problem is that people are extrapolating many unrealistic expectations from these initial successes of AI, without recognizing the many constraints surrounding their achievement. Human intelligence existed since the dawn of humanity and is further honed by millions of years of evolution. AI is merely 70 years old. Machine intelligence is still pretty dumb, most of the time. So, to say AI equal human experts is a big statement to make. It's far too early for the human race to throw in the towel.

Today, people are misled by AI because of a phenomenon known as the Eliza effect. Eliza is an early NLP computer program created in 1966 at the MIT Artificial Intelligence Laboratory by Joseph Weizenbaum. Eliza responded to people's typed statements in the manner of a Rogerian psychotherapist. The computer was executing some very simple programming logic. But the people interacting with it ascribed emotional intelligence and empathy to all Eliza’s replies. The same phenomenon happens today in our reactions to the successes of AI. People overestimate the achievement of AI and underestimate human performance because we don’t take a second to think about how much human already know. People should not be misled by initial success of AI and believe in words of prominent figures of AI who does not have faith in humanity and God. There is already a lot of hype in the computer science world, that one day, AI would take over humanity. An AI takeover is a hypothetical scenario in which AI becomes the dominant form of intelligence on Earth, with computers or robots effectively taking control of the planet away from the human species. Some popular public figures, such as Stephen Hawking and Elon Musk, have recommended research into precautionary measures to ensure future superintelligent machines remain under the control of a human.

No matter how powerful AI system is, they are better than human at only one thing - at matching known patterns. They can only match the patterns they have learned, and they have limited capacity to learn more than just a few patterns. Humans are optimized for learning unlimited patterns and then selecting the patterns we need to apply to deal with whatever situation we find ourselves in. This is a skill that's human possesses. For example, Booking.com, a travel e-commerce company with 198 offices in 70 countries worldwide used machine learning to autonomously tag images. Technology giant like Google, Amazon and Microsoft Azure already offer generic image tagging services. But the problem Booking.com encountered was that those services don't tag images in a way that's useful in the Booking.com context. They may identify attributes such as 'ocean', 'nature', 'apartment', but Booking.com needs to know whether there's a sea view, is there a balcony and does it have a seating area, is there a bed in the room, what size is it, and so on. The back-end developers have had to train the machines to work with a more detailed set of tags that matches their specific context. If the task was given to image tagging human expert, it would have tagged the images flawlessly as per the requirement.

No matter how superintelligent AI products are, taking over humanity is a distant dream for a soulless AI machine. The core of AI such as machine learning and deep learning algorithms which consist of the thousand lines of code is a by-product of intelligence of the human designer. Unlike a human, AI systems have a limited capacity and can do and excel in what it is programmed to do and in its domain. Nothing more, nothing less. Human, on the other hand, can do unlimited things.

In the paper, “A comparison of deep learning performance against health-care professionals in detecting diseases from medical imaging: a systematic review and meta-analysis”, the authors suggest AI is able to interpret medical images using deep learning algorithm and equal human expert in medical diagnosis. AI may be able to perform better than a human expert in some domains like medical and scientific research, to claim that AI equal human which is as good as saying human is outwitted by AI, a full real-world testing in every field is required as suggested by Dr Raj Jena, an oncologist at Addenbrooke’s hospital in Cambridge. As a machine, AI will be able to do thing effectively and efficiently if it is repetitive and mechanical. A human expert that we compared with a machine 1 minute is not the same human expert. He would have learned something new in that second. The machine cannot do that. If a new learning library is to be added into AI, we have to design, code, and implement the algorithm. AI systems cannot be exhaustively tested. If they fail, they will fail in an unpredictable way. For example, consider a hypothetical example of AI preparing Emadatshi rice (unofficial national dish of Bhutan). Let’s compare an AI chef who can prepare Emadatshi and Bhutanese top chef who is curious to learn new things in cooking. If we make AI chef prepare Emadatshi 100 times, it is going to prepare the same menu 100 times with same taste and flavour. Human chef, on the other hand, the menu he prepares yesterday will be same as today, but the taste and flavour will be different because he will have learned the new style of preparing Emadatshi. AI chef is simply going to prepare the menu mechanically. Human chef, on the other hand has learned new way of preparing Emadatshi and is going to continue. Till his life, he would have learned hundreds of different ways to prepare Emadatshi rice. An AI chef, on the other hand, does not have to capability to learn on its own to prepare Emadatshi rice. It has to depend on the developer or it is programmed already to prepare in different ways already. Predicting things is not easy for machine, but human can learn. So, here it is the human who learns things first than AI. If AI can excel like a human in all aspects, not just some selected domains, then, we can rightly say that AI equal human experts.

In an article on “Artificial intelligence tool 'as good as experts' at detecting eye problems” from Guardian, scientist says, a new machine-learning system is as good as the best human experts at detecting eye problems and referring patients for treatment. The eye checking AI system was capable of correctly referring patients with more than 50 different eye diseases for further treatment with 94% accuracy, matching or beating world-leading eye specialists. According to the article, the five separate machine learning system is trained with 877 clinical optical coherence tomography (OCT) scans of a patient’s retina. Today people are living longer, stay in different geographical location, and are going to experience different types of eyes diseases which were not earlier detected and programmed into AI system. New detection can be easily read by human experts and accordingly carry out the treatment with experiences gained from working with the prevailing eye diseases. AI system needs to be redesigned, coded, tested, and implemented to read the new diseases. The process of creating new release or version of the AI system is going cost time and resources. Just like all the conditions must be met for the AI system to perform diagnosis like input should be correct, process is fully validated and right output is generated, if all the conditions are also met for the human expert, then, definitely, then human expert will be able to perform better than AI. People should avoid tunnel vision and do a thorough analysis of the place of AI system in human expert.

From Buddhism perspective also, Buddhist believe human possesses an extraordinary knowledge, power, and wisdom which cannot be matched by anything in the universe. Just like the universe is infinite, human intelligence is also infinite; every second, a new knowledge is gained. This is the very reason, why human is considered the most beautiful creation. No matter how intelligent AI system is developed and perform mind-boggling work or solve problems which outperform human in their specific domain, they are developed by intelligence of human and has limited capacity. If we look at the system as a white box, their internals are all product of human intelligence. Every bit and pieces of algorithms are designed, coded, tested, and implemented by human. An AI which can pass Turing Test will be able to perform flawlessly only in its domain, for example, medical diagnosis, but it is bound to fail, and require human assistance for the smooth functioning. Buddhist also believe AI can be compared to a human expert, but should not make a statement like AI equal human experts.