Producing More, Understanding Less

We ended the previous course

in our AI for lawyers and other advocate series

by grappling with the problem of

delegation remorse. That feeling of

regret that comes when you realize the

task, assignment, or activity you delegated away is actually something you

wish you could keep doing. The initial example we used came from Economist

Steve Levitt, the co-author of Freakonomics, and a past recipient of the

Prestigious Bates Medal, which is given to the

top American economist under the age of 40. Here's a reminder of

how Levitt articulated the delegation remorse

he experienced after. In an effort to become more

efficient as a scholar, he began to delegate to

his research assistants, the extremely time-consuming, but also to Levitt, very joy-producing

task of data analysis. I ended up delegating

away what I loved in pursuit of

higher productivity. In the end, I was

productive, but honestly, all the fun went out of it, and I soured on

academic research. A related concern comes in the paper by

the anthropologist Lisa Messeri and

the neuroscientist Michelle Crockett, we

mentioned in course two. They're the ones who worry

that an overdependence on AI, particularly among scientists

might lead us to a world in which we produce more

but understand less. Messeri and Crockett

are not AI Domers, early on in the paper, which was published in

the journal Nature, under the title Artificial

Intelligence and Illusions of Understanding

and Scientific Research they make clear that we do

not take the position that AI should never be used

in scientific research. Soon after the paper was

published, Crockett, who specializes in

machine learning and natural language processing, told the reporter at the Science Magazine

ARS Technica that I have used AI in my work, and I'm really excited

about its potential. Yet both Crockett and Messeri, nevertheless worry about

what may happen should AI-assisted research come to dominate the production

of scientific knowledge. One of their particular fears is a situation where because of the seductive convenience

and efficiency of AI tools, more and more people rely on these tools for more and more

of the research process. Reading, summarizing, and analyzing existing

data and information, generating new ideas

and hypotheses, describing, editing, and

disseminating findings. This intellectual

homogenization might result in a scientific

monoculture that lacks crucial elements

of creativity and diversity and is

vulnerable to error, bias and missed opportunities

for innovation. In other words,

if everybody uses the same co-author

and co-brainstormer, and co-editor, we could find ourselves with a

dangerously narrow, stagnant, and incomplete

understanding of the world. Messeri and Crockett's

misgivings may remind you of a few of the red flags raised by somebody else we talked

about in course 2, Ethan Mollick, a business

professor at Wharton, and the author of

Co-Intelligence, living and working with AI. Here, as a review, is how he explained

the risk that AI's alluring ease might have a stifling effect on invention

and original thinking. The implication of having

AI write our first drafts. Even if we do the

work ourselves, which is not a given are huge. One consequence is

that we could lose our creativity and originality. When we use AI to generate

our first drafts, we tend to anchor on the first idea that

the machine produces, which influences

our future work. Even if we rewrite the

drafts completely, they will still be tainted

by the AI's influence. We describe this process

as a path dependence. We're starting out in

a certain direction constrains the next moves

you might decide to take. If we rely on AI to set our

initial options for us, we miss out on the

chance to explore different perspectives and

alternatives, Mollick writes. Some of which may lead to much better solutions

and insights. We'll want to keep

these considerations in mind throughout this course, as we turn our attention

to the role AI may have and already has had an

education and leadership. How should teachers approach AI? How should students approach AI? How about the leaders

of law firms, government agencies, and

nonprofit organizations, all of whom know that a key driver of

success is making sure their employees have the skills necessary to thrive in

a fast-changing world? Many like Sal Khan, the founder of the Online Education Powerhouse

Khan Academy are incredibly

optimistic about AI. In brave new words, how AI will revolutionize education and why that's

a good thing he details how Khan Academy has used AI to create a

platform that offers every person an

opportunity to engage deeply in the education

process in entirely new ways. Among other things, it provides a personalized and patient

tutor that focuses on the learners' interests

or struggles and empowers educators to better understand how they can fully

support their students. There's also the

book Teaching with AI by Jose Antonio Bowen, a former president

of Goucher College, and C Edward Watson, the current Vice President

for Digital Innovation at the Association of American

Colleges and Universities. While acknowledging that, AI is already increasing inequity, both in education and beyond. They also make a case for its potential as a

powerfully beneficial tool. It can increase

human creativity and customize materials for groups

or individual students. Others like Dan Meyer, who earned his doctor in math education from

Stanford and later became the chief

academic officer at the Online Learning

Company Desmos, are more apprehensive, particularly when it comes

to chatbot pedagogy. Generative AI is best he has written at what something

teachers need least, Namely personalizing the text of a word problem to fit

a student's interest. He then cites multiple

studies that suggest that this kind of superficial intervention, where,

for instance, the math questions given to a student who likes basketball, will include examples

about basketball, and the math questions given to a student who likes puppies, will include examples about puppies don't really

work that well. You'll get to engage with these and other perspectives throughout the rest

of the course, beginning with one

I try to convey, whenever I talk about artificial intelligence,

with students, lawyers, teachers, administrators,

government officials, and pretty much anyone else. AI, when used well, can be an incredible

feedback machine.