硅谷教育颠覆者发出警告：AI将淘汰不愿改革的学校｜硅谷101年度线下大会（全英）

[music] There is a devil's bargain between teacher or professor and student, which is the less work they both do, the happier they both are. And AI can

turbocharge that like nothing else.

The majority of the incoming class in the Ivy League for the last 30 years were millionaires coming in. That is who these universities select.

Fore [music] 23 and me are Turns out that lecture is the least effective way to educate. People

remember 1% of what they heard in a lecture. The second least effective way

lecture. The second least effective way is reading a book. That's I think they retain about 5%. So what is the most effective way

accelerator for learning is about how And uh universities have tended to focus

on one specific age group which are those generally 18 to 25 plus year old young adults when in fact you [music]

have this entire demographic that's growing. That's also we're looking at

growing. That's also we're looking at continuous and lifelong learning.

>> [music] >> All right, good afternoon. Thank you so much for being here with us. My name is Joy Joy Chen. I am very honored to be the moderator for today's panel about AI and education. And also I'm very

and education. And also I'm very delighted to share the stage with three of my dear friends, Ben, Isabelle, and Esther. Um so let's just dive into the

Esther. Um so let's just dive into the conversation. Uh over the past several

conversation. Uh over the past several years uh we have witnessed how AI has transformed how we live, how we work and

of course how we learn and um I'd like to ask uh three uh panelists when you think about AI and education especially

when we look forward uh look towards uh 2030 and beyond what excites you the most and what concerns you the most?

Maybe I can start with uh Esther.

>> So, what excites me the most is that AI can be a tutor for every single student.

They can ask questions. They can ask it over and over again. They don't have to worry about being stupid or anything.

And they can get answers right away. And

that was not the case before. And I

think all teachers should use AI as tutors in the classroom. Um,

unfortunately, I think, you know, we all have exciting ideas over here. I think

that what we need to do is try to get those ideas across to the teachers because, um, right now, I don't know how many of you have kids in the school

system, but they're still doing the same old thing, memorizing answers to last centuries problems. >> And what concerns you the most with AI?

Well, with AI, what concerns me, it's growing really, really quickly and um hopefully ethically.

I think the ethics and the deep fakes and all the fake news and fake information out there, that concerns me the most. So, I think it's important to

the most. So, I think it's important to teach kids how to think critically. And

um they learn to do that by actually making mistakes and doing it again, not by you telling them what to do.

>> Totally agree. And what about you, Isabelle?

>> Yeah, what excites me the most is um um education has not worked for many many children uh and adult learners. Um and

so here we have a technology that sally unlocks profound questions, profound possibilities.

Um we call AI um the WD40 for education. It's unlocking new

for education. It's unlocking new questions uh such as you know what does it mean to be future ready uh in this age of AI? What is the purpose of

education? Things that we have been

education? Things that we have been wanting to ask for a long time. So I'm

very very excited about all those new possibilities.

And then what concerns me the most is um right now AI is used for a lot of efficiency applications in education. Uh

which is good. Um but if we stop if we stop there, we risk replicating the past as opposed to rem reimagining the

future, which I think is uh what I'm most excited about with this technology is new possibilities.

I'll give you a um my concern first and then maybe what I'm excited about. The

concern is based on the state of our education system. When when I was

education system. When when I was looking for a school for my uh eldest daughter about 12 years ago, I tried a

school that was an arts a performing arts school for academically gifted children. And it was that's kind of an

children. And it was that's kind of an interesting combination. And I asked the

interesting combination. And I asked the the son of the founder who was giving the tour as a principal. I said, "How did your mother come up with this idea of of a art school for academically

gifted children?" He says she didn't.

gifted children?" He says she didn't.

She wanted to start an art school. She

had no care in the world about the academics whatsoever. So in 1981, she

academics whatsoever. So in 1981, she used the California state standard curriculum that everybody was using and

we never changed it. And now, which was 12 years ago, we're a year and a half ahead of the rest of the state.

And why do I bring this up? I bring it up because there is a devil's bargain between teacher or professor and student, which is the less work they

both do, the happier they both are. And

AI can turbocharge that like nothing else.

