Radio waves, blue sky, touch screens: How LLMs can answer questions about the world around us
I have always been very curious about the world around me. I like to think that as a kid, one of the reasons I was talking so much is that there were so many things to understand.
Even today, one of the best decisions I made in 2025 was to learn something new every day for 15 minutes, using technical textbooks and articles. As a side note, this has been the source of inspiration for many of my blog posts this year.
After the release of DeepSeek, I realised that LLMs had now become much smarter than I was in many areas. They could write better, code better, and sometimes, even “think” better. What I mean by “think” here is: solving tough problems with step-by-step reasoning.
The idea
This made me think. How could this help me learn new things about the world around me? As a Computer Scientist, I love experiments. I decided to give this a try for 30 days. Every day, I would ask a question to DeepSeek or GPT-4 about something about the world around me.
I first asked it to generate some questions to give me some inspiration. I made sure to ask for questions on various topics such as Physics, Chemistry, Biology, Computer Science, Engineering, etc…
Before we go further, despite the average nerdiness of my usual articles, I never formally studied Sciences. I graduated from High School with Economics and Maths, studying Economics, Political Science and Maths at university. I only later moved to Computer Science. These basic questions are far outside of my circle of competence. This is why this exercise was so valuable.
Some example questions include:
- How do touch screens work? (see example conversation below)
- Why do compounds dissolve so well in water?
- How do radio waves transmit data?
- What is electricity? And how does it arrive at the power outlets in your flat?
- How does a battery work?
- How do phones know which network to connect to?
- What is the difference between wifi and cellular data?
In the following section, I will describe my experience and include some useful tips to try this for yourself.
Putting it into practice
Generating an initial list of questions
With a basic prompt, ask any Large Language Model to generate a list of questions on the world around you. To generate more questions, you can select the ones you like and ask the model to generate more similar ones. Once you are satisfied, consolidate this list in a document, and start one question at a time.
I generally instruct models to answer these questions:
- Assuming no previous knowledge
- Using simple words and examples
You can experiment with prompts of your choosing such as:
- “Explain this to an undergraduate student” or any other level
- Add simple numerical examples, if you are into that
- Etc
Leave no stones unturned
If the explanation contains something you do not understand, do not panic, you just found another question.
This is what happened when I was reading an LLM’s answer to the question: “What is the difference between wifi and cellular data?”. It kept talking about how data was transmitted using electromagnetic waves of different frequencies. I am able to understand what a wave and a frequency is, but I still could not picture exactly how data was transmitted. This led me to the question on how waves can transmit data; through which I finally understood the acronyms AM (amplitude modulation) and FM (frequency modulation), on the radio on the breakfast table in the morning.
I can already hear what some readers may be thinking: how can I finish my list of questions if I keep adding new questions? This is the beauty of it, you do not have to finish any list. There is always something new to learn in the world we live in.
Noticing connections
As the days went by, I noticed that I was getting better at answering, or getting close to the answer of the questions on my list, before asking the LLM for an explanation. The more you understand (even at a superficial level) about the world around you, the easier it becomes to imagine how nature and machines may work.
This is what truly excited me and motivated me to write this blog post to share my experience.
To get the most out of it, I would recommend starting to think about a question in the morning, and only checking the answer in the evening. The more you practise thinking about problems you do not know the answer to, the easier it will become to solve new ones.
Final Thoughts
What I find crazy about this whole experience is that none of it required the technological leap of Large Language Models. One could have done this using a Google search or a public library.
The release of powerful Large Language Models obliterated all remaining frictions of acquiring new knowledge. Before, we had to go to a library, or do a Google search and go through a few sources. With LLMs, you simply need enough motivation and energy to type (or speak) a question and read (or listen to) its reply.
If you would like to start today, I would recommend you pick one of the questions below - or any question that you have always wanted to know the answer to - think about it for the day, and ask a good LLM for an explanation this evening. I am sure you will be amazed at the result of doing this for 30 days. Wishing you all the best in our learning journey!
Example questions:
- Why do compounds dissolve so well in water?
- How do touch screens work?
- How do radio waves transmit data?
- What is electricity? And how does it arrive at the power outlets in your flat?
- How does a battery work?
- How do phones know which network to connect to?
- What is the difference between wifi and cellular data?
- Why is the sky blue?
- How do aeroplanes stay in the air?
- How do magnets work?
- Why do we see lightning before we hear thunder?
- How do microwaves heat food?
- How do glasses correct vision?
- Why do leaves change colour in autumn?
