With that survival mechanism came experience. With experience came experimentation. We tried something, we observed what happened, we learnt from it, and from that came an inference. Over time, these inferences became knowledge, fact, rules, data, whatever you want to call it, in whatever order. In the world of man-made rules, everyone has their own theory and everyone is right in their own world. I want to call this “The Many Rights Theory”. Well, who coined it? Me. Just now. While I was typing this. Yes, I don’t rely on AI to make sentences for me. And that is my theory. You might think that is wrong because we need to make the best of the tools we have around us. Your friend might think we must use the best of all the things we have around us. No one is correct. No one is wrong. This is exactly “The Many Rights Theory”.

This is what sociologists have done. They observed the world from their viewpoint and documented it. Then we read it, discuss it, and debate around it. We debate and discuss what is right and wrong. Is that foolish? Is that smart? We all have our own theories again. The more we read, the more we may lose some of our originality. At the same time, the more we read, the more we build perspectives. Everything around us is still debatable. But then, what must one remember is that one can make sense of things, whether they agree or disagree, because they have done a lot of exploration and experimentation. So, “The Many Rights Theory” is: In a world governed by man-made rules, there is no single right; everyone is right within the world they have constructed for themselves.

Now what is happening with AI? Everyone is making predictions. Where do these predictions come from? Experience? Inferences? Or just random assumptions because I feel so? Someone who has spent years observing, learning, experimenting, and understanding a field might have a better chance of seeing what is coming. But even then, how much can one really predict? How biased are we? Time will tell. The real principles will eventually stand out, and most of us may not even recognise them when they do. Reading history teaches us wisdom. Reading principles teaches us why something works and why something breaks. That is the need of the hour. AI will change. The way we use it will change. Our understanding of it will change. Perhaps our predictions will change too. Maybe that is the nature of everything we know. We make sense of the world from where we stand, with what we have experienced and what we have understood so far. But if everyone is right in their own world, then who is actually wrong about AI?

I fear research is slowly walking the same path. In our attempt to make research teachable and accessible, we have given it structure. Structure slowly became a process. The process became a ritual. Today, we often mistake following the ritual for doing research. Somewhere along the way, we seem to have taken away the very essence of what research was meant to be.

The heart-aching part of research is that many who graduated through this methodical approach of how research has to be done push the very same approach onto their juniors. Instead of guiding them towards the intuition and magic that research can bring, they guide them through a step-by-step process of how to do research. That is not research. Instead of making one think and explore, they make you work section by section and follow a one-stop guide. That is not research. Instead of making one curious, uncertain, and willing to live the magical journey of research, they make you believe in methods and procedures. That is not research. Yes. We have massacred the meaning of research and the grace it once carried by attaching it to the number of publications. It has been passed on to newer generations, turning research into a timeline-driven, milestone-oriented, yet another course.

What is research then? No, it cannot be defined. Like the name itself says, it is searching yet again. That is all. We search for new meanings. We search for new ways to think. We search for new ways to see. We search for new ways to understand. We search for new ways to bring order to chaos, only for that order to reveal a new chaos waiting to be understood. Research is about living a passion that slowly makes you philosophical. Research is a journey where you get to decide when it has made an impact and when you are ready to stop. Research is a journey that cannot be explained in words. It has to be lived. Research is a freedom that one gets to choose, and not a prison where one is made to live.

Let a researcher decide how they want to organize the document. Let a researcher decide how they want to demonstrate the impact. Let research live a messy life to discover what works and what deserves another experiment. Let the supervisor become a philosopher again, not a gatekeeper. Let academia celebrate curiosity before it celebrates compliance. I hope the age of AI helps us regain what real research is. I hope AI frees researchers from rituals that have little to do with discovery. I hope AI throws away the meaningless paperwork that exists only to satisfy a process. I hope AI dismantles systems that reward compliance more than curiosity. I hope AI rebuilds the very essence of research, where questions matter more than templates and discovery matters more than documentation.

The same idea appears in any programs that we build. Consider a simple bank account application where customers are required to maintain a minimum balance of 1000/-. A customer may deposit money, withdraw money, or view the account balance. Although the account balance changes frequently, a rule must always be applied: the balance should never fall below 1000/-. Every operation in the program must preserve this rule. If a withdrawal would cause the balance to drop below 1000/-, the transaction must be rejected. This property remains true before, during, and after every operation. Such a property, which must always hold throughout the execution of a program, is called an invariant.