AI can make the teacher's job much, much easier. It can make the students job

easier. It can make the students job much much easier and there's no learning that happens on either side and that's the biggest risk because the incentive

system of everyone involved in the certification or otherwise what's known as the education sector is to do as little work as humanly possible. The

opportunity and the opportunity with every new technology to Isabelle's point is not how to take that old technology and jam it into the current system. It

is to say, "Wow, this old system doesn't work whatsoever.

What kind of system, what kind of educational path is now enabled that couldn't be done before?" And and that is the big opportunity that AI unlocks

for those bold enough to think in that way.

Well, thank you so much for sharing your excitements and concerns. Um, with all the things AI has brought to our life and will continue to bring to our life,

I do think our education education system will have to go through lots of change, lots of transformations to meet the new learning uh uh needs and also

the shifting workforce expectations. Um

so if especially universities actually because they are preparing our uh students for their future career. Um so

if today you guys were given an opportunity to redesign universities um what would your university uh be like uh

maybe I I'd like to start from Ben because you have already start this vision and mission ever since you founded uh Minurva University which by the way is a huge success. Uh

congratulations.

>> Thank you. Um well, Manuva was certainly built uh because of my mistaken assumption or one of the reasons it was built my mistaken assumption that basically AI would be widely available

circa 2012. So I I was off by a few

circa 2012. So I I was off by a few years in that. But I built a university that asked the question of what does it mean to be human when access to

knowledge and not just knowledge but to solutions is readily available. And when

you think about that paradigm, what's left for human beings is to judge, is to discern, to make decisions.

Because we may outsource a lot of work to the machine, but I don't think anybody here would be excited if a machine determined your tax rate,

>> right? You want to be able to decide.

>> right? You want to be able to decide.

You want to be able to participate, to govern, to vote.

And the discernment not in the black and white because it's obvious when something is true or false or at least it used to be but in the black and gray

when there's certainly wrong answers though again we are getting worse and worse in identifying what wrong answers are and there are many potentially

correct answers. An AI is going to

correct answers. An AI is going to produce in a blink of an eye a whole number of completely plausible scenarios. And so like what the video

scenarios. And so like what the video said, which I disagree with, it's not about knowing how to ask a question.

It's knowing how to evaluate the answer that comes out of that asked question.

The ability to to discern actually means that students would need to learn significantly more than they did in the past, not less. And that really is the

big challenge. How do we turbocharge

big challenge. How do we turbocharge what it is that students are able to do coming out of education? and the

disparity in that world between those who are well-educated, not well certified, well educated, actually know how to discern versus those that don't is going to be

colossal.

>> Thank you. Isabelle, you have worked for Stanford for several years. What is your uh future university look like? Yeah, I

would love to see universities take advantage universities and by the way innovators in the room take advantage of um one major demographic and then work

force change that's happening under that's already underway which is that we are living longer lives

uh all around this planet and uh universities have tended to focus on one specific age group which are those

generally 18 to 25 plus year old um young adults when in fact you have this um entire demographic that's growing uh

that's also we look looking at continuous continuous and lifelong learning. So I think there's a huge

learning. So I think there's a huge opportunity that uh universities and new models of adult learning should take advantage of. Um it's also the nice

advantage of. Um it's also the nice thing about it is that also consistent with what we are seeing in the future of work or this future that's already here

where um the number of occupations that any one of us will hold in the future will continue increasing and rising. So

not only discernment um which will be key and critical thinking and all these uh uh critical skills but also adaptability

um and frankly upskilling and reskilling that will need to happen a lot more in our future. So I would love to see um

our future. So I would love to see um and there are already many models that are focused on this but universities and new models uh focus on these um older

adults that will need uh more more and more uh and want to continue learning.

>> That's a very good point. What about

you?

