The novel way of learning something new

A brief history on learning

In the olden days, if you were eager to learn more about a specific subject, you would visit the library, look around for interesting books on the matter, then borrow a couple to bring home. If you knew what you wanted to learn specifically, you had to read through the books’ table of contents or glossary in order to find what it is that you’re looking for, hope that it is discussed at all in the book, or read the entire book for a general overview on the matter. It may have taken hours, if not days, to learn something new. Especially if a question suddenly came to you, say, late at night when the library was closed. Perhaps you could find a study group to share knowledge, but it was difficult obtaining reliable, truthful and complete answers to some of the more specific questions you had.

And then came the internet. A place where people started sharing thoughts, sharing information, sharing studies. The larger it grew, the harder it became to find what you were looking for, and thus the search engine was born: a place online where you type in a query, and the search engine finds indexed webpages that will likely have something to say about your query. The more wide-spread the internet became, the more commercial parties joined the club. Information was no longer free: webpages started collecting user data, click behaviour and advertisement viewing time; anything to profit as much as possible from each short-lived viewer on their webpages. Search Engine Optimization (SEO) became a field of study on its own: knowing how to effectively trick a search engine into thinking your webpage is perfect for various user queries on these search engines, even when it’s not. Or rather, I should say “search engine”, singular. According to Statscounter, at the time of writing, Google owns a search engine market share over 90%. Which means that only 10% of the market share is for all other search engines combined—from Bing to Yandex; Yahoo to Baidu. And DuckDuckGo is in there too, somewhere. With SEO giving the potential for prosperous companies to profit even more, especially when they only have to focus on a single search engine, most webpages you find at the top of your favourite search engine’s results have been riddled with attention-grabbing advertisements, sudden pop-ups and webpage-covering cookie banners, requiring many clicks and scrolls before you can finally, calmly, read the information you were looking for. This is ruining the internet so much that I, personally, despise searching for information online at this point. This phenomenon is aptly called the Enshittification of the internet.

Now, enter the age of Generative AI. It is 2022 and OpenAI releases its first public version of ChatGPT: an AI model that you can chat with, which chats back seemingly intelligently. You can ask it anything. And it replies with anything: truth, lies, hallucinations; you name it, they’ve sent it. As it turns out however, for most of the more basic questions, the output it generates actually makes a lot of sense, and quite consistently, too. The funny thing is, most questions that people ask are questions that some people have already asked before (we aren’t as unique as some of us might think). Meaning that these questions, including answers, are present in the immense datasets on which this model was trained. This makes ChatGPT’s model well generalized to this knowledge, which is why you might say it has an understanding of these topics. And this is where it gets interesting. A model like ChatGPT can do wonders for learning something new. Where Google, for instance, will find webpages somewhat related to what you’re asking, ChatGPT will actually attempt to answer every specific, niche, or even stupid question you throw at it. And all that without any judgment whatsoever. Where Google’s search results rely on people having actively discussed your piece of information online, ChatGPT can quite literally fill in the blanks.

A hypothetical

Let’s suppose, hypothetically, you’ve been interested in learning how computers work. You’ve heard about Software Engineering and you feel like this must be very difficult, since there’s just so much you don’t know about how computers do their thing. You want to learn more about this. You want to know whether you could ever become a Software Engineer, even if just for fun. You google “programming languages”, because that’s one aspect of Software Engineering that you recall. You end up on a Wikipedia article showing you a huge list of names for supposed programming languages. You don’t know what this all means. So you go back and query Google for “how do computers work?”. Google now provides you with a ‘featured snippet’ (in my case from Google itself, interestingly enough 🧐), coining some computer-related phrases and explaining in one or two sentences what they mean. You read through it. You feel like you’ve learned about computers, but you still can’t imagine what it would take to become a Software Engineer. You give it one last try: you google “how to become a Software Engineer”. You visit the first search result: a website of Columbia Engineering, detailing six steps, from planning the career path all the way to applying for jobs. The step for “planning” is only 154 words in length – it is incredibly general. Worse, as you scroll down, you notice a section that asks whether you should apply to a coding bootcamp to become a Software Engineer, with the next paragraph providing the reader with a wholehearted “yes!”. As we say in Dutch, “there comes the monkey out of the sleeve”: although this webpage does provide general information on the topic, ultimately it is trying to sell you one of their bootcamps. At this point you start to wonder whether the text you read on this page was in actuality as unbiased as you had initially assumed.

