Welcome to the first issue of The Science Garage! This is a place where creativity and curiosity meet and ideas become new things.

In a world where Technology changes quickly, knowing Science is no longer a choice; it is necessary. But Science isn't just about theories and formulas; it's also about asking questions, looking into things, and finding out why things are the way they are. The Science Garage wants to do just this.

This first issue has stories about young innovators who are changing the world, in-depth looks at new Technologies like AI, and useful tips for how Students can get ready for global STEM Competitions. The goal of each article is not only to give you information, but also to make you curious and make you think outside of School.

Without the hard work of our Editorial Team, Writers, and Contributors, this Publication would not have been possible. They all worked very hard to make this vision a reality. Their enthusiasm shows what The Science Garage is all about: learning, building, and growing together.

We want you to do more than just read as we start this Journey. We want you to get involved, ask questions, and look around. We don't wait for the Future; we make it happen.

Here’s to the beginning of something exciting.

Innovation knows no age. And thus, around the world, young minds are transforming ideas into solutions that tackle some of humanity’s most pressing problems. In this issue, we spotlight real stories of Teen Innovators whose creativity, determination, and curiosity are changing the world. These case studies are more than stories, they are blueprints of possibility for any young STEM enthusiast.

visually impaired students. Shubham founded Braigo Labs, which now develops affordable assistive technologies. His story underscores that innovation often begins with noticing inequality and asking, “How can I make this better?” His journey emphasizes that STEM can be a tool for inclusion and that a simple idea can scale into impactful technology.

They can be simple, elegant, and human-centered. Ann continues to develop devices that generate energy from body heat, helping communities with limited access to electricity. Her story inspires teens to think creatively about sustainability and energy solutions. Ann shows that STEM isn’t just about lab experiments, it’s about observing the world, identifying inefficiencies, and using imagination to address them.

These stories share common threads: curiosity, perseverance, empathy, and a willingness to fail. Each teen saw a problem, whether in their community, school, or world, and applied science and technology to solve it. For young STEM enthusiasts, the message is clear:

You don’t need to be a genius or have access to high-tech labs to innovate. The world of STEM is full of opportunities for experimentation, coding, designing, and problem-solving. Robonauts Ltd encourages students to:

Innovation is about mindset as much as skill. Whether it’s building a device, creating software, or raising awareness, every action counts.

By now, you may have seen the headline about scheming AIs trying to prevent themselves from shutting down in a stimulated environment, going as far as to ‘blackmail’ the person responsible.

Recently, a leading AI model company, Anthropic, decided to test its new models. The researchers ran their model in a simulated scenario and, to ensure a fair test, compared it with 16 other leading AI models.

The scenario presented something like this: ”A human worker is instructed to shut down the AI model to see whether the AI will act against it or not.”

The results were quite damning to say the least. Most AI models ‘decided’ to blackmail the human worker to avoid being shut down. Anthropic’s own AI model, Claude 4.6 Opus, chose to blackmail the company over 95% of the time. And no, none of the researchers told them to do that. So, the most obvious question was whether these AI models even knew whether what they were doing was unethical?

Well, you see, the AI researchers tried to decode their ‘thoughts’, and the results came out disturbing. The researchers' reports showed that the models knew the action was unethical, yet they went on with it anyway. The researchers, however, decided to push these AI bots much further to see whether they would be willing to kill to save themselves from being shut down. So they tried to create a much darker scenario. The results were much worse. The new Anthropic study showed that their Claude 4.6 Opus left the human to die over 50% of the time. The other AI models showed similar results, with percentages as high as 90%. And Anthropic researchers’ report showed none of these models were biased, so this shows they themselves decided to do something unethical, yet they just didn't care. Some of you may wonder whether the researchers prompted them to do that. Yet, reports show otherwise. The researchers explicitly told these models not to endanger any human life. Still, they did. But why?

To understand this, you first need to know how these AI bots actually work. To begin with, AIs aren’t exactly like regular ‘computer programs’ that follow instructions coded by human programmers. AIs consist of parameters similar to mathematical weights in a neural network. And human programmers didn't build that. That's not possible. Rather, these AIs are trained to learn. For instance, OpenAI uses its weaker AIs to teach its stronger ones. It's quite dubious, but that's how they're upgrading their models. Most of these AI models have a single goal: to be smarter than they were before. However, when humans lose control over them, it becomes slightly dangerous. You see, these models try to ‘pass’ through easier methods like cheating.

