Comments from Colleagues
I was very pleased that my recent blog, “Five Minds…Rethinking Education in the Era of AI General Intelligence” elicited comments and—with me largely in the background—a rather vigorous exchange among three individuals: my wife Ellen Winner (Emerita Professor at Boston College, and lifelong member of Harvard Project Zero) and longtime friends and colleagues Mindy Kornhaber (Emerita Professor at Penn State) and Susan Engel (Professor at Williams College).
In what follows, I present highlights of these exchanges, with a bit of context as appropriate.
- Howard Gardner
Mindy: On the challenges of making sense of text provided by Large Language Models (LLMs):
The problem for me is not just that people will not have the background information (and/or inclination) to evaluate the output from an LLM. In addition, LLMs are mostly going to be asked for, and provide, declarative knowledge, not procedural knowledge. Therefore, they are not typically going to spit out how to analyze, evaluate, and question that declarative knowledge. And, even if LLMs are asked for procedural knowledge (e.g., how to fly an airplane, write a novel), the list of steps they generate will not enable the development of actual skills. As we all know, procedural knowledge—for flying a plane, playing the piano, evaluating arguments, critiquing a novel, etc.—is a matter of practice. In my opinion, reliance on LLMs is not likely to cultivate practice. LLMs provide "one-and-done" encounters (or, via 7 LLMs, 7 "ones-and-dones") unless the user is otherwise inclined. Many/most LLM users are seeking time-saving shortcuts and not skill building. (I'd bet that 98% of students who are using LLMs want to reallocate time and effort from acquiring declarative knowledge and building procedural knowledge towards sex, drugs, and rock and roll.) There are programs that do cultivate practice/procedural knowledge (e.g., chess-playing programs, go-playing programs, flight training programs), but those consulting LLMs are not typically looking to advance procedural knowledge: Humans are energy-conserving animals, and LLMs enable that v. the expenditure of effort needed to do things well.
I recognize that from ancient Greeks on, scholars have decried the toll on thinking wrought by tools ranging from writing to calculators. My thoughts on AI/LLMs may be eventually, or soon, assigned to that same dust heap. Still, I can't help but think that we're dealing with a different sort of technology. LLMs do not just reduce relatively mindless drudge work or enhance the storing of, and access to, the wisdom of the ages. At this point, LLMs make it easier for people to avoid acquiring knowledge, to bypass the procedural skills needed to evaluate the declarative knowledge that LLMs spit out, and to skip the development of procedural skills in a host of other areas. Per Wechsler, judgment is essential to intelligence, and IMO, development of judgment occurs via ongoing encounters with both the "what and how" of domains.
Howard is more convinced than I am that people will be inclined to gain knowledge and skills in at least those areas of interest to them, and LLMs may help them do that. That is likely true for some segment of the population. However, having spent too many years in the hinterlands, I am more convinced that the majority of the US population is inclined toward watching sports, vicarious encounters with the lives of the rich and famous, and other forms of entertainment. (The latter was also a dominant focus during the fall of the Roman Empire. Ah, well.)
Given the pace of LLM development, if chaos doesn't overwhelm the world and us, we may live to get some answers to the foregoing.
Ellen: On the need for mastering traditional disciplines in the pre-collegiate years
Where I disagree with Howard is that I think at least in grades 1-12, kids should be exposed to a number of disciplines (in addition to reading, writing, and math at least up through algebra, world history, study of another culture, literature, biology, chemistry, physics, a foreign language. These train the mind. They make you think.
Howard: Just a word on possible entry points to the traditional disciplines
Inspired by an HPZ initiative "Open Canopy" and discussions with "Skip" Gates, I favor having young people investigate their own family history and using that as a “soft” entry point to genetics, history, geography, anthropology, etc. I think that kids will sort themselves easily enough in terms of which entry points they enjoy and want to explore further and which should be left for another time and another place.
Also, since I think that colleges/universities should be gradually converted into Aspen/Berkshire kinds of lifelong learning territories, you can always revisit the aforementioned disciplines and topics at ages 30, 60, 90, who knows!
Howard: A word on how different people gravitate toward different disciplines, hobbies, pursuits
Of course, we value people who like to use their minds in the way that thoughtful persons, including intellectuals, do. But I think this is a preoccupation of WEIRD societies, and certain demographies within those societies—faculty members, those who go to APS, legal experts etc.