So I'm very practical because I'm a teacher and I can tell you uh in the university the majority of teachers

professors lecture and it's very hard to change old patterns of behavior and so how are you going to change that that's the thing that I am concerned

about turns out that lecture is the least effective way to educate people remember 1% % of what they heard in a

lecture. The second least effective way

lecture. The second least effective way is reading a book. That's I think they retain about 5%. So what is the most

effective way? Most effective is

effective way? Most effective is peerto-peer projectbased learning. When you do

projectbased learning. When you do something that you want to do together with your peers, a project you come up with yourself,

then you remember. So what I think AI can do is help teams of kids working on projects and help them understand what

the mistakes are, what the things are that they should not do or whatever. And

I think all AI, all interaction should be done in groups. Group of two or three because we're then training kids to work

in the real world. All of you work in teams. I'm sure that none of you are working totally by yourself. So why not train the kids for the world that

exists, the world after they go to school. train them to work in groups on

school. train them to work in groups on projects that they personally care about. Also, I'm very realistic that

about. Also, I'm very realistic that changing the school, it's kind of like changing the church. Okay? Never going

to happen.

>> So, may I shouldn't say never, but it's a little hard. So, I think we should change just a small percentage, maybe 20% of the school time at once and then

maybe move to 30%.

Let's see what we can do to actually change the school so that more kids can do at learning that matters.

>> Well, thank you so much. Actually, I

can't agree more. Um when you say like it's very difficult to change the school system and especially nowadays actually we are still having this is this very

much linear learning journeys uh from K12 to higher education. uh people

collect high school diploma um um um a bachelor degree and then master degree and doctor degree. So uh in your uh uh perspective, what do you think about how

the future will hold the values for those degrees? Uh do you think they will

those degrees? Uh do you think they will still have those uh that much value or you think something new will come up?

Some new learning pathways will come up and who want to take the question?

All of you guys actually >> you can you can pick it up. Yeah, if you want.

>> Well, I sorry to say I think that the value of these degrees is going to be diminished as long as those degrees are not tied to the real world and are not

tied to things that you can actually do to make the real world better. So I

think we need to change the way we educate and I think every single class in all universities should have a project-based element that kids can work

together. So one of my grandchildren

together. So one of my grandchildren goes to UC Berkeley. I can just tell you that the classes there are about 500 >> too big.

>> 500. They're sitting there listening to a lecture. Actually, I can tell you both

a lecture. Actually, I can tell you both most of those kids are thinking about what they're going to do after they get out of the lecture.

[laughter] So, we need to do something else. Give

them an opportunity to work on whatever subject it is that they're studying in teams. And that's what I think is important.

I I think that the four-year degree will continue holding a lot of value for some uh people um and then you we will see continued rise of alternative

credentials um that are much more focused on skills. I mean the future is um um showing us already that they will

be uh more and more smaller credentials or carrier technical skills that are rising uh already today with AI. So I

think this will probably be a mix of the two in our future. Um and I'm I'm hopeful as um you know I'm not as negative because I'm teaching at

Stanford the class that's project based.

Uh so maybe I'm seeing already some [laughter] some elements of Estelle's vision happening um in big universities.

>> So does anybody here know if you look back uh 30 years on the incoming class of an Ivy League uh class Ivy plus class

Stanford etc. What is the only characteristic that was shared by the majority of the incoming class? Maybe we

Oh, >> gradepoint average >> over the last 30 years.

>> No, not grade point average. Not for the Well, for the majority, I guess they had some high class rank >> test scores.

>> Some high test scores.

>> Well, that's what they value. I'm

>> talking about demographically.

>> White.

>> Nope.

No, not white. Nope, not the majority.

Millionaires, the majority of the incoming class in the Ivy League for the last 30 years were millionaires coming in.

That is who these universities select.

And in the world before AI, >> that's crazy. to hire the idiot children of very rich people was fine because they were maybe 20% less productive than deserving students who weren't born into

wealth.

>> But that's not true for UC Berkeley. The

UC >> UC is not is not Ivy League. So not Ivy Falls.

>> That's different. [laughter]

>> Well, UC is better.

>> Yeah. So that's different. Different

organization. And so you have a a a very low tax in the past on hiring

well-connected kind of dumb people.

>> In the world of AI, when your employee base is not producing 10x on a per person basis of what the

current employee does, your company is out of business. Now this may not be next year, the year after, but certainly within a decade.

