By Ryan Ripsman
If you haven’t been living under a rock these past couple of months, there is a good chance you’ve heard of, or used, chatGPT-3, a deep-learning text generating model. ChatGPT was created by OpenAI, a company dedicated to producing computer systems that can outperform humans at most conventional jobs. Snippets of chatGPT’s writing can be found all over the internet, with people in awe of how accurate and human-like chatGPT responses are. This software can respond to prompts that range from “how was your day” to “write a poem about bananas in the style of Shakespeare”. ChatGPT is a relatively developed example of artificial intelligence (AI), the performance of tasks and predictions associated with human intelligence by machines.
ChatGPT was trained using reinforcement and supervised learning. Reinforcement learning is a machine learning strategy where there is no “right” and “wrong” output. Instead, the algorithm’s output is given a score or “reward”. Over time, the algorithm learns how to maximize its reward. To evaluate the score of chatGPT’s output, OpenAI used a set of human AI trainers who evaluated the naturalness of chatGPT’s responses to various prompts. Supervised learning, on the other hand, refers to a type of machine learning where the programmer has a desired or “right” answer. Over time, the algorithm learns to select the right answer.
The advent of chatGPT raises many questions about the future of education. The most prominent question is with regards to the future of essay writing. When chatGPT can produce university-level essays in a matter of minutes, it is questionable how schools might assign essay-style assignments. Students who use chatGPT will receive an unfair advantage over those who chose to abide by the rules of academic honesty. Because of the quality of chatGPT’s writing, it will be difficult for teachers and professors to distinguish between AI- and human-generated essays.
There are solutions being developed to ensure and reinforce academic integrity in a post-chatGPT world. OpenAI, the company that produced chatGPT, has created a classifier that can predict the likelihood of a piece of writing being AI written. However, the classifier is not fully effective, and it only works for longer pieces of writing (at least 1000 characters). Furthermore, over time, AI will improve, and it will become more and more difficult to distinguish between AI- and human-written pieces.
Beyond academic integrity issues, chatGPT raises questions about our education system. ChatGPT has been majorly successful at completing writing tasks which only humans could do before. In one preprinted study, raters compared the helpfulness of answers to questions generated by chatGPT and actual humans. They found that the chatGPT answers tended to be more helpful, particularly in the fields of psychology and economics. A different preprinted study found that chatGPT’s performance on a suite of medical exams is comparable to the expected performance of a third-year medical student. Right now, at best, chatGPT could be used as an aid to experts. But over time, as artificial intelligence continues to improve, more and more jobs will start to become unnecessary. This future societal change raises an important pedagogical and
philosophical question: what do we teach a generation of students who may not spend most of their lives working conventional jobs?
Traditionally, school has been a means of training youth to enter the workforce and society at large. But now computers learn in minutes what we spend years trying to understand. So in a world without a typical workforce, a world where people are not forced to work and interact with each other, what role does school play? One could argue that school itself should eventually be phased out. However, we saw during the pandemic that youth need a place to congregate with their peers. Being deprived of human contact for long periods of time leads to a myriad of mental health issues. Furthermore, school provides students relatively safer opportunities to grow as people and learn about the world, which are hard to find elsewhere.
While the advent of a post-work world is still far off, it is time educational systems start to prepare. Instead of preparing students for jobs of the future, prepare students for living a meaningful life. Teach students skills that will make their lives more enjoyable, like social skills and finding their passions. Encourage students to explore their hobbies even if they aren’t economically viable. Teach students to tell stories, not so they can be marketed and sold, but for their own interest and enjoyment.
I was raised in a capitalist world, where every skill I learned was learned to produce value. I cannot imagine a life without work, I can’t imagine a world where value is assigned internally rather than through the whims of the market. I can’t imagine a world where I don’t need to worry about receiving the necessities to survive. But with new innovations in artificial intelligence, there may come a day, when all these necessities are handled by advanced artificial intelligence. The students of tomorrow will need to be prepared to live in that world. Future generations will have to learn to find their own meaning or they may find that much of the meaning humanity currently attributes to itself has been rendered obsolete.
