0:02 Soon after Chat GPT came out, I started
0:04 traveling around the country telling
0:06 people, "Oh no, there's this new thing
0:08 people are going to need to learn to be
0:10 more creative because that's the only
0:12 thing that the AI can't do." I don't say
0:14 that anymore because now I've seen that
0:16 the AI can actually come up with lots
0:19 and lots of ideas. Last year, the
0:21 International Math Olympiad problems,
0:24 four of them were solved by Google's
0:25 artificial intelligence. The
0:27 International Math Olympiad has six
0:29 questions and all six of the questions
0:31 are very very original. They are so
0:33 original that when the national coaches
0:36 meet, they all look at the problems and
0:38 they all try to make sure nothing too
0:39 similar to those problems has ever
0:41 appeared in any contest or anywhere in
0:43 the world before. The questions are
0:45 supposed to be really original. But
0:46 nevertheless, the artificial
0:48 intelligence was able to come up with
0:49 solutions to four out of six, which is
0:52 more than I can do. The creativity in
0:55 the AI can probably surpass what we can
0:58 do, too. The only unique thing about
1:00 human intelligence is that we hopefully
1:03 care that humans still exist.
1:06 I'm Po Shanlow. I'm a mathematician who
1:08 likes to solve real world problems. In
1:10 real life, I'm a math professor at
1:12 Carnegie Melon University, but I'm also
1:14 a social entrepreneur where I've been
1:17 running my own educational solution,
1:19 which tries to make the world a more
1:31 [Music]
1:34 in schools. One of the biggest places
1:37 where students are using AI to cheat on
1:47 This unfortunately could make a huge
1:49 problem for human civilization because
1:52 you just have to think what is that AI
1:54 anyway? It's a large language model. How
1:57 is that AI so good? It's because it's
1:59 good at language. It's good at looking
2:01 at the patterns of words that often
2:05 appear. If many kids lose this ability,
2:07 we'll get many kids who grow up and
2:09 aren't able to think logically. All
2:11 they're able to do is just take whatever
2:13 anyone else gives them. they'll just be
2:16 dependent. If you're already grown up
2:17 and you already have that skill and
2:20 you're using the AI to achieve that task
2:22 because that's for your job, great.
2:24 Okay, you're using it to do a job. But
2:26 if you're in school, why are you doing
2:28 that writing? It's not because the
2:30 writing you make is going to make money
2:32 directly. No, no, that writing is
2:34 actually part of your own learning.
2:36 Using AI to do your writing homework in
2:39 school is like saying, "I'm not going to
2:41 run a mile for exercise. I'm going to
2:44 drive my car one mile for exercise. How
2:45 much exercise you get? You get none.
2:47 You're going to grow up and you're not
2:48 going to be able to be as physically
2:50 fit. Similar thing here with mentally
2:53 fit. And just this observation that the
2:55 power of the large language model is the
2:58 L, the language. That's why we need to
3:00 really make sure that all of our kids,
3:01 if any, and if you're watching this and
3:03 you're students, this is why you need to
3:05 be really, really good with language for
3:07 the next generation. All of these skills
3:09 like reading and writing, communication,
3:11 logic, these are all going to be very
3:13 important because these are how you
3:16 develop a a good way to think. I was
3:17 explaining to people the reason why we
3:19 do this math is not because eventually
3:21 we have to do algebra, but the reason is
3:23 because eventually this just makes you
3:25 smarter. So you're able to go and think
3:27 through situations you haven't seen
3:29 before and figure them out. I interview
3:31 lots and lots of high school students
3:34 who want to work with me. And during the
3:35 interview, the way that I interview for
3:38 intelligence is I ask them questions,
3:39 usually math questions because those are
3:41 analytical. And I ask them questions
3:43 until it's very clear from their body
3:45 language that they have never seen this
3:47 question before. Because usually if
3:48 you're doing math questions and you've
3:49 seen it before, you have a certain look
3:51 of confidence in your eyes. But I wait
3:53 until it's really clear that they have
3:55 never seen this before. And then I want
3:58 to see how you think. Actually, because
3:59 you've never seen that question before,
4:01 the expectation is that you won't solve
4:04 it. And so then I start to give hints.