I would be willing to endorse the idea that we all have bodies and all have minds and that we should be encouraged to use and develop both. But I don’t believe for a minute that we all can or should do so in the same way. I might have a knack for table tennis but none for sumo wrestling. And if you had tried to get Bobby Fischer to write jazz music or Scott Joplin to play chess, you would have been very disappointed by the results.
And where I disagree with you and Ellen, is that I don’t think that there is a great future for those of us who would like to be professors and to write articles for publication…maybe for the select few.
Susan: On capabilities, dispositions, and the appeal of deep thinking
I have a hunch that I think somewhat differently about this than what either of you said in your messages. See if this makes any sense. Perhaps the kinds of ideas that show up at an academic conference come from a very small subset of the population. But many, many people have the capability if not the disposition or experience to consider an idea (their own or someone else’s). You don’t need to be capable of outstanding work in a domain to be able to fruitfully engage in it, wouldn’t you agree? As I recall from what Howard wrote about MI all those years ago, outstanding examples help us understand the structure of that kind of thinking, rather than providing a metric, no? So, I think many, many more people are capable of thinking about something (a topic, a problem, a question): over time, incorporating new information, revising to strengthen, and applying to a new problem. It may not be about the mind/body problem, or the best economic structure for equity, or the limits to evolutionary theory, but it might be whether a different system of health care would be better for more people, whether there’s anyway to know if God exists, or whether there is only one or many kinds of love.
I am not sure, Ellen, that I think the classical forms of knowledge that schools have included are the essential ones. Perhaps it is enough to say that children in school must attain a depth of information in at least say, two different domains (could be five, who knows), so that they learn what it's like to build new ideas based on what’s already known. All I want to argue is that research in neuroscience, developmental and social psychology, shows that most children and some adults feel pleasure when they engage in “deep thinking” (producing new ideas and knowledge). Further, I want to argue that if our schools focused on capitalizing on the inherent pleasure caused by that kind of thinking, there would be a great chance that more adults would love it as they did when they were little, and if it continued to give them pleasure, they’d do it more, and that would be better for society?
Howard: Flexibility in use of our cognitive potentials
Two points:
Enjoying ideas and being able to engage in them—including disagreements—is valuable. I have often wondered whether the same people who turn their noses up at academic discussion DO enjoy similar kinds of discussions when they are about sports or movies.
I don’t really know, because I loathe talk radio and its online equivalents. From the emails that I receive about multiple intelligences, there is more ATTITUDE and PLATITUDE than serious engagement with crunch ideas.
It is important to know whether it is ever too late to kindle serious interests and in-depth discussions—but it's clear to me that you can turn off those motivations when you try to force or cajole young people into reading the works of Jane Austen or going to the Metropolitan Museum or Symphony Hall.
Susan: On what it means to be immersed in a topic
I guess here the word that worries me is “acquainted”, because I never know what that really means, in high school or in college. I want kids to be immersed, not acquainted. And that requires coming up with intellectual experiences they want to immerse themselves in.
A few years ago, my youngest son Sam and I came up with a new set of themes for organizing academic work in high school. When I was the Gaudino Scholar at Williams, developing the Conversation Project, I asked college students what was the hardest thing to talk about with their parents. To my surprise the most common answer was: money. That revelation collided with my sense that we shouldn’t focus on topics at all. Most people do their best thinking when they’re solving a problem, and for the problem to be sufficiently sticky to people below the age of 18, the problems must be problems to them. So, what are the biggest “problems” to teenagers? Money, Sex, and Self. That would be my organizing principle for a high school curriculum.
At its worst, this is just window dressing- topics like economics, math, history for Money; biology, literature and psychology for Sex and every discipline for the Self. It could easily devolve into a trick to make kids think the topics are more relevant than they otherwise would be.
But if it’s taken seriously, this restructuring would lead more students, more of the time, to experience the potential of these disciplines to help them solve the problems that most concern and fascinate them. They would produce rather than absorb knowledge. Instead of a slog that is only meaningful to high achieving kids because it gets them into college, and not meaningful in any way, to all the others, students would experience the power and pleasure of knowledge and good thinking.
