sds podcast episode 277: the new age of reason · 2019-07-10 · inspire, and guide you. there are...
TRANSCRIPT
Kirill Eremenko: This is episode number 277 with Serial Entrepreneur,
Khai Pham.
Kirill Eremenko: Welcome to the SuperDataScience Podcast. My name
is Kirill Eremenko, Data Science Coach and Lifestyle
Entrepreneur. Each week, we bring inspiring people
and ideas to help you build your successful career in
data science. Thanks for being here today, and now
let's make the complex simple.
Kirill Eremenko: This episode is brought to you by our very own data
science conference, DataScienceGO 2019. There are
plenty of data science conferences out there.
DataScienceGO is not your ordinary data science
event. This is a conference dedicated to career
advancement. We have three days of immersive talks,
panels and training sessions designed to teach,
inspire, and guide you. There are three separate career
tracks involved, so whether you're a beginner, a
practitioner or a manager you can find a career track
for you and select the right talks to advance your
career.
Kirill Eremenko: We're expecting 40 speakers, that’s four, zero, 40
speakers to join us for DataScienceGO 2019. And just
to give you a taste of what to expect, here are some of
the speakers that we had in the previous years:
Creator of Makeover Monday Andy Kriebel, AI Thought
Leader Ben Taylor, Data Science Influencer Randy Lao,
Data Science Mentor Kristen Kehrer, Founder of Visual
Cinnamon Nadieh Bremer, Technology Futurist Pablos
Holman, and many, many more.
Kirill Eremenko: This year we will have over 800 attendees from
beginners to data scientists to managers and leaders.
So there will be plenty of networking opportunities
with our attendees and speakers, and you don't want
to miss out on that. That's the best way to grow your
data science network and grow your career. And as a
bonus there will be a track for executives. So if you're
an executive listening to this, check this out. Last year
at DataScienceGO X, which is our special track for
executives, we had key business decision makers from
Ellie Mae, Levi Strauss, Dell, Red Bull, and more.
Kirill Eremenko: So whether you're a beginner, practitioner, manager or
executive, DataScienceGO is for you. DataScienceGO
is happening on the 27th, 28th, 29th of September
2019 in San Diego. Don't miss out. You can get your
tickets at www.datasciencego.com. I would personally
love to see you there, network with you and help
inspire your career or progress your business into the
space of data science. Once again, the website is
www.datasciencego.com, and I'll see you there.
Kirill Eremenko: Welcome back to the SuperDataScience podcast, ladies
and gentlemen. Super excited to have you back here
on the show today and we've got an incredible guest
joining us, Khai Pham, who is a serial entrepreneur.
This is a person who has both an MD and a PhD in
artificial intelligence. Khai founded the company called
DataMind, which was in 2000 acquired by Epiphany
for, wait for it, $400 million. That's $400 million.
That's the second highest AI-based company
acquisition after DeepMind.
Kirill Eremenko: Currently Khai is working on a very cool, very exciting
project called ThinkingNode Life Sciences.ai. And lots
of knowledge bombs. Such an exciting podcast.
Literally just got off the phone. Can't wait for you to
check it out. Here's some previews of what you're going
to hear about.
Kirill Eremenko: Entrepreneurship and data science. Why data science
is an advantage in terms of mindset even to be an
entrepreneur. General artificial intelligence versus
super intelligence and what are the differences and
why you don't really need general artificial intelligence
to get to super intelligence. Democratization of
expertise. Questions are more important than
answers, and hence the reasoning engine versus a
search engine. Becoming a founder of companies and
what experience Khai got out of that. Why companies
need to move from data-driven and machine learning-
driven to reasoning-driven, and what is this whole idea
of reasoning?
Kirill Eremenko: Those are just some of the insights that you'll get from
this episode. It was such an amazing conversation. I'm
really excited for you to check it out. I personally
learned a ton and Khai is a very thought provoking
person with very philosophical ideas, so I think you'll
find this interesting. Without further ado, I bring to
you serial entrepreneur and founder and CEO of
ThinkingNode Life Sciences.ai, Khai Pham.
Kirill Eremenko: Welcome back to the SuperDataScience podcast, ladies
and gentlemen. Super excited to have you on the show
here today. We've got a very exciting guest joining me
for this episode, Khai Pham, calling in from San Diego.
Hi, how are you going today?
Khai Pham: Very good. Very good. I mean, how can you not be
good in San Diego with this weather?