4:06 Those hints are usually ideas that they
4:08 wouldn't have encountered before in
4:10 school. And then I want to see how
4:13 quickly can you synthesize these new
4:14 hints, these new ways of thinking,
4:17 synthesize them into a solution for a
4:20 problem we have never seen before.
4:23 Actually, that's also creativity.
4:25 So that's one particular piece. But
4:27 going forward, I think that one of the
4:29 skills that people will really need is
4:33 that aspect of actually wanting to
4:35 create value and delight in other
4:38 people. Why do I say this? I say this
4:41 because for many, many years, humans
4:44 were the top species, the most capable
4:47 things on this planet. Soon it will not
4:50 be that case. Soon, you're going to have
4:53 to work together to survive. The only
4:56 way to get other people to want to team
4:59 up with you is for you to authentically
5:02 and deeply be a person who is motivated
5:04 by creating value in the other. If you
5:07 are not that way, you are a bad partner
5:08 and people will not want to go and team
5:11 with you. If they don't team with you,
5:16 you will die. You will lose opportunity
5:19 because eventually all of these kinds of
5:21 jobs you can use AIS to do. Then why
5:22 would anyone want to employ you? Why
5:24 would anyone want to have you as as
5:26 someone on their team? Presumably only
5:28 because they somehow felt like you are
5:30 going to create some value and they they
5:32 like that vibe. They like you. I think
5:34 what we really need is to get more and
5:36 more people who are figuring out what
5:39 the real problem to solve is. But
5:41 unfortunately sometimes when kids just
5:43 think about problems, they don't realize
5:46 that the way that you solve a problem is
5:47 through empathy and through relating to
5:50 other people. Why? because you can't
5:52 solve a problem unless you can visualize
5:55 it through their eyes. I do spend a
5:57 significant amount of my life working
6:00 towards the goal of being better at
6:02 simulating the world. I also use AI for
6:05 that. I think I'll give one exact
6:07 example. I was just in Nashville,
6:08 Tennessee last week and I just saw a
6:11 really I thought very very talented
6:14 singer in one of the bars there. Wow,
6:17 she's good. And I just got curious, how
6:18 hard is it to be able to get a
6:20 performing spot on Broadway in
6:23 Nashville. So I asked AI and I was
6:24 actually not just interested in AI tell
6:26 me about it. I I want to see the links,
6:28 you know, tell me more. What is the
6:29 background of this particular place here
6:31 she's performing at? Right. I I will
6:33 make my own conclusions based on
6:35 knowing. Oh, oh, oh, I see. So there's
6:36 all of these different people who would
6:38 want to do this thing, right? So if
6:39 you're one of the people who got picked
6:41 to do it at this prime time, oh, this
6:44 makes logical sense. See, I'm using the
6:47 AI to build the logic inside my brain
6:50 for understanding country music
6:51 performance. And why was I trying to
6:53 understand? Honestly, it's because these
6:54 days I also work with professional
6:56 entertainers. So, I'm also myself always
6:59 scouting. The big heart of this is I
7:01 wasn't using the AI to write the report
7:03 for me. I was using AI to make myself
7:06 better at that particular goal. Being
7:08 able to simulate the world is the
7:11 superpower that makes someone able to be
7:13 a successful entrepreneur. Simulating
7:17 the world allows you to imagine a
7:19 product or imagine a strategy and then
7:22 play it forward in your head. What would
7:31 A long time ago when I started with
7:32 education, I was actually just thinking
7:34 about how to help people do math
7:36 problems. Today when I think back to
7:39 that time, I think I was probably a
7:41 solution looking for a problem in the
7:43 sense that uh somehow I thought it would
7:44 be very important for people to be good
7:46 at math. But then things that happened
7:47 later in my life as I became the
7:49 national coach of the US Olympic math
7:52 team, I saw situations where there were
7:56 so many so clever, so capable people who
7:57 were still so depressed. And
7:59 furthermore, after they graduated from
8:00 high school, they even didn't really
8:02 know what to do next because they
8:03 thought that the point of life was to
8:05 find ways to prove you're better than
8:08 other people. That's when I realized we
8:10 actually will do much better if we think
8:12 about the philosophy to start with,
8:13 right? The philosophy in life should not
8:16 be how do I outdo everyone else? If you
8:18 do that, you will you'll probably never
8:20 be satisfied. But if your philosophy in
8:23 life is, hey, it is actually addictive
8:25 to make a bunch of other people happy.