And so the signaling value of a degree which adds impirically no benefit to its graduates zero

is going to quickly match its real output. The signaling value of these

output. The signaling value of these degrees is going to become a negative so long as those institutions do not educate. And in a world where you do not

educate. And in a world where you do not have to go to class because the lectures are useless, you don't have to do anything in the exam because a 36 will

get you an 36 out of 100 correct will get you an A because of a curve.

And where you cannot flunk anything, the value of that degree is going to be reflected in society. And in fact, that is what we are seeing today. the broad

attack on American higher education whether you agree with those tactics or not I happen to not agree with them but

there is a reason there is not a public outro uh outro about what's happening today to universities it's because we

have lost faith in those institutions and that will be reflected in employment >> very interesting sharing thank you so much and putting degrees aside, uh, in

my personal opinion, I think the most important thing will always be skills and the competencies. Um, and in the AI world, I like to ask you three, what do you think will be the most important

skills and the competencies we need to have to help us stay relevant and also competitive in the AI world? Um,

>> Ben, you have the microphone. You can

start. Well, when when I built Manurva University, um we identified more than 80 of these skills and competencies

which made up our core curriculum. And

the reality is that in order to be able to understand the world around you, you cannot really distill your uh your needs, your capabilities

into something very narrow. The reality

is you have to be a systems thinker. And

if you want to be able to understand things systematically, you have to take a proposed solution or a set of data, you have to break it down into a component parts that make sense. You

have to then be able to reformulate those component parts and come up with a unique solution. You have to take that

unique solution. You have to take that unique solution and understand what happens when you actually implement it in the real world. what are unintended consequences or second or third order effects that can come up withh with

that? And you have to then be able to

that? And you have to then be able to explain what it is that you've come up with. Otherwise, anything you've come up

with. Otherwise, anything you've come up with is irrelevant. And that process has dozens of skills that are built into it.

Dozens. And you not only have to know what those skills are, you not only have to be able to become an expert at deploying them, but you have to be able

to know how to deploy them in radically different context because you don't know what is the context that you're going to encounter it in. And in order to do

that, that takes years of study, practice, and deliberate recontextualization.

And that's of course what what we've done at Manurva, but it is not how traditional education institutions are are set up.

>> Esther, you about to say something?

>> Well, I Yes. I think that all kids coming out of

Yes. I think that all kids coming out of school don't have enough training for social emotional skills. They don't know how to work with each other. And so

that's why I think it's really important to work with each other. And one of the things Google focuses on is teamwork and working with other people in on the

team. And some of the most amazing ideas

team. And some of the most amazing ideas have come out of teamwork. But you have to practice. You you know as a child you

to practice. You you know as a child you learn these skills as a child and that's why I think it's really important for them to work together in teams.

>> Yes. I was I was actually going to build

>> Yes. I was I was actually going to build on on Ether uh Esther's comment. So

right now with the rise of AI, we clearly have an increase in the need for all of us to be actually increasingly

human. Uh so human skills are on the

human. Uh so human skills are on the rise. Certainly the concept of

rise. Certainly the concept of collaboration or what I call relational intelligence is on the rise. um as our economies will be increasingly already

uh uh demanding communication collaboration as top skills.

The second element is adaptability.

Um, which gets a little bit to what Ben described, but uh I think adaptability will adaptability is actually very very closely tied with creativity,

which is a skill that many of you entrepreneurs have in this room. Um, but

this this concept of being adaptive is a is one that we'll see more and more. And

then there is a third one which we touched on already on throughout all all these comments which is learning how to learn.

>> I mean this capacity this metacognition uh that we humans hold uh will continue being increasingly important. So

question is then how do we teach more of those skills but uh those three I think will continue rising.

So can I just say that most companies that fail most startups so 90% of them fail and you know the main reason they

fail lack of communication skills lack of people being able to get along so if they fail for that reason why aren't we teaching those skills why why are

universities not allowing kids to have more interaction and social skills so um that's you know I also recommend that

everybody that he read John Door's book.

Do you know who John Door is? Yeah. So,

because I think it's important to learn how to empower the people you work with.

You know, the leader is really important and that the person needs to learn how to empower the people that they are working with and that is involved in

social emotional training.

