Go Sensible Science

Artificial Intelligence today is brilliant — it recommends your songs, drives your car, and writes your emails. But here’s the awkward truth: we still don’t fully know why it does what it does. AI systems can make decisions that are accurate, useful, even life-saving — but when you ask why a particular choice was made, you often hear silence. That silence is what engineers call the Black Box Problem . 🔒 The Black Box Problem Modern AI, especially deep learning models, are built from layers upon layers of mathematical transformations. They’re great at recognizing patterns — but their reasoning is buried under millions of parameters and neurons. So while the output makes sense (“brake now,” “approve loan,” “reject image”), the logic behind it remains hidden. It’s a bit like asking an artist, “Why did you use blue here?” and getting a shrug that says, “It just felt right.” Except in AI, that “feeling” comes from statistical weightings that not even the algorithm’s creator c...

There’s a quiet revolution underneath ChatGPT, Gemini, and every large language model you see today. It began with a simple but powerful paper from Google Research in 2017 titled “Attention Is All You Need.” Those five words reshaped artificial intelligence forever. Let’s explore what it really means — without going too deep into computer science, but deep enough to appreciate the elegance behind it. When Machines Read Like Humans (Almost) Before 2017, machines processed language like slow readers — one word at a time. They used models called Recurrent Neural Networks (RNNs) and LSTMs, which remembered past words as they moved along a sentence. It worked, but only up to a point. Imagine trying to understand this sentence: “The scientist who won the Nobel Prize in 1998 was from Sweden.” By the time the model reached “Sweden,” it might forget “scientist.” So it lost the relationship between who did what. That was the problem: AI could see the words but not hold them together in one coher...

Artificial Intelligence is no longer the future — it’s here, and it's reshaping how we think, work, and build. But for many people, especially those without a background in coding, AI can feel intimidating. Here's the good news: you don’t need to be a software developer to use AI tools like ChatGPT. In fact, if you understand problems and have ideas — AI can be your most powerful partner. LLMs: The Mind That Has Read Everything Imagine this: you’ve studied 10 books on a topic. Your friend has studied 30. Clearly, your friend might know a bit more. Now imagine a model that has read millions of books, research papers, and internet pages across every field imaginable — from quantum mechanics to philosophy to architecture to car repair manuals. That’s what a large language model (LLM) like ChatGPT has been trained on. This is why it can answer questions, generate code, write summaries, translate languages, simulate conversations, and even explain tough engineeri...

Language has always been more than just a medium of communication. It is a carrier of identity, access, and — most importantly — power. When we look at how power is distributed globally, it's easy to forget how central language is to this equation. The influence of a language often parallels the economic dominance of its speakers. English, for instance, owes much of its global status not just to colonial legacy, but to the economic and technological supremacy of the US and UK. But this linguistic power has long created inequality in unexpected ways — especially in countries like India, where language often acts as an invisible filter, separating the privileged from the marginalized. Let me illustrate this with something I observed firsthand. In Kolkata, one of my school teachers came from a tribal background. His knowledge was deep, and if you spoke to him, you'd instantly sense his insight and compassion. But his English wasn’t fluent — a limitation that often over...

As engineers, we take pride in solving problems. We optimize systems, debug code, design robust architectures, and think in terms of precision and logic. But there’s one skill that quietly makes or breaks everything — a skill often underestimated in technical circles: communication . Yes, I’m talking about the good old art of expressing ideas clearly. This blog is a reflection on how communication – especially in the age of AI and prompt engineering – is becoming a non-negotiable skill, and how my own experience as an engineer (and a former theatre student) shaped this realization. The Engineering Mindset: Facts First, Communication Later? Engineering teaches us to focus on accuracy, efficiency, and function. We’re trained to get things right . But when it comes to sharing what’s in our mind — whether in design discussions, stakeholder meetings, or team emails — we often falter. Why? Because we tend to think logic alone should be enough . But here’s the truth: if o...

Artificial Intelligence (AI) is transforming the automotive industry, from autonomous driving and predictive maintenance to smarter manufacturing and personalized in-car experiences. However, beyond the technical advancements, AI is also shaping the social, geopolitical, and professional landscape of automotive engineering. In this evolving world, one truth stands out: learning new things is not just an advantage—it’s a necessity . AI has democratized knowledge, making cutting-edge information accessible to anyone willing to learn. But this also means that as engineers, we must actively adapt and grow . Otherwise, in 5-6 years, we may find ourselves outdated, unable to solve the modern problems of the automotive world. 1. The Global AI Race in the Automotive Industry The automotive industry is no longer just about making better cars—it’s about winning a technological race. Countries like the U.S., China, and Germany are investing heavily in AI-driven transportation, and...