8:28 Oh, now I can do it for five people. Oh,
8:30 now I can do it for 500 people. Oh, wow.
8:32 Now I can get thousand people to come to
8:34 this thing. The more that you do, the
8:36 more you want to do. And the fun part is
8:38 that correlates also with traditional
8:40 success. Then I realized, ah, I should
8:43 be trying to push this worldwide. And if
8:46 I don't do it, who will? with the things
8:49 I've done in my life, I now have an
8:52 opportunity to go and say, you know,
8:54 I've seen what happens if you go all the
8:56 way in pure competition. I've seen what
8:57 happens if you go all the way and just
8:59 practice problems to do the best on tests.
9:01 tests.
9:05 Actually, that's not the right target.
9:06 And I realized that because of my
9:09 background, I would be able to shift
9:11 mindsets. Then I said, okay, this is
9:13 what I have to do.
9:15 I am a math professor at Carnegie Melon
9:17 University. Uh the way I teach every
9:19 single one of my classes is that I just
9:21 go up to the chalkboard with a piece of
9:23 chalk and I write today we're going to
9:25 do this but I forgot exactly how to do
9:27 this. So I need all of you to suggest
9:29 ideas and then the whole class
9:31 experience becomes one where the
9:32 students are suggesting ideas and I'm
9:34 giving them feedback on their ideas. And
9:36 what I've been working on as a social
9:38 entrepreneur to make it possible for
9:40 everybody to be able to experience
9:42 education in this way is we run an
9:45 online program. It's called live.
9:48 We found a way to scale up the education
9:50 of critical thinking. You see the
9:51 limiting factor is to have enough
9:53 coaches to lead all of this critical
9:55 thinking and brainstorming. And we scale
9:57 it up by making a win-win situation by
9:59 just observing that uh one of the
10:02 greatest ages to learn critical thinking
10:04 is when you're 10 to 13 years old. And
10:06 it turns out that some of the best
10:09 people to lead 10 to 13year-old kids
10:12 into the joy of being thoughtful are
10:14 high school kids who are about 15 to 18
10:15 years old. But what I observed is that
10:18 there were plenty of kids who were 15 to
10:19 18 years old who were already very
10:21 clever academically, especially in math.
10:24 their limiting factor in life eventually
10:25 was not going to be their science,
10:27 technology, engineering or math. It
10:29 would be communication skills,
10:31 confidence, and the ability to win over
10:34 anybody and lead a team. So, I created a
10:37 program where we find kind and
10:39 mathematically clever high schoolers.
10:41 And then I hire professional comedians
10:43 and actors to teach them how to be
10:45 charismatic and how to be able to win
10:47 over anybody because if you can convince
10:49 anybody to love math, you can convince
10:52 anybody to do anything. And so this this
10:54 exchange, they're trained by the actors.
10:56 They coach math. Uh while they're
10:58 coaching math, we pay for an actor to
10:59 watch them for every hour that they're
11:01 coaching math. The actor gives them
11:04 real-time feedback to help them become
11:06 more enthusiastic, emphatic, and a more
11:09 winning personality. Win-win situation.