Invariants appear in all algorithms. Consider a program that searches for the largest number in a list. At every step, the variable max_so_far must store the largest value encountered up to that point. Similarly, when searching for the smallest number, min_so_far must always represent the smallest value seen so far. When calculating a total, sum_so_far must equal the sum of all processed elements. If the task is to count records, count must accurately reflect the number examined. Even when computing an average, the running sum and count must always correspond to the elements processed so far. In each case, the invariant acts as a trusted fact that remains true throughout the execution of the algorithm and guides it toward the correct result.

Invariants are important for AI-assisted programming and Vibe Coding. Human programmers often carry important rules in mind while designing a solution. An AI, however, focuses on generating code that appears to satisfy the requested functionality. Without clear constraints, it may produce solutions that work in some cases but violate critical business rules. Should we therefore include invariants carefully in our AI coding prompts? Consider an AI-powered e-commerce application. The AI may decide which products to recommend or how to personalize search results, but certain rules must always hold true. A customer should never be charged twice for the same order, and the payment collected must always match the order value. These invariants act as guardrails that help the AI generate safer solutions. If AI can generate thousands of lines of code in minutes, are we spending enough time defining the few rules that must never be broken?

There was a time when we searched for restaurant phone numbers, made phone calls to place orders, asked whether delivery was available, explained locations multiple times, and waited without knowing when the food would be delivered. It also came with many hurdles, like lines being busy, addresses being misunderstood, and there being no clear estimate of when the food would arrive.

A food delivery app as we have today absorbs most of these complexities. There is a clean menu, prices clearly listed, several food options to choose from, payments happening within seconds, and continuous delivery tracking. The person ordering food only thinks about getting the food delivered, and that is all. That is what abstraction does. An abstraction hides complexity and presents something simple enough for us to use a service comfortably. Most people using a delivery app never think about payment systems working, traffic prediction, GPS coordination, database management, or delivery optimization. The experience feels smooth because the difficult parts stay hidden from the user.

Food delivery systems today process thousands of orders simultaneously, identify efficient delivery routes, reduce communication mistakes, estimate preparation times, and coordinate deliveries across entire cities. A manual phone call may take several minutes before an order is even confirmed. The same process through an app happens within seconds. The applications have made the overall experience faster and more efficient and reduced several uncertainties that earlier existed as part of the process itself.

A similar shift is happening in programming today through Artificial Intelligence. AI is abstracting technical complexities and making content creation significantly simpler. A person can ask an AI system to summarize reports, generate ideas, explain concepts, organize information, or create content using ordinary language. Users don’t want to see the mathematical models, large-scale computing systems, or complex training processes underneath. Those layers remain hidden behind a simple interaction. For a long time, abstractions were associated with overhead. Earlier, we believed that hiding complexity would naturally make systems slower or less efficient. That is no longer true.

This idea is called negative overhead abstraction. It is slowly becoming one of the defining traits of modern technology. AI is helping build systems that absorb complexities so people can focus on thinking, creating, deciding, and living. These systems were never supposed to feel complicated to the end user. Simplicity was the destination all along. The responsibility of handling complexity was meant to stay with the engineer designing the architecture behind the scenes. And for the first time in a long while, it feels like we are finally moving in that direction.

Inductive thinking starts from small observations and moves toward a bigger understanding. We notice patterns from experiences and slowly form an idea or conclusion. Deductive thinking works the other way around. It begins with an idea or rule and applies it to specific situations to see if it makes sense. Good thinking happens when both work together, back and forth. Real understanding develops when people continuously move between experience and reasoning, testing ideas, questioning conclusions, and refining their thoughts through reflection and experience. I mean a mix of induction and deduction.

Imagine, what if the doctor always consulted a diagnostic system before observing the patient? What if the carpenter always searched for the next step before understanding the wood in front of him? What if the teacher always waited to see what experts had already said before responding to a student? Would their skill deepen with experience? Or would they eventually become dependent on an external intelligence? What does it mean to be a doctor, carpenter or a teacher? Would these crafts still stay distinct, or would they all slowly become the same act of looking elsewhere for answers?

The easiest thing for the mind is to accept the first reasonable answer it sees. Real thinking probably begins only when we resist that urge and stay with uncertainty a little longer. What if, instead we started reflecting more? There is a difference between using AI to think and using AI instead of thinking. Ironically, as machines become more intelligent, human reflection may become even more valuable. Thinking is not something humans naturally do well. It is something we slowly learn through practice, experience, mistakes, and repeated encounters with problems that matter to us. Somewhere between not knowing and knowing, the mind is forced to work, and that space matters the most. The deepest habits of thinking are built when something matters enough that we struggle with it, stay with it, and try to figure it out ourselves. What do you think happens to the human mind when we remove the very friction that once made thinking possible?