- How does sunscreen protect your skin?
- How do vaccines work?
- Why does metal feel colder than wood at the same temperature?
- How do submarines dive and surface?
- How does GPS determine your location?
Example conversation
User: How does the screen detect your touch? Is it the same for all types of screens (e.g., LCD vs. OLED)?
Deepseek: Great question! Touchscreens are a marvel of modern technology, and they work in different ways depending on the type of screen and the underlying technology. Let’s break it down step by step, including how touch detection works and whether it’s the same for LCD and OLED screens.
How Does a Touchscreen Detect Your Touch?
There are two main types of touchscreen technologies used in smartphones and other devices: 1. Resistive Touchscreens 2. Capacitive Touchscreens
Most modern devices, including smartphones and tablets, use capacitive touchscreens, so we’ll focus on those. However, I’ll briefly explain resistive touchscreens for context.
1. Resistive Touchscreens
- How it works:
- A resistive touchscreen is made of two layers: a flexible top layer (usually plastic) and a rigid bottom layer (usually glass), separated by a small gap.
- When you press the screen, the two layers touch at the point of contact, completing a circuit.
- The device measures the change in electrical current to determine the touch location.
- Pros:
- Can be used with any object (finger, stylus, glove, etc.).
- Cheaper to produce.
- Cons:
- Less sensitive and responsive than capacitive touchscreens.
- Poor visibility in bright light (due to multiple layers).
- Where it’s used: Older devices, some ATMs, and industrial equipment.
2. Capacitive Touchscreens
Capacitive touchscreens are the standard in modern smartphones, tablets, and other consumer electronics. Here’s how they work:
How It Detects Touch:
- Conductive Layer:
- The screen is coated with a transparent conductive material, usually indium tin oxide (ITO), which forms a grid of electrodes.
- Electric Field:
- A small electric charge is applied to the grid, creating an electrostatic field across the screen.
- Touch Detection:
- When you touch the screen with your finger (which is conductive), it disrupts the electrostatic field at that point.
- The device measures the change in capacitance (the ability to store an electric charge) at the touch location.
- Location Calculation:
- The device’s controller calculates the exact coordinates of the touch based on the changes in the grid.
Multi-Touch:
- Capacitive touchscreens can detect multiple touches simultaneously, enabling gestures like pinch-to-zoom or two-finger scrolling.
- This is done by tracking changes in capacitance at multiple points on the grid.
Pros:
- Highly sensitive and responsive.
- Excellent visibility and clarity (since there’s only one layer).
- Supports multi-touch gestures.
Cons:
- Requires a conductive input (like a finger or special stylus). Gloved hands or non-conductive objects won’t work.
- More expensive to produce than resistive touchscreens.
Is It the Same for LCD and OLED Screens?
The touch detection mechanism (capacitive or resistive) is independent of the display technology (LCD or OLED). However, there are some differences in how touchscreens are integrated with these display types:
LCD (Liquid Crystal Display):
- LCD screens use a backlight to illuminate the display.
- The touch layer (capacitive or resistive) is typically placed on top of the LCD panel.
- In some designs, the touch layer is integrated into the display itself (called in-cell or on-cell technology), making the screen thinner and more responsive.
OLED (Organic Light-Emitting Diode):
- OLED screens don’t require a backlight because each pixel emits its own light.
- The touch layer is usually integrated into the display (in-cell or on-cell), similar to modern LCDs.
- OLED screens are thinner and more flexible, which allows for better integration of touch technology.
Advanced Touchscreen Features
- Force Touch (3D Touch):
- Some devices (like older iPhones) can detect how hard you press the screen by measuring the area of contact or using pressure-sensitive layers.
- Haptic Feedback:
- Devices use tiny vibrations to simulate the feeling of pressing a button or interacting with the screen.
- Palm Rejection:
- Modern touchscreens can distinguish between intentional touches and accidental contact (like your palm resting on the screen).
Summary
- Capacitive touchscreens are the most common in modern devices and detect touch by measuring changes in capacitance.
- Resistive touchscreens are older technology and detect touch by physical pressure.
- The touch detection mechanism is the same for LCD and OLED screens, but the integration of the touch layer may differ.
- Advanced features like force touch and haptic feedback enhance the user experience.
Let me know if you’d like to explore any part of this in more detail! 😊
PS from Eliott: This led me to two other questions:
- Why is a finger conductive?
- In both cases, how exactly is the location of touch determined?
Let’s not be afraid of asking the dumb questions.