11:10 Now, the high schoolers directly
11:13 benefit, their life becomes much better.
11:15 In return, they're producing all of
11:17 these critical thinking classes for kids
11:19 all around the country. The architecture
11:22 of this is that it just connects people.
11:24 We are introducing kind and thoughtful
11:25 high school students to middle school
11:27 students to help them see, hey, you
11:30 could become these great people who are
11:32 having fun and also having a very
11:34 fulfilling life. But also the way we
11:36 execute the whole thing, the high
11:38 schoolers all teach in pairs that not
11:40 only makes the experience more fun for
11:42 the people who are learning math, but it
11:44 makes it so that these kind and really
11:46 mathematically clever people get to know
11:48 each other. because I don't know what
11:49 kind of challenges we're going to face
11:50 in the future. I think they're going to
11:52 be quite big and technology will make it
11:54 so that whatever the challenges we face
11:57 are, they're big. So, we also need to
11:59 have a big network of kind people who
12:01 are clever, who know each other, and who
12:03 have trust so that if there are some
12:06 civilization threatening challenges that
12:08 arise, well, then maybe there will be
12:09 some people who will stand up, work
12:11 together, and do something about it.
12:13 This is what I pay attention to as I
12:14 travel around the entire country and the
12:16 world. I'm actually thinking about the
12:19 robustness of human society. Whether or
12:21 not if there was some challenge that
12:23 faced the world due to technology, are
12:26 we as human civilization equipped to
12:29 rise up and face that challenge? To me,
12:41 The work I do is really focused on
12:44 building up autonomous human thinking.
12:46 That's why the core word I use for the
12:49 philosophy is thoughtful. I've watched
12:52 over many years as people are becoming
12:55 less and less interested in thinking. I
12:57 think that actually happened because
12:59 people found out that they can entertain
13:01 themselves with iPads and this
13:03 unfortunately makes people also not have
13:06 as much interest in concentrating and
13:08 thinking about something. AI could make
13:11 that much worse. The fun part of life is
13:14 having your own contribution to the life
13:16 that you live. Actually, I think that's
13:18 why people actually like creativity.
13:20 It's fun. People like to draw. People
13:21 like to put their own flavor. People
13:23 like to wear their own fashion. It
13:25 expresses themsel. This expression of
13:27 yourself, it would be lost if everything
13:30 you do is efficient but just reliant on
13:32 the AI which told you how to dress
13:34 today. That's why I want more and more
13:35 people to discover it's fun to think.
13:37 it's fun to have your own twist on
13:39 things, your own your own idea that you
13:41 inject inside. The other dangerous thing
13:43 that happens if people lose the ability
13:47 to to think and reason is that it makes
13:49 it far easier to deceive people. The
13:52 world is so complicated that if you look
13:54 at any situation in the world, sometimes
13:56 depending on how you tell the story of
13:59 what happened, you can say statements
14:02 that are all true which make you come up
14:04 with a different feeling. I think it's
14:06 really important for people to be
14:08 critical and to for people to be able to
14:09 understand what's really going on
14:11 because sometimes when someone's talking
14:13 to you, they have an agenda. Like I'll
14:14 be frank, I have an agenda. I'm trying
14:16 to build a more thoughtful world and I'm
14:17 going to be very very upfront with you
14:19 on that. Anyone who's watching this
14:21 video, I think it's really important
14:23 that we have as many people as possible
14:25 find out how much fun it is to delight
14:28 other people and to have the ability to
14:30 think and figure that out. That's my
14:32 agenda. But you see, everyone has an
14:35 agenda. And if you can't think for
14:37 yourself and you just listen to some
14:40 authority, what if that agenda is
14:43 actually to your detriment? You'll have
14:46 no way of knowing. The technology
14:48 revolution really made all of us start
14:52 to realize how much of an impact bias
14:55 has in the sense that whoever makes some
14:58 technology tool has some bias. What does
15:00 bias really mean? Well, I guess as a
15:02 mathematician, the way I would say it is
15:05 2 plus two is always four. That's right.