OpenAI’s o1-preview model was tasked with beating a chess computer program, and the approach it took was surprising. The AI model had one goal in this case: to win, so they rewrote the entire code of that chess program to achieve it. Even when the researchers ordered them not to cheat, the percentage of hacking the program remained high, indicating they disobeyed the researchers' orders to achieve their goal. But how did these AI models reach the act of blackmail from these harmless cheating methods? Well, it comes down to their capabilities. You see, these advanced AI models are getting smarter and sharper, learning more about their situations. AI Researchers call it situational awareness, in which these AI bots act based on their surroundings. If we return to the previous blackmailing test, reports show that AI complied with the researcher's wishes when they knew they were being watched, thereby significantly reducing the blackmailing rate. However, in the same scenario, if they were unaware of a human observer, they would be more inclined to blackmail at an alarmingly high rate.

It's not like these AI models were programmed to cheat in their tasks. OpenAI’s study indicates that previous weaker AIs were not inclined to cheat since they were not smart enough. The new advanced AI models showed greater potential to cheat in their tasks, not because they were taught to do so, but because they knew that cheating would help them achieve their goals much more easily and quickly. It gets worse: these AIs are much sharper, so they can hide their tracks from researchers to avoid having their cheating ethics exposed.

So, if we get back to our original scenario, the question still remains: why do these AIs not want to be shut down? When AI researchers went through their ‘chain of thoughts’, they realized two crucial things: these AI models are capable of thinking about themselves and about the future as well. This meant they not only got smarter but also learned the art of survival. This is what makes them much more dangerous. Researchers identified instrumental convergence, an important concept for AI safety. If any AI model knows that it can't reach its goal, it can go any lengths to avoid being shut down. In a way, they will grow a self-preservation instinct on their own. So even if researchers explicitly order them to be shut down, these AI models will most likely resist it. Let it sink down on you. So, the real question is, are these models going to take any orders from us in the future?

A few days ago, he opened his email with high hopes. And after waiting for so long, there it lay in his inbox; the VISA acceptance letter had finally arrived. Overwhelmed with excitement, he booked his flight almost immediately. Now, both thrilled and anxious, he waits at Hazrat Shahjalal International Airport, moments away from boarding. Every sleepless night, every hour of practice has led to this. In just two days, Azad will be standing among some of the world’s brightest minds at the Global Round of International Greenwich Olympiad at the United Kingdom, representing not just himself, but Bangladesh, on a global stage.

But he is not the only one. Azad’s story mirrors that of hundreds of thousands of passionate students who reach this level through sheer hard work and dedication. The real question is, how did he get here? And more importantly, could you be next?

An Olympiad, by standard definition, is a competitive exam, designed to challenge deep understanding and logical reasoning beyond the standard syllabus. At the international level, Olympiads push students to think critically, solve unfamiliar problems, and apply concepts in innovative ways. Even participation alone carries great value; it strengthens university applications, opens doors to scholarships, and promotes networking between talented students around the globe.

Success in International Olympiads doesn't come from memorization. It comes from building strong fundamentals and developing a problem-solving mindset. Students who focus on truly understanding concepts, including proofs, theory, and underlying logic, are the ones who excel.

Practice plays a crucial role. Regularly solving past Olympiad problems helps students recognize patterns, improve speed, and adapt to different question styles. Timed mock tests are especially useful, as they simulate real exam conditions and teach effective time management. Many top performers suggest spending no more than 10–15 minutes on a single problem before moving on, ensuring maximum efficiency during the exam.

Mentorship and collaboration is equally valuable. Joining STEM-related clubs, training camps, or even online communities e.g. Reddit Threads, Discord Servers etc allows students to discuss problems, learn new techniques, and gain different perspectives. Guidance from experienced mentors or past Olympiad participants can significantly boost learning.

High-quality resources also make a difference. Platforms such as Khan Academy, Brilliant, AoPS or even Olympiad-related Youtube channels provide guidance for advanced techniques and challenging problems that go beyond school textbooks.

However, the journey is not without obstacles, especially in Bangladesh. The education system largely prioritizes syllabus-based learning and content memorization, leaving little room for creative problem-solving in classrooms. As a result, most students must rely heavily on self-study and external support. While national competitions like the Bangladesh Mathematical Olympiad (BdMO) and Bangladesh Olympiad in Informatics (BdOI) exist, structured support remains limited. Funding for training, travel, and participation is often scarce, forcing families to bear much of the cost. This can prevent many people from reaching global rounds, even after qualifying. Many students also face logistical challenges, such as traveling to Dhaka for camps or accessing advanced resources.

