Artificial Intelligence

A World full of Magic

Patrick Quanten

I don’t know about you, but I am captivated by magic and illusions. To me, it is fascinating to experience being present and attentive, and still being deceived. How do they do it? Repeatedly. The allure of the seemingly impossible, combined with the skill and mastery of magicians, has left me spellbound, questioning the very fabric of reality. At the core of this craft, it turns out, are eight fundamental types of magic tricks. Things disappear or appear; things float in the air; things penetrate solid surfaces; restoring or transforming things; things swap places; predicting the future. But what are the intricate methods magicians employ to perform their astonishing tricks? What are the secrets of misdirection, sleight of hand, and the psychology behind creating illusions that seem impossible to the naked eye?

• Misdirection is one of the magician’s most powerful tools. By diverting the audience’s attention to one thing, a magician is able to conceal his secret moves. Through carefully planned gestures, words, and actions, magicians lead the spectators’ focus away from the real secret behind the trick.
• Sleight of hand is another crucial skill that magicians develop. It involves the dexterous manipulation of objects to create seemingly impossible transformations and disappearances. Magicians master delicate movements and precision required to execute magic.
• Psychology plays a significant role in magic tricks as well. Magicians understand the human mind’s vulnerabilities and cognitive biases, and they use these insights to their advantage. They exploit our propensity for selective attention and our tendency to perceive patterns even when they don’t exist.
• Customer experience, service and support – My experience as a customer is that no chatbots or virtual assistants have ever solved the problem I presented them with. In many cases these computer programmes didn’t even understand the problem.
• Fraud detection – I am now being investigated for fraud because I transferred some money from one bank account to another. This has been ‘identified’ by AI as an unusual transaction. And here is the good part. Because a computer programme says I have done something wrong, I have to prove my innocence. With the use of AI, one is guilty until proven innocent, because AI is always right and is unable to identify its own errors.
• Personalised marketing – Being bombarded with offers for things I haven’t asked for is an invasion of my privacy. Even the fact that ‘they’ know what I look at, read, buy, is an invasion of my right privacy. I do not feel I am benefitting from personalised marketing. I didn’t ask for it. And I cannot opt out of it.
• Human resources and recruitment – Selecting the best person for the job can be better done by AI. Nobody needs my opinion anymore, not even about who I want to work with. How ‘best’ is defined in AI is lost in the algorithms.
• Streamlining repetitive coding tasks associated with application development and accelerating migration and modernisation of legacy applications – Basically, automatically updating systems. AI determines when that is required, and it wants things done quickly. I cannot recall how many times a computer programme has failed after an automatic update, which was decided by either a build-in system within my own computer or by a company providing my computer with a service.
• Predictive maintenance: AI can analyse data from sensors to forecast when maintenance will be due and predict equipment failures. – Of course, making the decision that maintenance is due or that failure is imminent is depending on what limits one sets on the reading of the sensors. An automated programme that is to be trusted because the industry itself now decides when I need to buy a new piece of equipment. The toothbrush itself now instructs me when it is time to replace it!

“The magician and the politician have much in common: they both have to draw our attention away from what they’re really doing.” – Ben Okri

Magicians often play with our expectations and manipulate our attention. They know that our brains have limitations in processing information, and they capitalise on these limitations to create moments of wonder. For example, a magician might perform a trick that relies on our inability to accurately track multiple objects simultaneously, using misdirection to shift our focus away from the secret move. In addition to cognitive biases, magicians also leverage psychological principles such as suggestibility and visual perception to create their illusions.

And then there are the stage illusions, the grand form of magic. Levitation is made possible through the use of strategic choreography, hidden supports, and misdirection. Sawing a person in half is performed by using cleverly positioned mirrors, hidden compartments, and intricate mechanics. The disappearing act is done through well-timed distractions, secret trapdoors, and carefully orchestrated lighting. One of the most essential elements behind the scenes of magic is the presence of hidden equipment. From invisible threads to hidden compartments, magicians employ a myriad of clever devices to execute their mind-boggling tricks seamlessly. By strategically concealing these tools, they maintain an air of mystery while delivering breath-taking performances that captivate audiences. The construction of props is another secret revealed behind the scenes. Magicians have perfected the art of designing and building props that not only look ordinary on the surface but also harbour secret mechanisms.

Magic often seems like pure deception, but there’s science behind it too. The art of creating illusions is rooted in the principles of psychology and physics, making it a fascinating blend of artistry and scientific understanding. One of the key factors in magic is perception. Magicians exploit the way our brains interpret visual and auditory stimuli to create illusions that seem impossible. They understand how to manipulate our attention and trick us into seeing what they want us to see. By studying human psychology, magicians can control our perception and make the impossible appear real.

Oh my goodness, my mind has floated up in the air just thinking about the amazement of magic and trickery. I apologise. What were we going to talk about? Oh yes, artificial intelligence. Okay. Clear my throat and try again.

