Could GPT Be the Ultimate Teacher? We Think So

We’ve talked about why GPT is a tool that’s a vital part of Arcwise’s product. 

Today, we want to talk a little bit more about why it fits so neatly into our ethos. GPT is a teacher here to level you up. And it’s the latest iteration of a group project of shared, accessible knowledge pursued by humanity for the last 40,000 years, give or take. 

GPT isn’t going to make you an expert in spreadsheets or anything else overnight. If you’re like us, you’re probably wary of anyone who insists it will. (Just like we’re wary of any newly appointed experts in the physics of superconductors or deep ocean explorations or Eastern European geopolitics…or AI.) 

But we think its value lies in how it allows you to test assumptions in seconds, fail faster than ever, and free up your time to learn new skills.

Human Advice Technology is 40,000 Years Old, and We’re Just Getting Started

In this way, GPT is only the latest, most advanced tool in the pursuit of knowledge. And it’s a pursuit we’ve been chasing for a long time.

We’re not going to get too philosophical, but we will get archaeological. As far as we know, human interest in documenting and sharing information started with petroglyphs somewhere in the realm of 40,000 years ago. Over the next 35,000 years (and a few disasters later), it progressed to early forms of writing. When humans started writing around 3400 B.C., the world took off. 

Writing alone wouldn’t take us to where we needed to go; the early days of digital media are a modern example of just that. Because learning and advancement are not just about how we document information – they’re about how we compile and share it so everyone else can access it.

We began compiling libraries. The world’s oldest known library was compiled by Assyrian King Ashurbanipal in the 7th century B.C. Ashurbanipal was obsessed with learning and using knowledge to aid him in ruling the Assyrian empire. 

When we hit 100 B.C., technology really speeds up, relatively speaking. Forty thousand years after the earliest documented petroglyphs, we progressed from: the Library of Alexandria to printing books in fewer than a thousand years and to the Gutenberg printing press in only 600 more. From the industrial revolution to the digital revolution, we got the Dewey Decimal system so we could categorize and find all the books we’d been busy making, and then with the Internet there was a new big bang: modern internet publishing demanded a system for cataloging it and we got Google.

GPT is a Teacher, Not a Source

GPT isn’t an information source; it’s not a scroll or book or blog or forum. So in some ways, it doesn’t matter that it’s not that smart. Instead, like the Dewey Decimal system or an internet search engine, it connects you to those information sources.

So where does GPT really level us up and take its role as a teacher?

GPT is a teacher because it’s interactive. It goes beyond the library, the Dewey Decimal system, and internet search engines.

Rather than hunting down the source using an opaque system and then finding a way to answer your question, you ask GPT a question and see what answer you get in seconds. If the answer is unsatisfactory, you ask again, maybe using different parameters. Even better, unlike your major search engines, GPT isn’t bogged down by an opaque algorithm that millions of people around the world are dedicated to gaming.

It’s not the exact equivalent of putting your hand up and asking the teacher for help, but it’s a similar concept. You’re getting a single actionable output rather than a list of books or articles that a search engine thinks might be relevant based on an esoteric idea (i.e., Google’s search algorithms).

Ultimately, the faster access is making it easier to fail faster and learn faster, at least in the right conditions.

Speedy Access to Information

GPT makes it incredibly easy to get an answer to virtually any question, even unfortunately questions that may not have answers yet.

It’s important because finding and talking to an expert is slow. It can take weeks. If that expert has written a book or a guide, then you’re getting closer to the time you need. A digital asset is even better, as long as you know what you’re looking for.

But GPT is basically instant. Your questions are asked and answered in seconds. It’s not always right, but people aren’t always right so that’s why faster failure is why GPT is revolutionary.

Rapid Failure(s)

Faster access to information means rapid failure. And failure is everything.

If we go back to the expert route of information access, you can see how it’s a real time-suck. You send an email, wait for a reply, then send one or two more follow-up requests if you’re lucky. Ad infinitum. And that’s assuming you’ve found the right expert. Books have the same problem.

The internet isn’t bad, but you might need to dig through a few pages of search and then read someone’s life story to find precisely what you need.

GPT just answers your question in the here and now. And if it’s wrong, you can go find out right away. 

You didn’t spend three hours and ten search terms to find something that is out of date. You didn’t read a book published in 2008 to find out that the author recanted most of the lessons from that book in the 2019 edition.

ChatGPT isn’t perfect at this, but when you move GPT to spaces where you can instantly evaluate the results, you’re failing fast. Your spreadsheet gives you instant feedback on what's wrong and some breadcrumbs on how to fix it.

Focused Upskilling

Faster upskilling comes from the instant information access and the rapid failure. You learn what works and what doesn’t quickly. But with GPT and the applications it offers, you’re not limited to those two features when you’re applying it to a use case.

Whether you’re using GPT in your Google Sheets with Arcwise’s Co-Pilot or you’re using it directly in a tool like Code Interpreter, you get the added benefit of cutting out hours of busy work. When you can eliminate time trying to solve the problem and clean up the problem area, that’s when upskilling becomes more available.

Where We See GPT Going

We see GPT as a teacher, and like real teachers, the answers it provides can be wrong. It’s still up to you to evaluate the response to understand whether you have good information.

This gets so much easier when you’re using GPT in situ as we do in Google Sheets. 

When you put GPT in a space where you can immediately evaluate the results, you get far more from it. All the benefits, faster access, faster failure, and faster upskilling begin to kick in. 

What’s next? The problem with GPT as it stands is the answers can be wrong. The next era of GPT is turning it into an expert teacher rather than a student teacher. In short, it’s about making GPT wrong less often.

Everyone who uses GPT is giving it more sources and examples, and GPT4 is going to eliminate a lot of the accuracy issues.

We’re here to give GPT more sources and examples that we know are correct to help train the model. We can also get feedback when it’s wrong to correct future responses.