A doctor who has seen thousands of patients develops a certain kind of attention. A carpenter's hands know things before his mind articulates them. A teacher who has sat with struggling students learns to read confusion in the room. These aren't formal skills written in a manual. They are habits built through lived experience and reflection on that experience.

There is much to learn from the past but it is not clear on what to learn at. It feels like we ended up having more confusions than finding answers. Is it a progress? BIG yes. Did it happen every time when there was a paradigm shifting technology takeover? Maybe. But for now, we are all figuring it out. We all have our own theories. We have our own interpretations and inferences. Based on the usage we have developed our own mental models as well.

When mobile phones became common, people worried they would reduce face-to-face conversations and increase distraction. When television spread widely, concerns grew that it would reduce reading habits and critical thinking. When social media platforms became popular, people warned about addiction and shallow interactions. When the internet first emerged, many worried it would spread misinformation, reduce deep thinking, and weaken real-world social interactions. All our worries and concerns have occurred at varying scales and will continue to. Same applies to AI as well.

Keeping all that aside, the point of discussion has to be on how we can use AI for deeper thinking and long-term learning. Yes, it reduces effort by instantly generating answers, explanations, and solutions. There is, of course, a gap created here. While we rely on AI and offload our cognitive effort to it, can it instead become a cognitive amplifier that creates space for deeper thinking? Is AI a cognitive scaffold instead? Can we design interactions where thinking remains central? Can AI design different kinds of tasks based on the need, and guide each individual along their own path? Yes, capable and already doing, just like it happened with every other invention.

What we are calling today confusion is not coming from the tool. It is coming from the person who is interacting with it. Every new invention is first accused, then feared, and finally absorbed. Every new invention has changed the human nature and also revealed aspects of our thought process. Every time when something changes, we think the change is the problem. But often, it is just exposing how we were already thinking. The confusion, the dependency, the shortcut and as we name and label it, they were always there. Now they are more visible. So maybe AI is not changing thinking as much as it is forcing us to notice it. And then, like always, we slowly learn how to live with it. And of course make a better life with it.

Now, this is the core idea behind transformers that are used in tools like ChatGPT and Google Translate, and also form a key part of voice assistants such as Siri and Alexa, helping them understand language, generate responses, and power many of the AI systems we interact with every day. Here comes the most important question: is this the only way to construct a sentence? There is probability involved, but should it be the only process we rely on for everything? Is it the only way forward? Or can we step back and explore entirely different ways of thinking?

We often optimize what already exists without asking why it exists that way. Is that a problem? Yes. Is it something to worry about? Not always. A healthy system needs different kinds of thinkers. Not everyone has to wear a research hat. Some build, some refine, some question, some use and very few really research. That said, many of us skip the foundations. We don’t sit with the problem long enough to understand it. We explore, tweak, and add our cents, but that is not the same as solving. Replacing two layers with three is not research. Making something slightly faster or slightly better is not always insight. We rush to improve before we understand. We optimize outcomes without questioning the assumptions underneath.

Take ReLU as an example. It is a simple function that turns all negative values to zero and keeps positive values as they are. This helps models learn faster and avoid certain issues like vanishing gradients. Did we question why and how? Did we think of alternate ways of learning faster? Why do we need a learning rate? Are we really trying to understand machine learning or are we working with the statistics? It works well, so we use it everywhere. Why don’t we think whether something entirely different could do better?

Did you ever think of transformers from first principles? Did you ever question what and why for each layer that made it so? Let us return to the core question: do you accept what we call AI today? Not the definition we write down, but the way we have realized it and works in practice. I mean yes, it has solved a lot of our problems but can we use it to solve all of our problems?

AI can clearly impress us. Not a topic of discussion anymore. Can we stay longer enough to let the truth reveal and explore if that was the right approach to solve it? What does it actually mean to work with AI? Imagine a teacher who is not as smart, but has abundant of information. The teacher carries an abundance of knowledge, layers of concepts, connections, and insights built over years. When it comes to guiding someone through a problem, the teacher and student both feel something is missing. There is knowledge but not enough ways on how to use it. The teacher has the knowing-doing gap. We have the “Translation Void” here. This is our AI teacher. While the student learns, the teacher is also trying to learn.