Artificial intelligence (AI) is ‘technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy’. Why would anybody want to simulate human learning if you can possess the real thing? Applications and devices equipped with AI can see and identify objects. They can understand and respond to human language. They can learn from new information and experience. They can make detailed recommendations to users and experts. They can act independently, replacing the need for human intelligence or intervention (a classic example being a self-driving car). If there is no need for human thinking or human action anymore, then I better disappear because ‘you’ – whoever is creating and using AI – don’t need me anymore. You, with the help of your artificial tool, can work out the solution and execute the necessary action without me.

But in 2024, most AI researchers and practitioners - and most AI-related headlines - are focused on breakthroughs in generative AI (gen AI), a technology that can create original text, images, video and other content. To fully understand generative AI, it’s important to first understand the technologies on which generative AI tools are built: machine learning (ML) and deep learning.

Machine learning involves creating models by training an algorithm to make predictions or decisions based on data. It encompasses a broad range of techniques that enable computers to learn from, and make inferences based on, data without being explicitly programmed for specific tasks. The simplest form of machine learning is called supervised learning, which involves the use of labelled data sets to train algorithms to classify data or predict outcomes accurately. But one of the most popular types of machine learning algorithm is called a neural network (or artificial neural network). Neural networks are modelled after the human brain's structure and function.

Deep learning is a subset of machine learning that uses multi-layered neural networks, called deep neural networks, that more closely simulate the complex decision-making power of the human brain. These multiple layers enable unsupervised learning: they can automate the extraction of features from large, unlabelled and unstructured data sets, and make their own predictions about what the data represents. Because deep learning doesn’t require human intervention, it enables machine learning at a tremendous scale. The only thing that is missing in the decision-making process is emotions and senses. The machine cannot evaluate on the basis of ‘I have a feeling this is not okay’.

Generative AI, sometimes called ‘gen AI’, refers to deep learning models that can create complex original content - such as long-form text, high-quality images, realistic video or audio and more - in response to a user’s prompt or request. Generative models have been used for years in statistics to analyse numerical data. But over the last decade, they evolved to analyse and generate more complex data types.

And then the first important question: Why would you choose not to have any human involvement anymore in the thinking and acting processes of life? Well, the people who develop and fine-tune AI provide us with an answer. I am not quite sure why AI doesn’t answer this question itself!

AI offers numerous benefits across various industries and applications. Some of the most commonly cited benefits include:

• Automation of repetitive tasks.
• More and faster insight from data.
• Enhanced decision-making.
• Fewer human errors.
• 24x7 availability.
• Reduced physical risks.

So, numerous benefits for various industries and applications. No benefits to the ordinary person then?

Are they assuming AI will never encounter any (technical) problems, will never make mistakes and will always have a superior outcome?

The real-world applications of AI are many.

Scientists are enthusiastically imagining ways in which artificial intelligence (AI) tools might improve research. Why are AI tools so attractive and what are the risks of implementing them across the research spectrum? Here we develop a taxonomy of scientists’ visions for AI, observing that their appeal comes from promises to improve productivity and objectivity by overcoming human shortcomings. But proposed AI solutions can also exploit our cognitive limitations, making us vulnerable to illusions of understanding in which we believe we understand more about the world than we actually do. The proliferation of AI tools in science risks introducing a phase of scientific enquiry in which we produce more but understand less. 

Interpretability is important because it allows humans to place trust in a machine when used in the real world. If a robot or AI can explain its actions, then humans can decide whether it needs adjustments or can be trusted to make fair decisions. An interpretable system also enables the users of technology - not just the developers - to understand and trust its capabilities. However, interpretability has long been a challenge in the field of AI and autonomy. The machine learning process happens in a ‘black box’, so model developers often can’t explain why or how a system came to a certain decision. But they remain convinced, and they want us to be convinced, that AI has come to the ‘right’, to the ‘best’, solution.

Weizenbaum, who was a mathematician, computer scientist, and a student of psychoanalysis, was one of the founders of modern artificial intelligence who not only invented the first chatbot (Eliza), but also built early (mainframe) computers (back when they used vacuum tubes and took up entire rooms) for the University he was studying at, General Electric, and for the Navy. He said. “Artificial Intelligence is an ‘index of the insanity’ of our world“.

The reason for such a statement is that artificial intelligence is not something that occurs in nature. It cannot, therefore, be part of humanity as human beings are encompassed within nature. AI has no part in that. If it isn’t real, in that sense, it can be seen as ‘an illusion’. The dictionary defines artificial as ‘made by people, often as a copy of something natural’ and as ‘not sincere’. So it looks like nature but it isn’t. It is not ‘sincere’, not genuine, not real. It looks like the real thing, but it isn’t. Just like a magic trick in which something disappears, while in reality it is still there. Truly believing a magician has made someone float up in the air is insane.