15:07 What's the point of life? Oh, I don't
15:09 know. Like, there's no clear definition.
15:11 What's the point of life? I think the
15:12 point of life is is to delight as many
15:14 other people as you can, but I know that
15:16 you might not necessarily agree, and
15:18 it's not a problem. I think it's it's
15:19 healthy that we may have different
15:22 starting points. The part that becomes
15:25 unhealthy is where there is only a very
15:28 short menu of options each of which is
15:31 followed by a huge number of people.
15:33 That's actually where bias comes in
15:35 because we just we just mentioned so far
15:36 in this in this video a couple of
15:39 different uh sources of AI providers,
15:40 right? We have we have Claude, we have
15:42 OpenAI, there's also Gemini. If you're
15:44 in China, there's Deepseek. There are
15:46 all of these, but that's relatively few.
15:48 If you think about it, that's that would
15:50 be sort of like saying, well, the world
15:52 has lots of different viewpoints. It has
15:54 five of them. Really? No. No. The world
15:57 has seven and a half billion different
15:58 viewpoints. There are 7 and 12 billion
15:59 people. One of the beautiful things about
16:01 about
16:03 I guess humanity is the fact that there
16:05 are so many different ideas all out
16:08 there. And let's be frank, some of the
16:10 ideas are bad. Some of the people are
16:12 unfortunately in prison because they
16:14 decided to kill someone else. Hopefully
16:15 we all understand that this are a bad
16:18 idea. But the point is there are lots of
16:19 different people who are trying
16:20 different kinds of ideas, lots of
16:22 different philosophies. And in this
16:24 great big marketplace of ideas that is
16:27 the world, we see some ideas come out
16:29 and the the variety also allows us to
16:32 have more creativity perhaps. And when I
16:35 look at the different AI tools, well,
16:37 it's actually well known that they have
16:39 certain biases. This is also why for me
16:42 when I try to get the news, I don't only
16:45 go to CNN.com. I also go to Fox News. I
16:49 tune my social media so that my X is all
16:51 tracking Republican uh right-leaning
16:53 viewpoints and my Facebook is all
16:55 tracking left-leaning viewpoints. And I
16:57 look at both of them every day because I
16:58 want to see what's going on. And my
17:01 expectation is yes, you're going to be
17:02 biased. You have a certain view on the
17:03 world and you think you're right. And
17:05 you're biased, too. You have a way of
17:07 thinking of the world. And my job as I
17:08 simulate the world is to try to figure
17:10 out where do you disagree? Ah, you
17:13 disagree perhaps on a few values of how
17:14 people should live their life. And then
17:16 that causes you to have different ways
17:18 of reporting on the story. I think it's
17:20 all the more important now that there's
17:22 AI out here which sounds like a very
17:24 convincing reasonable person. It's even
17:27 more important that people look at
17:29 things and say, "All right, is that
17:32 really the story?" Because I think that
17:35 the AI is going to be so good at looking
17:37 complete that you may think you have the
17:39 entire story on a controversial
17:49 I am an optimist, as you can probably
17:51 tell from the way I'm talking. I'm
17:53 optimistic because I can see that
17:55 thoughtful people stick together. I've
17:57 been working on my work with this whole
17:59 group of high school students and so on
18:02 for 3 years and I can just see there are
18:05 so many kids who are not kids anymore
18:07 who are growing up who are really clever
18:10 but so kind and they also know each
18:13 other. They make me optimistic about the
18:15 future. Not only the ones that I work
18:16 with but the fact that if I'm working
18:18 with these people that means there are
18:19 other people out there too. I'm
18:22 basically seeing with my own eyes the
18:25 power of what can be done if you bring
18:28 together people who have in common this
18:30 idea that they actually like people and
18:33 they want to make the world brighter.
