
You’ve likely heard the term ChatGPT by now. Since its launch, this past November 30th, much has been said and written about ChatGPT and its implications on the way we live, work and learn… Sometimes even by the Generative Artificial Intelligence chatbot itself.
Within five days of its launch, more than one million people had signed up to test ChatGPT, described by one New York Times article as a “highly-capable linguistic superbrain” – a head-spinning “mix of software and sorcery.”
Coming from a digital media background, and interest in the way emerging technologies apply to teaching and learning as EdTech, I was curious what the implications were of ChatGPT on education in the not too distant future – neigh…now. (Because these impacts are coming whether or not we are ready for them and have already started to massively shift what’s taking place in classrooms, at homes, and on campuses in Canada and abroad.)
What We Know About ChatGPT
ChatGPT, which stands for Chat Generative Pre-trained Transformer, is based on pre-existing Large Language Model technology called GPT3 that essentially provides exceptionally-humanlike, cogent and original text responses based on what prompts you give it. But while this sounds simple, what the tool can do is much more complex than that; It can write academic essays, summaries, poems, raps, memes, and jokes (that are actually funny). It can answer trivia, translate languages, produce course syllabuses, generate original recipes (with moderate success), spit out functional if imperfect computer code or drive interactive text-adventure games, mimic pop psychologists and celebrities, produce plausible cover letters and government bills, provide inspiration to creatives, and much more, giving even its makers pause. With ChatGPT, the more specific and detailed the prompts, the more fleshed out and human-like the results.
Some even say that this invention will render search engines like Google obsolete, and put many out of a job (some say even teachers), across various creative fields – challenging our perception of just how replaceable by robots we may ultimately become.
ChatGPT comes by way of the world’s leading artificial intelligence lab, San Francisco-based OpenAI, and is partly funded by Microsoft, as well as Twitter and TESLA CEO Elon Musk. It is a neural network – a mathematical model loosely based on the human brain that scans vast amounts of digital data and recognizes patterns in that data, allowing it to learn to recognize what it’s looking at (say a cat), as well as to anticipate what comes next in a sequence of text.
It essentially makes very good probabilistic guesses about how to splice together different bits of text in a way that is nuanced and makes sense. It has learned to do this by analyzing billions of examples across the Internet, including Wikipedia articles and books. Interestingly, ChatGPT won’t give you the same results every time, making each work original and difficult to trace. Depending on how you word your prompt, you’ll likely get a unique writeup, even if the context remains the same. And amazingly, it does this in seconds and at speeds no humans can compete with.
We’re already using some of this tech in facial recognition technology and spoken commands you give Google, Siri or Alexa. And OpenAI is also behind DALL-E 2 — an image-generating application that creates completely original work based on what you tell it you want to see (not without its own controversy); it builds these images with bits and pieces created elsewhere (and by someone else) on the web and raising questions about copyright and just what constitutes original work (can we say the same of remixed music?).
OpenAI’s latest model, GPT-3, scanned more digital data than its predecessors (it trained on 540 billion words to be exact), and its programmers are working to resolve issues with problematic language prior models generated (more on this in a bit).
Not only did ChatGPT disrupt what many of us thought currently possible for an AI-fuelled chatbot to accomplish, but it’s also only the beginning, as the tech continues to evolve and learn (cue semi-jokes about HAL3000 and Isaac Asimov-esque plot twists).
All this raises questions about whether we are at that inflection point of singularity – an irreversible future history in which the sentient robots we created cease control and reign supreme. To be clear, it’s hard to make a case for this at present. But still, we are nearing the text equivalent of what roboticist Masahiro Mori termed the “Uncanny Valley” in 1970 (the idea that there is a point past which a robot’s physical likeness to a human becomes deeply unnerving).
How It’s Revolutionizing Teaching, Learning, and… Well… Everything
Now let’s talk about how ChatGPT can impact teaching and learning. Paradigm shifts in teaching and learning aren’t unusual. Just think of the shortcuts calculators, grammar and spell-checkers, reading guides such as Cliff Notes and even the World Wide Web introduced. Students regularly carry the Internet in their pocket, and access to resources has never been easier for much larger segments of society (though still, unequally and incompletely).





