At a deeper level, I guess I think the biggest change from elementary school to high school is the capacity to do deep dives into more than one domain or discipline, and then find connections across those disciplines, in order to solve interesting problems. So rather than mandate certain subject matter or bodies of knowledge, I would say high schools should be set up more like colleges. Each school would offer courses in a variety of disciplines (I have thoughts about that too, but will keep it to myself for now). Each student would be required to go deeply into at least two of the available disciplines. Then each would be required to find connections and integrate information, ideas, and methods, across at least two of those disciplines.
Ellen: On how much or how little is needed to think well
This is a brilliant and very persuasive reply.
Lots to think about.
You agree that thinking can’t occur without info.
So, what disciplines should we get kids acquainted with post elementary school? (By subject areas, I meant disciplines—like history, science, literature, math...)
I’m glad I’m asking you this, rather than you asking me this!
Susan: Towards a curriculum for thinking
I tend to think more in terms of the disciplines and kinds of intellectual experiences that kids need, rather than subject areas, along with a few specific essential skills, though even those may soon be totally irrelevant.
I think kids should learn to/get better at:
Pose questions that can be answered with information
Pose questions that can be answered by contemplation (and know the difference between the two)
Identify or construct, and try out, various ways to answer a question
Know when a question is answered, or evaluate the satisfaction provided by a given answer
Think counterfactually
Possible alternatives to all of the above:
Pose and answer “What if-s” and “What are-s”
Have a feel for designing and conducting actual experiments (in the world), and thought experiments (in the mind)
Speculate, then imagine what would make you wrong, and imagine how to find out if you’re right
Consider a text (fiction or non-fiction) from the author’s point of view
Use numbers to answer questions
Consider an idea fully without embracing it
Change one’s mind based on reasons
Be alive to aesthetics
Know what it is to try to make something beautiful
Experience some level of expertise in a complex domain (the million-dollar question here of course is what I mean by “some”?)
Have sustained conversations
The ones I’m less sure of:
Read
Write to think
Write to communicate
To be clear, following Jerry, I don’t think it’s possible to do any of these without acquiring a great deal of information and turning it into knowledge. Being interested in and attuned to information and knowledge is necessary. But it’s not sufficient. And I don’t think it matters all that much which disciplines a given student focuses on. I never learned a whit of geology and nothing of chemistry or physics. Does it matter all that much? One of the most brilliant mathematicians I ever knew had never read a novel. Again, a great privation from my perspective, but it didn’t keep him from being a deep or flexible thinker.
What makes the classic disciplines so compelling is they seem to offer unique ways to view the world, and each is specific in its potential for answering big questions. I am just not sure they are the only ones, and I don’t think we have any evidence whatsoever that there is a certain known level one must have of each of the classical domains, in order to be a good, avid, or deep thinker. So, I don’t put much stock into figuring what content or which discipline is most important.
The biggest problem, I think, is that I have yet to come across a contemporary culture or region where it’s possible to get most kids interested in the domains you studied when you were in school. It seems misguided to me to work so continue trying so hard to educate lots and lots of kids, growing up in a wide range of settings, when in fact it hasn’t worked at all so far. Why do we insist on trying to get them to think about certain topics, in ways that don’t lead anywhere, while simultaneously stifling their inherent impulse to think a lot, but about other topics?
Ellen: A final word on thinking and thinking well
What I meant by “made you think” was that teaching kids about a traditional school subject (history, biology, etc.) in an interesting thought-provoking way will get kids to think. And will also give them the content knowledge that I still think kids should get in school, even if they forget it later on, which they surely will, but they will have had the experience of thinking and as you say, they then may use the skill of thinking well to learn about something else.
Howard questions whether kids need much of the traditional subject areas of school, and that is where I disagree with him (as long as they are taught in a thought-provoking way).
None of this goes against your clever experiment idea—I think, as do you, that kids will start to think right away as soon as they are given a challenging fun problem, not too easy, not to hard (same as for flow, it only happens when not too easy and not too hard).
I’d just like the teaching they receive in school to be like this—give them challenging fun puzzles to think about—about history, or about biology, etc. I don’t want to throw out the subject areas!
Howard: Final word
I hope that readers who have been able to eavesdrop on this exchange have found it worthwhile. If you want to add some points in the comments below, so much the better. I think that the various points cry out for systematic investigations by committed researchers and educators. And if such work is carried out at Harvard Project Zero, that would be gratifying.