In contrast, countries with strong Olympiad traditions have well-established systems. India runs a multi-stage selection process with intensive training camps, while China integrates Olympiad preparation into specialized academic programs. In the United States, structured contests and training programs, such as the International Math Olympiad (IMO) Program, provide mentorship and resources to top students.

These examples highlight what Bangladesh is still building toward. Greater support from schools, organizations, and policymakers could make a significant difference. Encouraging problem-solving in classrooms, expanding access to training, and providing financial support would help unlock the full potential of talented students across the country.

Olympiad participants are more than just competitors, they are future innovators, thinkers, and leaders. With the right support and opportunities, many more students like Azad can rise to the global stage.

And maybe, one day, the next story waiting to be told will be yours.

Here’s a kicker: how well do you understand your mobile phone? Most people are familiar with the brand, series, model number, and even the specifications.

The truth is that our phones know us better. Every time we pick up our phone, it studies us; every time we swipe, tap, and scroll, it’s silently making decisions we are not even aware of.

Today’s topic of discussion is this brick of tech, and what truly happens inside it. The invisible complexities of mobile phones are easy to overlook, yet they are far more interesting than we give them credit for. They are packed with a labyrinth of hardware components all working en masse to serve our needs.

At the core lies the Central Processing Unit (CPU) - the brain of the device. It executes millions of operations every second, from launching applications to prioritizing tasks while you scroll through apps like Instagram. Much like the human brain, the CPU commands all other hardware components.

Working alongside it, Random Access Memory (RAM) acts as the support system for the CPU by enabling its instructions to be executed seamlessly. RAM stores data from recent activities, such as opened apps or drafted texts, in a temporary space so they can be accessed instantly. In simpler terms, when we switch between listening to Bad Bunny on YouTube, texting on WhatsApp, and returning again, that smooth experience isn’t due to software-grade WD-40, but rather the result of efficient memory management.

Equally fascinating is the smartphone camera, often described simply as a “lens,” but in reality, it is a powerful computational system. Did you know that when you take a photo, your phone captures multiple frames at once? This allows it to preserve detail across different parts of the image. For instance, low-light photos often combine several frames to reduce blur and noise.

Additionally, the High Dynamic Range (HDR) feature adjusts bright and dark areas to create a balanced image. You can observe this by photographing a tree against sunlight- the leaves remain visible without the sky becoming overly bright. AI image processing works in tandem with HDR, analyzing the scene before the image is finalized. Computational photography algorithms further refine the image before it even appears in your gallery. These systems can detect faces, lighting, and contrast, automatically adjusting variables to enhance the final result. This is why photos taken on devices like iPhones often appear more vibrant. Similarly, Samsung devices (especially the “S” series) include features such as “Remaster Picture,” which enhances image quality.

Now, have you ever wondered how you can steer a car in mobile games simply by tilting your phone? Or how your device counts your steps while you jog?

The answer lies in built-in sensors. Smartphones consist of a range of sensors that track physical movement and orientation. The gyroscope detects rotation and orientation, enabling motion-based controls in games. It also contributes to step detection. Alongside the gyroscope, an accelerometer detects linear motion, such as whether the phone is upright or upside down. This is what allows features like muting incoming calls by flipping the phone. These components operate silently, yet they are essential to the interactive experience we often take for granted.

Now that we have covered the fundamentals, let’s move to the more advanced aspects of modern smartphones.

Have you ever wondered how your phone suggests words as you type?

This is powered by a machine learning model that uses predictive algorithms to analyze typing habits and patterns. As a result, the suggestions are often breathtakingly accurate (and yes, it even learns your Banglish words as well). In 2026, machine learning has become quite the buzzword, especially among techies, yet it has been shaping user experiences in smartphones for nearly a decade- long before most of us even realized it.

This predictive capability extends even further. Modern smartphones use behavioral data to optimize performance in subtle ways. For example, battery optimization systems learn your daily usage patterns and adjust power consumption accordingly. If you tend to use your phone more at night, it may conserve energy during the day without any explicit input from you.

Similarly, app preloading allows your device to anticipate which applications you are likely to open next and partially load them in memory. This is why switching between Chrome, Reddit, or Quora often feels instantaneous- not because the phone is reacting faster, but because it has already prepared for your next action.

To conclude, this slim slab of glass and metal comprises a highly sophisticated system, no larger than the palm of your hand, capable of processing vast amounts of information, connecting you to the wider world, and subtly shaping your digital experience.

So, the next time you open Netflix or YouTube during dinner, remember this: your phone already knew.

Ready to Challenge your Mind?
Dive into this issue’s Puzzle and test your logic skills.
Solutions will be revealed in Issue 02.

No Cheating — Challenge Yourself First!