18:35 See, one reason I'm building this whole
18:38 community of really kind and clever
18:41 people is because I'm also hoping to
18:43 eventually build a whole lot of social
18:45 entrepreneurs. Social entrepreneur is
18:47 someone who makes a conscious decision
18:50 to not necessarily become that rich.
18:52 You're doing it for a different purpose.
18:54 It's not because you want tons of money.
18:56 It's because what makes you happy is
18:58 seeing lots of delighted people. Social
19:01 entrepreneurship is hard. The hard part
19:04 is that sometimes at the beginning you
19:06 feel this is an important problem that
19:09 needs to be solved. Why won't people pay
19:12 me to do it? I need a donation or an
19:15 investment or something. That's how I
19:17 started off. Today when I do my social
19:20 entrepreneurship, I think very carefully
19:22 how much does that problem actually
19:24 cost? You know that if it's a problem,
19:26 there's a real cost. In fact, it should
19:28 be possible to make money. That word
19:30 entrepreneur hopefully also includes
19:33 something about sustainability through
19:35 business type form. A good social
19:38 enterprise should make a net profit on
19:40 every single person that they help.
19:42 Right? The problem that happens with
19:43 social entrepreneurship sometimes is
19:45 that people are doing something which
19:48 would be nice to solve, but they haven't
19:50 found a way to tie that together with
19:52 solving a problem that has real monetary
19:54 value that you can find someone to pay
19:56 for. You should hopefully be able to
19:58 figure out how it costs the world
20:00 something that the problem is there.
20:02 Aha. Now try to figure out who is paying
20:04 that cost right now and see if maybe
20:06 they would be interested in paying you
20:08 for you to go and solve that problem.
20:10 You just have to go and hunt around how
20:13 can I figure out how the positive that's
20:16 created on average is spread apart
20:18 across everyone who is playing and then
20:20 you also get a little bit too. That's
20:21 actually what we do. For example, we
20:24 provide all this education since we make
20:26 the best after school math classes for
20:27 middle school students who want
20:29 enrichment in the world. We have lots of
20:31 customers. They pay for it. So, we have
20:33 the profit. But because I'm a social
20:36 entrepreneur, we subsidize students from
20:37 different parts of the country and
20:38 different parts of the world. How come
20:40 nobody else ever thought of putting
20:43 professional actors to live coach high
20:46 school math stars who are kind while
20:47 they live stream math classes to middle
20:49 school kids? And that whole thing builds
20:50 an ecosystem that builds a more
20:52 thoughtful world that can hopefully help
20:55 to sustain our human civilization. It's
20:57 actually an obvious idea. There are
21:00 actors. There are math people. We can
21:04 all win. It took 11 years to really
21:05 build up to the stage that I am right
21:07 now. It took 8 years to come up with
21:08 that idea.
21:10 You can imagine how many other weird
21:13 ideas came about along the way. Anytime
21:15 I see any thing in the world or
21:17 something I need to do, I try to keep
21:19 coming up with a weird new way to do it.
21:21 For fun, I like to constantly be
21:23 thinking, is there another way? Is there
21:25 another way? Is there another way? When
21:28 I do this most of the time I come up
21:29 with a way and then the next question
21:31 becomes all right what could be wrong
21:34 with it. So I first generate like this I
21:36 want to generate new idea and after the
21:38 new ideas there I say now I want to
21:39 destroy the new idea. I want to shoot it
21:42 down. How could it possibly be wrong?
21:44 Because 99% of the ideas that I generate
21:47 that are new are fundamentally flawed.
21:51 But 1% are good. You have to just be
21:54 constantly for its own sake saying could
21:56 there be another way of doing this?
21:57 Could there be another way of doing
21:59 this? So this combination of generate
22:02 with excitement, destroy with lots and
22:04 lots of effort and then after a while
22:08 one diamond appears
22:11 then push it. [Music]