Artificial Intelligence (AI) has revolutionized the way we live, work, and interact with technology. From healthcare to finance, AI has brought unprecedented advancements, making processes faster, more efficient, and often more accurate. However, like any powerful tool, AI has a dark side. Misinformation, deepfakes, and other malicious applications of AI are emerging as significant threats to society. As engineers and technologists, it is our responsibility to understand these challenges, mitigate their risks, and ensure that AI is used ethically and responsibly. The Rise of Misinformation and Deepfakes Misinformation and deepfakes are two of the most concerning byproducts of AI's rapid development. Misinformation refers to the spread of false or misleading information, often amplified by AI-driven algorithms on social media platforms. Deepfakes, on the other hand, are synthetic media generated using AI, where a person's image, voice, or actions are manipulated to c...

As an engineer, I’ve always been fascinated by how technology can solve real-world problems. Artificial Intelligence (AI) is no exception—it’s not just a buzzword; it’s a toolkit that can transform lives. But here’s the catch: while developed nations are racing ahead with AI, developing countries are often left playing catch-up. The question isn’t just about building fancy algorithms; it’s about making AI work for everyone, everywhere. And as engineers, we have a unique role to play in this mission.   Let’s talk about how we can bridge the gap, why AI literacy matters, and what it takes to build a future where technology is truly democratic.   Why Engineers Should Care About AI in Developing Countries When I think about AI, I don’t just think about self-driving cars or chatbots. I think about the farmer in Kenya who needs better crop predictions, the nurse in rural India who could use AI to diagnose diseases, or the small business owner in Brazil who need...

Artificial Intelligence (AI) has made remarkable strides in recent years, with Large Language Models (LLMs) like GPT-4 pushing the boundaries of what machines can achieve. However, as AI systems grow more complex, they also become more resource-intensive, raising concerns about scalability, energy consumption, and computational limits. Enter quantum computing—a revolutionary technology that promises to redefine the future of AI. But what exactly is quantum computing, and how could it transform the AI landscape?   What is Quantum Computing? Quantum computing leverages the principles of quantum mechanics to process information in ways that classical computers cannot. While classical computers use bits (0s and 1s) as the smallest unit of data, quantum computers use quantum bits, or qubits. Qubits can exist in multiple states simultaneously, thanks to a phenomenon called superposition. Additionally, qubits can be entangled, meaning the state of one qubit is intrinsically li...

Artificial Intelligence (AI) has become one of the most transformative technologies of the 21st century, revolutionizing industries from healthcare to finance. However, as AI models grow in complexity and capability, so does their demand for computational power. This "power hunger" has sparked debates about the environmental and economic costs of AI development. Yet, as NVIDIA CEO Jensen Huang pointed out in a recent podcast, the energy consumed by AI models is not just a cost—it’s an investment. This investment, he argues, accelerates research and drives the discovery of practical energy-saving solutions across various domains. Let’s dive deeper into this topic and explore the multifaceted implications of AI’s power consumption. The Power Hunger of AI Models AI models, particularly large language models (LLMs) like GPT-4, require massive amounts of computational resources for training. Training these models involves processing billions of parameters across vast d...

Artificial Intelligence (AI) has become a cornerstone of modern technology, powering everything from virtual assistants to autonomous vehicles. However, behind the scenes, developing and deploying AI models is an expensive endeavor. Whether it's training a Large Language Model (LLM) like ChatGPT or building a custom AI solution, the costs can run into millions of dollars. In this blog post, we’ll explore the reasons why AI models are so costly, the factors driving these expenses, and the perspectives of engineers who work on these cutting-edge technologies. The High Cost of AI Models: Key Factors The cost of AI models can be attributed to several factors, ranging from data collection to computational resources. Let’s break down the major contributors: 1. Data Collection and Preparation     Data Acquisition: High-quality datasets are essential for training AI models. Acquiring large, diverse, and labeled datasets can be expensive, especially for niche domains.     Dat...

In recent years, the field of artificial intelligence (AI) has witnessed groundbreaking advancements, particularly in the domain of natural language processing (NLP). Among these advancements, Large Language Models (LLMs) like OpenAI's ChatGPT have emerged as transformative technologies, reshaping how we interact with the internet, process information, and even think about creativity. This blog post delves into the technical underpinnings of LLMs, their impact on the digital landscape, and the perspectives of engineers who work with these models. What Are Large Language Models (LLMs)? Large Language Models are a class of AI models designed to understand, generate, and manipulate human language. These models are trained on vast amounts of text data, enabling them to learn the statistical patterns, grammar, and semantics of language. The "large" in LLM refers to the enormous number of parameters—often in the billions—that these models use to capture the complexi...