Kirill Eremenko: That's awesome. How's the weather there?
Khai Pham: Fantastic as usual. Blue sky, perfect.
Kirill Eremenko: You live in San Diego, right?
Khai Pham: Yeah. Yeah. I live in San Diego. I moved here for about
seven years now.
Kirill Eremenko: Okay. Very cool. It was such a surprise. For our
listeners, I'm in Paris right now, in France, and I said
to Khai, "I'm in Paris." And you just started talking
French to me. That is so cool.
Khai Pham: Yeah. I mean, when people see me I don't look very a
French guy, but I grew up in France. Where I did all
my studies, my MD, PhD over there. It's my mother
language, I would say.
Kirill Eremenko: Oh, mother language. Wow. That is really cool. Maybe
we can have a podcast in French one day. I'm really
still improving my French, but it would be interesting.
Khai Pham: Well, the problem is, I was born in Vietnam, so I forget
my Vietnamese and I start to forget my French, and
my English will never be good, so I don't speak any
good language today.
Kirill Eremenko: Oh, wow. Wow. On the other hand you've traveled the
world and lived in so many countries, so that's exciting
I guess as well. Yeah, you have both an MD and a PhD
in AI. For our listeners, MD is like medical, in
medicine, a doctor of medicine. PhD is a PhD in AI.
That is such a rare combination. How did you end up
having those two degrees?
Khai Pham: Well, the reason is, because I have an Asian mum. As
you know, Asian mum, you have to be a doctor,
physician.
Kirill Eremenko: Very straight to the point.
Khai Pham: I didn't have a choice. Anyway, I started medicine, but
rapidly, I don't have a lot of memory, so it was tough. I
said to myself, "Yeah. Why computer cannot just
remember everything and just get the information I
need?" Each time I went questioning my chief of staff,
"How can you be sure that you make the best decision
for the patient? Were you able to explore all the
combination?" This is why, this kind of frustration
drive me to AI.
Kirill Eremenko: Very, very cool. Also, I was very impressed to find out
that you were the founder and CEO of DataMind, a
leading AI company that was sold later. It was
acquired in 2000 for $400 million. That is the second
largest AI acquisition after DeepMind, which was
bought by Google I think not that long ago, but that is
really cool. Congratulations on that. That's a massive
accomplishment and breakthrough or like a massive
way you've made an impact in the space of artificial
intelligence.
Khai Pham: Well, yeah. When I started, I didn't even have money to
buy a PC. I started not to make a great exit or
whatever. I just wanted to prove that the technology
idea really works. I wanted to go beyond the academic
environment to show that it can work in the real
world. So yeah. With passion and so on, you just
always find a new way to accomplish what your dream
is about.
Kirill Eremenko: That's fantastic. I find some of the most interesting
stories happen with people starting with nothing.
When you don't have, as you said in your example,
you didn't have enough money to buy the computer,
then you find the way, you breakthrough and you
create something incredible. I think, even though it's
hard at times, especially at the beginning, that's ... I
don't know, it creates some kind of hunger in you
when you want to really succeed and really make an
impact in the world, because you're seeing what
situation you're in and you want to improve that, not
just for yourself, but for others and make a difference
in this world.
Khai Pham: Well, at that time I was younger. I didn't picture in my
mind what kind of impact I can have today. At that
time, I just really believed in what I'm doing. That's it,
and just wanted to share it. This was the fundamental
engine for me to move on. It's not about, at that time
yet, okay, what kind of impact I can do with this or
that. I just believed in what I had and I think that for
everybody that has something that they believe in it,
then it become a passion.
Kirill Eremenko: Fantastic. I love that approach. What are you doing
these days? You sold or that company was acquired
back in 2000. What have you been up to and what is
your current passion?
Khai Pham: Yeah. Since the company has been acquired, it has
been renamed later on to Rightpoint and so on. I
decided to start something in social network, because
the idea is to gather information, data, so it can be
used for machine learning at that time. But then, if
you remember, there was the dotcom crash and then
there was the financial crash in 2008. It was a
rollercoaster. It was a tough time. After that, I decided
to really spend some time to think about, okay, what
really I care. I come back to my first love, which is AI,
and I work on this project for more than six years on
system reasoning, which have a business if you
consider that, it's going to be the next wave in the next
five years. I'm very excited to work on that and we
applied that for life science.