When AI gives a solution, we read a well-structured response, maybe a few paragraphs or bullet points, figures, tables, and we assume it is correct. We copy it, reuse it, and sometimes even build conclusions on top of it. In all of that we stop questioning, we stop explorations, we stop reading, and we settle at surface missing the depth. Sometimes, we question back to validate and that validation is done by the same teacher who said it in the first case. We even gave this interaction a name as “prompt engineering.” As if crafting a slightly better sentence is engineering. No. It is not! It worked for a while, yes. It helped us get cleaner answers. But calling it engineering almost hides the deeper issue that we are still operating at the level of asking and receiving, not understanding. We have refined the input; we have no skill to challenge the output.

This is also true with all of us. When we don’t understand a concept, we often write more. We try to explain in different perspectives; we try to cover all possibilities. We fill space. When we understand something clearly, we write in precise terms. We write what is needed and we stop. LLM currently is the first kind of ‘us’. There is a rush of associations, patterns, and possible continuations flowing through them. And in that rush, they assemble answers that fit the context (yes, probability). The fluency hides the uncertainty. Depth requires judgment, restraint, and an understanding of what not to say.

So the “wow” is not a problem. But if we stop at the “wow,” then it definitely is.

Taking a photo involved film rolls. After all the photos were taken, we had to take the roll to a studio and wait a few days to see the results. Then digital cameras arrived and photos could be seen instantly. Now, smartphones capture pictures and also give us options to enhance them. Where are the film rolls and dark rooms today by the way? We hardly see the postman, but more delivery riders are on the roads. We don’t need to ask for directions anymore. We have GPS systems, but we still rely on asking strangers for directions rather than completely relying on technology.

The written letters once took days to deliver. Parcel delivery still takes days. Searching for information once required long exploration. Real research still needs long investigation. Learning a new skill once required finding the right teacher. True mastery still requires guidance and practice. Sending money once required visiting a bank. Trust in transactions still takes time to build. Booking a ticket for a journey once required standing in long queues. Even now, travel takes its own time.

Do you see all these things do not come with a single pattern? Or do they? AI is attempting to do that. When people speak about artificial intelligence, the conversation usually happens around algorithms, computing power, and data. Most are still trying to figure out what AI is and how it looks like. We all know it has benefits. These are undeniably important, and they form much of what we see on the surface. Along with faster computations, humans must decide the direction that AI action must move towards.

Everybody brings their own pool of experiences and perspectives to the table, and from that they attempt to place their predictions about where AI is heading. Cautious, optimistic, and speculative are the kinds of discussions we are having around AI. Individual predictions alone do not shape the direction of AI. The AI market is rather influenced and grounded by the needs and demands of society. As technology has aided in solving real problems of people and organizations, AI technology will continue to do the same. While opinions and forecasts create discussion, it is ultimately the demands of society, industries, and everyday users that determine where AI advances and where it finds its place.

Let us say we go to an AI assistant and ask to write a poem about rain. AI would instantly predict the next related words and create a poem. A conscious human would rather recollect all the lived experiences like the smell of the wet soil, the sound of rain on different surfaces, the calm and chaos that rain brings in, etc., and then pen down the experiences. We humans write from lived experiences. From a scientific perspective, consciousness refers to the brain processes that create awareness when a person recalls memories, senses emotions, and reflects on them. From a neuroscience perspective, the hard problem of consciousness is how this self-awareness produces the inner experience. From a philosophical perspective, this situation raises questions like: if an AI can produce a considerably good poem, does it actually experience rain, or is it simply processing patterns in the language?

Keeping all this in mind, we can further debate whether consciousness arises from complex information processing or whether it requires something more than computation. Humans writing the poem have basic consciousness, in which they are aware of the world and their senses, and higher consciousness, in which they reflect on memories and emotions while composing the poem, whereas an AI or a large language model generates text by analysing patterns in data without any inner awareness or lived experience.

Human and AI architectures differ primarily in their foundations. Human intelligence arises from a biological brain that integrates perception, memory, emotion, and lived experience, and AI relies on engineered models trained on large datasets for specific tasks. AI demonstrates that problem solving can occur without subjective experience. Yet higher abilities such as reflection, creativity, and moral reasoning may depend on awareness. Functionalist theories argue that if a system perceives, integrates information, reflects on itself, and adapts, consciousness may emerge regardless of whether it is biological or artificial. Theories like Integrated Information Theory and Global Workspace Theory propose that consciousness arises from deeply integrated information and globally shared internal processing, which current AI largely lacks. The Hard Problem of Consciousness points out that explaining perception and behaviour does not explain why there is a subjective inner experience. Similarly, the Chinese Room argument suggests that a system may process symbols and produce correct responses without understanding or experiencing anything.