And while the effect of the trick may have a potential useful purpose, the way it has been achieved is unknown to the observer. The potential danger lies in the fact that ‘make belief’ is considered to be ‘real’. Making something disappear out of sight by putting it in a secret compartment may give the impression it is gone and that you therefore have to proceed without it. It is lost to you because you truly believe it is gone, while in reality it only has been hidden from you. And the reason you believe it is gone, is that you don’t know how it disappeared.

When the magician no longer knows how something has disappeared or floats up in the air or has changed place, then he can no longer perform. His life ends right there. Losing control of the process completely invalidates what is being shown.

Because a robot or AI does not explain how it does something or has come to a certain conclusion, you cannot understand how the effect has been created. In other words, you cannot evaluate what has happened. When you know you are watching a magician perform, your brain will tell you that it can’t be true and that it, therefore, must be a trick. So when he makes your wedding ring disappear you don’t panic and you patiently wait for him to return it, from somewhere, in some way. However, when you watch the effect of AI you are not aware that you are watching a magic show and you accept the fact that what is lost, is not going to appear again.

When AI creates images, sounds or texts you can no longer distinguish these from the ‘real’ thing. Of course, the images, sounds and texts themselves are real, as far as our observation goes, but how have they been accomplished? Who has created them, and why? Just as in a magic trick, the reality is that you have lost your wedding ring. But how you came to lose it is a mystery to you. Why is this important? When you know how something has happened you can either retrieve the situation or you can understand it. In the first instance you go back to reality. In the second you are able to accept reality. When you don’t know how, you no longer know what reality is. You will keep wondering whether there is a hidden compartment somewhere. And if you don’t, you have been sucked into an illusion. Believing the bank actually has your money because a figure on a computer screen says so, is an illusion. It is not the reality.

When things are thought to be part of reality, part of true life, while they are not, we begin to live in a world where fiction and reality are intermingled in an indistinguishable fashion. You are unable to separate reality from fiction. This way, you no longer know what is real and what is artificial. This is the basis for an insane world. Discussions will arise, conflicts will emerge, as a result of the merger of reality and fiction. And nobody will be able to settle the argument because there is no longer a distinction between arguments based on reality and arguments based on fiction. What we believe is, to us, reality, and nobody is able to prove or disprove your position. This is insane!

Illusion and reality have become inseparable. An illusion is something that is not really what it seems to be. In our modern context we call it virtual. The adjective virtual is used to describe something that exists in essence but not in actuality. It looks real, but it isn’t. And that wouldn’t be so bad as long as we could tell the difference. With AI we no longer can. This leaves us wide open to being taken over by a virtual reality, by an illusion, without any opposition on our part. We are no longer aware that we are watching a magic show. We quite happily take part in it and we believe that what our senses pick up is real. We are mesmerised by the show. We are overwhelmed by the unbelievable things we witness. So much so that we no longer search for how the trick has been achieved. Once you no longer consider the fact that it is all magic and unreal, you no longer question the reality of what you are seeing or experiencing. Now you are part of the illusion. Now you are defending the illusion. It is no longer the magician who tells us “this is real”. Now it is the spectator who insists that it is real, because I was there – I saw it with my own eyes.

Having computers that compute figures and analyses statistics is one thing. It is another to have computers telling you who is guilty of a crime, who is best suited for a job, when you need to replace your toothbrush, whether your faeces is normal. Decision making handed over to a computer is insane. When I analyse a situation and come to a conclusion, it is my conclusion. Anyone who knows me and my reasoning can follow how I reach my conclusion. When you survey the same situation, you may reach a different conclusion. That is your conclusion, based on how you approach the situation. And anyone knowing you can follow your analytic route towards that conclusion. And the same goes for Pete, Marry and Joe Blogs. However, when AI reaches a conclusion nobody knows how it got to that point. Not even the creators of the computer programme. It is a fact that model developers do not know how AI reached their conclusions. If there is no way of tracking down the path of reasoning, there is no way of evaluating the outcome. And more importantly, as far as the government and any authority is concerned, no way of criticising the outcome.

We are trapped in a mist of illusion and magic, created by AI. The only way to allow the mist to lift is to ban AI from our lives, to return to reality. The basic rule for making that judgement is that we need to know, in all circumstances, how the conclusion was reached. We need to be able to evaluate the rationality behind the process. We need to be able to make sense of it.

We are rapidly losing the freedom to think for ourselves, because we accept that a computer programme is ‘better’ at it than we are, that a computer programme ‘knows more’ than we do. This may be the case for fixed data, such as numbers. However, we are rapidly losing sight of the fact that no computer programme knows anything about life. And what is more: it will never know anything about life!

And humans live life. Humans live in a natural world, not in an artificial world. AI can never encompass all necessary elements to understand and evaluate life, because AI isn’t a natural thing. It is an intelligence that is artificial not natural. It can never possess the intelligence of my dog.

November 2024