Kirill Eremenko: Got you. What is this concept? We chatted a bit about
this before the podcast that, we want or companies
need to consider moving from being data driven, which
is a very trendy topic right now and very impactful as
well, but according to you, companies need to consider
moving from being data-driven or being machine
learning-driven to reasoning-driven. What is this idea
of reasoning?
Khai Pham: Yeah. Actually, this is a very interesting question that
sometimes some people ask me. What is reasoning?
Actually, it's something we are doing every day without
realizing it. There is mainly two things that are
important first, pattern recognition and problem
solving. Pattern recognition is what human and
animals are doing, which means to recognize
something. We recognize a face, we recognize a piece of
music. It's everything we are doing in a second.
Problem solving is when we start to have some
assumption, hypothesis, deduction, tests back the
assumption to see if it can be true or not and have
plan. Problem solving is really what distinguishes more
from the animal kingdom, even though some animal
has some reasoning but not at the level that we have.
Khai Pham: Machine learning data-driven is a statistical approach
and provide a very, very efficient tool for pattern
recognition, but if you want to go beyond pattern
recognition, which means predicting things, if you
want to understand things, if you want to be able to
intervene, you need reasoning, because you need to
understand the causality of things and you need to be
able to have inference in your mind, which means,
how to deduce things and how to check back if it's
coherent with your knowledge. So reasoning is what
you do every day to solve problem. It's not about just
recognizing an existing situation, but it's about
generating a new idea about generating new
hypotheses and try to solve it.
Kirill Eremenko: As we know, correlation and causation are not the
same thing.
Khai Pham: Yeah, we repeat that all the time, but I'm sure if you go
into a lot of conference and you start to ask people,
actually you will be surprised that sometimes people
confuse about it and how many time on TV, because
they give you some data and it's very confusing. I have
a very funny story about it. In the '50s, there was a
perfect correlation between the sale of ice cream and
the polio outbreak.
Kirill Eremenko: Yeah, yeah. I remember you telling that one the last
time we met at DataScienceGO. I got you, yeah.
Khai Pham: Yeah. At that time people even advised people to eat
less ice cream. Yeah, just because ice cream, yeah,
you eat more ice cream in the summer, and in the
summer the temperature is higher, so it's why you
have a ... the virus is more virulent. This is kind of
example to do not confuse.
Kirill Eremenko: Basically, the correlation was that people are eating
more ice cream and they were getting more polio, but
the common denominator was that it's summer. It's
just hot and that's why ... there is correlation, but
there's no causation between eating ice cream and
getting polio, even though doctors or there was advice
not to eat ice cream so you would avoid polio, is that
right?
Khai Pham: Yeah. Correlation, you just observe that something is
happening at the same time than another thing. They
observed that the sale of ice cream increased at the
same time than the polio outbreak is increasing, but
it's not the cause of polio outbreak. One of the cause, I
mean, one of the factor that participate to the cause of
polio outbreak is high temperature. So yeah, you're in
the summer, the temperature is higher, so it's why
people eat ice cream. It's very important to think about
that when you go to so many AI machine learning
conference in particular for life science, how many
times people are going to focus on the ice cream
instead of on the real cause.
Kirill Eremenko: Got you. Totally agree. Tell us a bit about your recent
or current company. Well, you're the founder CEO at
ThinkingNode Life Science.ai. What is the mission of
the company? What is the vision? Why did you create
it?
Khai Pham: Yeah. ThinkingNode Life Science, our mission is really
to build a global library of reasoning network for life
science. What does that mean? Today you have a lot of
knowledge and every day you have scientists all over
the world working very hard to make new discovery.
Once they have the discovery, it goes to a publication.
Then, at some point, it's end up into a very big
database where you accumulate all these different
knowledge. What we do is, we crunch all this
knowledge and generate a reasoning network that can
either solve the problem directly or help dramatically
the scientists to solve it.
Khai Pham: Because today, knowledge is static, human use the
knowledge to make the reasoning and to solve a
problem. In this case, we want to use the machine to
help human to use this knowledge, because human
can only process five to nine concepts at the same
time. How to make knowledge directly reasoning
capable, if I may say. The idea is to build a library
where we have different reasoning network for different
kind of domain of problem, in immunology, in
microbiome and so on. This is the goal of the
company, so then companies, researchers, can tap
into that like thinking as a service to get the
knowledge to solve the problem they need.
Kirill Eremenko: Okay. How are you going to apply data science or
machine learning to create this?