If we can’t solve the Hard Problem, is there a “Consciousness Test” that goes deeper than the Turing Test? Can we create a conscious AI or a system that behaves like a conscious being? What do you think will work? All this leads to a deeper question: is consciousness necessary for intelligence?

If you said “no,” then you need to take some time to consider the questions below and rethink your answer. You can also still do it even if your answer was “yes.”

Have you ever relied on information that did sound reasonable and later turned out to be fictional? Do you sometimes create facts the way people create imaginary arguments in the shower? Have you ever answered like someone in love, like the way they are confident even when they are imagining things? Do you remember how you accept the “I read the terms and conditions” section?
How many times have you trusted and forwarded that WhatsApp forward?

Our brain is very good at filling gaps. And it fills even those gaps that may not be necessary to fill. We all have experiences of seeing, hearing, or feeling things that might not exist. These experiences can feel very real and can also impact our senses. We might have heard our phone ring when it did not, we might have felt the phone vibrate when it did not, or we might have replied to a text in our mind when in reality we did not, and so on. That’s just a ghostly version of the phone we have, and we can collect many such examples from the things around us.

Something to reflect on now is this: did this hallucination help in evolution? That little fancy of imagining things and falsely creating things, did it allow for creativity? When we don’t observe patterns, we create stories. That’s how we have tools, art, and ideas. The human mind going beyond reality is the reason we dream, innovate, and survive. For a beginner, the imagination that makes them worry a presentation might go badly is the same thing that helps them think ahead and prepare answers. People who imagine possible problems often end up doing better than those who assume everything will be perfect. The same quick mental shortcut that sometimes makes us misread a message tone is also what helps us catch social cues in meetings. Imagination is when your mind creatively thinks about possibilities. Hallucination is when those thoughts feel real without proof.

This also reminds us of something important about the systems we build and the knowledge we pass on. Any system that learns from human examples will naturally reflect human habits, patterns, and mistakes. If the input is incomplete, emotional, or biased, the output cannot magically become perfect or neutral. When we plant mango seeds, we only get mango trees. Information does carry assumptions, shortcuts and blind spots.

For the love of God, we hallucinate. That’s how we evolved. That’s how machines may evolve too. If imagination helped us grow, could it be the very thing that helps machines learn to grow as well?

When someone says a Thursday morning 9:00 a.m. show will probably get cancelled because there won’t be much of an audience, this is mostly coming from a past experience and not from statistics and historical reports. We don’t label this ‘data’, yet we take decisions. We carry certain book quotes with us for years, recall fragments of conversations long after they happened, and hold onto specific events and memories while many others disappear from our life. These memories guide our choices, even though they were never formally recorded or quantified.

Humans have a natural tendency to compare. We subconsciously ask what works, what works better, what feels better, what is more reliable, etc. We compare routes, career decisions, daily routines and we inherently do that. These comparisons are patterns. They form our belief systems. Just like most of us believe Mondays are unproductive. We do not arrive at this conclusion after studying productivity graphs or time logs. We arrive there through lived experience. Experience teaches us patterns, and we trust them without demanding numerical validation. Similarly, we avoid grocery shopping on weekends simply because we know it will be crowded, even when we have never counted the number of people inside.

This human ability is strengthened by structured memory. Now, what? This is a right time to ask the question: ‘What is a structured memory? Is structure universal?’ When experiences are reflected upon, grouped, and connected, decision-making becomes better. A professional who pauses after each work-let, thinking about what worked, what didn’t, and why, slowly builds an internal structure. Without analytics or reports, the work quality begins to improve. This structure isn’t found in numbers; it exists in reflection.

Uncertainty sits inaudibly at the center of all of this. We never have complete information. Even when we believe we are completely prepared, there are things that remain unknown. This is not a weakness of human reasoning; it is the condition of life itself. Every novel, every film, every story worth telling rests on this uncertainty and that is the beauty of life. If everything were predictable, there would be no tension, no curiosity, no becoming.

AI enters precisely here. It offers memory at scale, pattern detection beyond human reach, and comparisons made at speed. Used wisely, it can support reflection and sharpen awareness. But it cannot replace our ability to live with uncertainty, to value meaning over measurement, and to decide even when information is incomplete.

Life never waited for perfect data. The question is, now that we have machines that seek it endlessly,
well,
will you ask the next right question!??