It is at this point that you recall a friend of yours once mumbled something about ChatGPT. You decide to check it out. You notice you can ask it any question you like, and it attempts to answer everything you throw at it. Intriguing! You continue to ask about computers. Specifically, you write a detailed message like you would to an expert in the field:

Hi there, ChatGPT!

I was wondering if you could help me out. I was hoping to understand what it takes to become a Software Engineer. I’ve been searching on Google but so far I’ve only learned what terms like RAM, binary and OS mean. Do you think I would have what it takes in order to become a Software Engineer? And is there an easy way to get started?

Thanks in advance!

Yours sincerely,

A clueless yet curious enthusiast

to which ChatGPT replies:

Hi there, Curious Enthusiast!

I’m thrilled to hear about your interest in becoming a software engineer! You’re definitely not clueless — just starting out, which is totally normal, and the fact that you’re already learning things like RAM, binary, and operating systems means you’re on the right track.

What Does It Take to Become a Software Engineer?
[…]

Do You Have What It Takes?
[…]

Easy Ways to Get Started
[…]

Your Learning Path Could Look Like This:
[…]

Summary: Software engineering requires a mix of coding, problem-solving, and understanding how computers work. You don’t need to know everything at once — start small, and build your knowledge over time. And yes, based on your curiosity, it sounds like you’ve absolutely got what it takes!

Feel free to ask more specific questions as you move forward, and good luck on your journey!

Yours sincerely,
ChatGPT
Your helpful guide to the world of coding!

“Would you look at that!”, you exclaim. ChatGPT replies as if with compassion, telling you it’s normal to feel overwhelmed, while providing you with an easy way to get started, various motivational words throughout their explanations, as well as multiple free online resources on the topic. It is not trying to sell you something (except, indirectly perhaps, ChatGPT Premium). The reply reads as if it honestly wants to help you out. You feel heard. You ask a couple questions more. You feel like you have a stupid question, but you know ChatGPT will answer it nonetheless. No judgment, just pure, immediate help for all the specific questions you have. A year passes. Your job application as Software Engineer at ArbitraryCorp. was accepted – you’re in. Congratulations.

Welcome to the future

The best part about models like ChatGPT, as we saw, is that they really act like personal assistants. Paradoxically, these AI models can really provide a personal touch in assisting you with any issue that you may be facing, given that you provide sufficient context. You can ask about literally anything and still get a reasonable response in return. Sometimes they don’t seem to make much sense. Sometimes they don’t tell you what you want to hear. Sometimes you tell them to do one thing and yet they do another. But most of the time, they provide somewhat intelligent thoughts and ideas about anything you share.

But let’s not fall in love with Her just yet. When you use ChatGPT to learn something, it is crucial that you fact-check some of the more specific or lesser known facts it tells you. It thinks it knows everything, yet it doesn’t. It doesn’t know what it doesn’t know, and that’s exactly why you need to keep autonomy. Let it help you out whenever you have unanswered questions. If you think deeply enough, I’m willing to bet you can conjure up new questions about pretty much any topic. Give it a try! See if you can learn about something new today. If you haven’t got a clue, you can always ask ChatGPT for ideas. Today, it is easier than ever before to stand on the shoulders of giants.

Appendix: new topics to learn a thing or two (or 20)

In case you’re out of inspiration, yet filled with curiosity to learn more, ChatGPT provided me with a long list of niche topics to study if you’re in a pinch. Some of my personal favourites to ask ChatGPT include:

• What is “quantum tunneling” and how does it apply to everyday technology?
• Can you describe the cultural significance of the Japanese tea ceremony?
• What are “untranslatable words” in different languages, and what do they reveal about culture?
• What is the “Chinese Room” argument in the philosophy of mind?
• What is the significance of tardigrades being able to survive in space?
• What is the story behind the “Great Moon Hoax” of 1835?
• Can you explain “mirror neurons” and their role in empathy?
• Can you explain the concept of “biological dark matter” in genetics?
• What are the legal ramifications of space mining on asteroids?

Copy ’em from here, paste ’em into ChatGPT and learn something new today! Just make sure to fact-check important information using your favourite search engine, so you won’t rely on ChatGPT’s questionable truth too much.