Do we need humans anymore?
Mark Schneider, Director of IES | January 4, 2023
I begin with a confession. In 2008, several months after I left my job as commissioner of the National Center for Education Statistics, I was invited to an international conference to discuss NAEP. Since NAEP is a big multifaceted operation, I was at a loss about how to organize my comments. My solution was to turn to a (then) innovative state-of-the-art knowledge management tool—Wikipedia, which had an excellent discussion of NAEP's history, scope, and topics covered. I didn't use the Wikipedia entry verbatim, but it was an essential (and I think uncited) contributor to my talk. It is no surprise that advances in AI-based technologies have improved since my long-ago foray into "tech-enhanced" research and writing.
This is a blog post about OpenAI's ChatGPT.
ChatGPT is a natural language chatbot driven by artificial intelligence that, among other things, can respond to writing prompts. If you haven't already tried ChatGPT, I'm going to suggest you pause your reading and give it a spin. It's fun, challenging, and a potential giant time sink.
As I planned my first blog of the New Year, I thought I would speculate about the trends that will likely have the biggest impact on education research this year. As with my NAEP project 15 years ago, I was faced with a large, complex topic and needed assistance to launch the actual writing. While I love Wikipedia, it would be of little use identifying coming trends in education research. I turned to ChatGPT, which seemed appropriate given the subject of the blog.
My first foray was to ask the chatbot to: "write 3 paragraphs about the future trends in education research in the voice of Mark Schneider Director of IES." The chatbot tapped out the following in about a minute:
As the Director of the Institute of Education Sciences (IES), I believe that there are several key trends that will shape the future of education research.
First and foremost, we will continue to see a greater emphasis on personalized learning. With advances in technology, it is becoming increasingly possible to tailor education to the specific needs and abilities of individual students. This will require education researchers to find ways to effectively assess and measure the effectiveness of personalized learning approaches, as well as to identify the most effective ways to implement them in classrooms.
Another trend that I anticipate will continue to grow in importance is the use of data and analytics in education research. By collecting and analyzing data on student performance, educators and researchers can gain a deeper understanding of what works and what doesn't in the classroom. This will be particularly important as we strive to close the achievement gap and ensure that all students have access to high-quality education.
Finally, I believe that we will see a greater focus on the integration of social and emotional learning (SEL) into education. Research has shown that students who have strong SEL skills are more likely to be successful academically and in life. As such, education researchers will need to find ways to effectively teach and assess SEL skills in the classroom.
Overall, these trends point to a future where education is more tailored to the needs of individual students, more data-driven, and more focused on the development of social and emotional skills.
My observations:
- It identified some key trends remarkably fast.
- It is quite well written but banal and without much personality (which of course led to all kinds of self-doubt—are my blogs really this bland?).
- It came remarkably close to identifying the trends I had already intended to write about: AI, big data/data analytics, closing equity gaps exacerbated by COVID, and the need to increase the speed of education research, statistics, and assessment.
- How was the chatbot counting? The chatbot did not follow the instructions to write 3 paragraphs (it crafted 5), but 3 were substantive and two were "ceremonial" throat clearing. (Looking at some of my previous blogs, I discovered a penchant for this pattern—so I will give the chatbot a pass.)
Part of the trick in using ChatGPT effectively is making your preferences known. If I wanted my blog post to have more personality, I'd need to clarify that for my new writing assistant.
Prompt attempt #2: "Write a blog about trends in education research in the voice of IES Director Mark Schneider that begins with a humorous parable about predicting the future." And the output begins:
[ . . . ] Predicting the future can be a tricky business. Just ask the old farmer who was once asked how he knew it was going to rain. "Well," he said, "I can't be sure, but the cows are all huddled under the trees, the pigs are all snuffling in the mud, and the chickens are all roosting in the barn. That's usually a pretty good indication that it's going to rain."
In a similar vein, I like to look for "indicators" in the world of education research to help me get a sense of what's on the horizon. And there are a few key trends that I think are worth paying attention to.
Better! Maybe not winning any Pulitzers, but a stylistic improvement over the original. (Although the joke is awfully lame.)
So, what if we try the voice of a real Pulitzer-prize winner? Prompt #3: "Write a blog post about trends in education research in the voice of Nicholas Kristof that begins with a vivid analogy about the perils of predicting the future." Result:
[. . . ] I've always been wary of trying to predict the future. It's like trying to navigate a rocky coastline in the dark: you can see the faint outlines of what's ahead, but you can't be sure exactly what lies in wait. This is especially true when it comes to education research, where new insights and technologies are constantly emerging. Still, that hasn't stopped me from looking for patterns and trends that might help us chart a course for the future . . .
Progress! I won't reprint my entire afternoon of robot exchanges. But suffice it to say that ChatGPT presages big changes in the coming year. The chatbot highlights how so many different innovations (such as AI, machine learning, deep language processing) are integrating and building on each other. I can't even wrap my (human-sized) brain around how all these innovations are going to change education, education research, and the world around us.
For those of you in the classroom, there are other challenges. One of the more interesting ones is the growing use of automated grading bots so that professors can assign more writing without the incredibly heavy burden of reading hundreds of papers generated in a large lecture class. Imagine: chatbots write essays and grading bots grade them! Where do humans fit in?
I haven't taught students for two decades, but even back then, issues of plagiarism and using Google to find material to "recycle" for essays were front and center (and I already fessed up to my "semi" plagiarism using Wikipedia). How do we police this ever-growing sophistication in producing coherent (grammatically correct and spelling error free) essays so that students are forced to think and write? Should we delegate much of that task to chatbots and instead focus on teaching students the skills they will need for an AI-dominated future? For example, how to identify issues that we need AI to solve for us? How to craft the right questions to pose? How to produce and feed data to AI to improve its accuracy? How to think about ethical dilemmas and the changing role of humans?
This blog was supposed to highlight the big trends of the coming year. But the most pressing matters for us all are simply determining how we can keep up with the accelerating pace of innovations and assessing whether and how they can be harnessed to improve learner outcomes across the life span. OK, more of a challenge than a trend, but that sentence was written by a human, not by a chatbot.
Here are best wishes for the New Year (and for those of you who observe dry January, good luck).
As always, feel free to write me (or have your favorite chatbot do it for you). Mark.schneider@ed.gov
BTW: I have kept the "ceremonial" ending so as not to confuse future chatbots.