I totally agree with all the skills and competencies capabilities you three just mentioned and I think to help our learners to develop all those capabilities and the competencies uh we

need to give them guidance um advice support and also inspirations and we all know those are the things um that our great teachers are always good at uh

providing. Uh but talking about

providing. Uh but talking about teachers, uh we also uh heard something from the uh video clip saying some classrooms uh has no teachers at all. Um

so um actually like I've heard more and more concerns these days about whether um in the future AI will completely eventually replace all the teachers in

the classrooms. So I'd like to hear what's your take on this. Maybe I can start from uh Esther as you have been a teacher for many many years.

>> I think the role of the teacher is really important and I don't think that that technology can take over the role of the teacher. You need a human being

there. So I do think that a lot of

there. So I do think that a lot of additional skills can be learned but I think the teacher needs to be there and support the students and help them

understand what they're doing. even if

it's um you know just a small percentage of the time. Uh maybe the teacher could be called a coach. You know that's what some of the people are saying. My

classes were very very large. They were

70 kids in one class. And the reason that they were that big is the kids worked in teams. Every group there were teams of four and five in every group.

And so it was easy for me to call roll because, you know, you just call team one, team two, and so forth. But they

also learned a lot about how to get along with each other. And I'll tell you, when they're in the ninth grade, they don't know how. By the time they're in the 12th grade, if they've been doing

this long enough, they know how.

>> Isaba.

>> Yeah. I mean, our brains are deeply deeply social. Uh we are wired to

deeply social. Uh we are wired to connect as a human species. So we need um human relationships around us including teachers and other other

people, peers uh other people surrounding us for all of us to learn and to thrive. Um actually isolation is

connected to major mental health issues um which we are already seeing. So uh

anyway relationships matter tremendously for learning and thriving. Um and I also believe very similar to my friend Esther

that uh teachers are here to stay. Uh we

need teachers um uh we need to equip teachers with the right AI tools so that AI can actually support teachers in u in

being increasingly uh more relational uh by maybe helping on some admin tasks and things that uh can free up time. Um but

teachers are here to stay because our brains are deeply deeply social as humans. The whole of teachers may evolve

humans. The whole of teachers may evolve as the way that we are transferring knowledge may evolve with some of these technologies. Um but the idea of one

technologies. Um but the idea of one tutor, one child, one AI tutor, one child without any human relationships, I don't see that vision myself. uh because

that would be a world where I don't want to live in. I don't want to have my children live in uh in this world where they are not connected with other humans

where they are uh not flourishing. Um so

um I I think that teachers will remain and um AI will hopefully make actually teaching one of the best profession it has ever been.

I'll give you a mathematical um or statistical response to that question.

Teachers like any population are distributed around a normal curve. Um

the only problem is that the midpoint of the normal curve of teaching is ineffective.

That's a fact. The fact of the matter is is that anybody here can think about a class you took in college, assuming it wasn't last semester, and if a final

exam were to be produced this instant and I would time you on passing that exam, you would all flunk because you didn't learn it.

And so, and that is true of high school, that is true of middle school. It's not

true of elementary school. you do wind up learning to read, write, um do basic math.

But the problem with a normal distribution where your midpoint is ineffective is you've got to get a standard deviation away from the mean in order to get to good teaching. And that

means that one standard deviation uh from the mean in one direction and above is roughly 16 17%.

16 17% of teachers are good or supremely good extraordinary problem is more than 80% are ineffective or much worse

so long as that is the case AI will replace teachers and if we reform the system

to actually allow teachers to have that midpoint become effective >> where more than 80% are effective to

great and less than 20% are bad to terrible then there will be a role for teachers in the future.

>> I can't agree more on what Ben just shared with us. Uh my personal view I think AI won't completely replace all the teachers especially those good

teachers and AI will empower them to make them even better. Um but AI um like teachers they do not know about AI or they do not know how to use technologies

they will be replaced by teachers who do okay actually um our time is up um I think AI and education is such a big topic we can't finish uh the

conversation within 30 40 minutes and I hope our panel has uh given you guys some new perspectives and also some uh advice to carry forward and thank you so much for your insights and uh

perspectives and also of wisdom and I hope you guys enjoy the rest of the uh event. Thank you.

event. Thank you.

[music] Here [music] [music] we go.

[music] [music] Bye.