Khai Pham: Yeah. At the beginning we do not apply machine
learning to do that, because today machine learning
start from scratch. It just used data to build a system
that can make some prediction based on pattern
recognition. For me, it doesn't make sense. You have
to start by building first the reasoning network, the
reasoning model. It's like in medicine, you go first to
medical school to build your mental model, your
reasoning model about medicine. Once you have this
reasoning model, then you practice medicine and you
can improve your reasoning model through
observation, through the different data and so on. So
we build first the reasoning model, or reasoning
network, and then we use data to improve this
reasoning model.
Kirill Eremenko: What will this reasoning network be based on?
Khai Pham: It's called system reasoning. It's completely proprietary
technology, but it's based on existing AI technology, in
particular intelligent agents, but the main thing is,
system reasoning is designed to have a human-like
reasoning. This is important for me because, if you
have a system that human cannot understand, its
limit a number of application. The second thing is, it
doesn't have a logic by itself. In addition to that, it's a
framework that can host different logic in it, because I
don't really that one logic can solve a very complex
problem. It's like human, we are using several logic to
solve a problem. We don't have just one logic in our
mind.
Kirill Eremenko: Okay. Basically, you're going to be aggregating all of
these different papers-
Khai Pham: Knowledge.
Kirill Eremenko: ... knowledge about the life sciences and allowing
researchers to ... helping or facilitating how they
navigate this research and put it together and get
insights for their specific applications or products or
further research that they're doing?
Khai Pham: Yeah. Well, we're kind of mimicking the way the
scientists will use this knowledge. For example, if you
are in synthetic biology and you want to genetically
modify an organism to produce something, what you
do is, you have to decide which organisms you have to
choose and then what kind of genes you are going to
put into this organism, and then to think about, what
can be the consequence of doing that? This take a lot
of time and a lot of experience, it take years for
somebody to master a different organism. In this case,
the system digest all the different organism into the
system so it can do the combination for you directly.
Kirill Eremenko: Okay, Got you. It speeds up the process, that makes it
a bit clear. The whole example situation.
Khai Pham: Another way to see it, it's like Excel for thinking. What
I mean by that is, you can still do accounting on the
paper or you want to throw everything into Excel and
then you can play with it. Like I said, we can only
process five to nine concepts at the same time, so it's
very difficult for us to combine all the criteria.
Khai Pham: Or if you take an example with the doctor, when you
come to see a doctor, you say, "Okay, you know what
doctor, I have this symptom, this symptom, this
symptom, and I take this medication and so on. And I
have in my family ..." At some point, "Okay, wait a
minute," because it go beyond five-nine concept.
Khai Pham: Beside that, the doctor is going to say, "Oh, you tell me
you take this medication for that, but it doesn't make
sense. Are you sure it's about this medication?"
Because the doctor has the reasoning network in his
or her mind, so can check back the consistency, the
coherency, of all these different knowledge to make
sense of it.
Kirill Eremenko: Okay. It can be applied in medicine as well?
Khai Pham: It can be applied in any domain where you have
reasoning.
Kirill Eremenko: Oh, it's not just life sciences? It can be in other
domains as well?
Khai Pham: Yes. Yes, but we want to focus on life science today.
Yeah.
Kirill Eremenko: San Diego is a great place to be for life sciences.
There's a lot of biomedical industry there.
Khai Pham: Yeah. I mean, it's a reason why I moved at that time
from Silicon Valley to San Diego, to be closer to the life
science community. For me, the big difference between
the two places, Silicon Valley is more technology and
San Diego is more science, if I may say.
Kirill Eremenko: I got you. Very interesting. At this stage of your
business, of this new company, you mentioned you're
... at the fundraising stage, tell us a bit about that.
This is very interesting, how much are you looking to
raise? You mentioned you are not interested in the
traditional venture capitalist approach, with the exit.
Can you provide a few comments about that? I found
that quite a interesting approach to raising money.
Khai Pham: Yeah. The thing is, we can either decide to grow the
company progressively through our customers and so
on, is one way to do it, or we can have enough money
to directly develop the major reasoning network that
we believe would be useful for the whole community.
For example, the immune system reasoning network or
maybe start to scratch a little bit more about the
microbiome. For that, we wanted to have a good
funding to just focus on developing that directly,
instead of growing progressively. VC are fantastic
engine for start up and growing up, but as you know,
most of them have the four or five years constraint,
because themself has to show the return at that time.
Khai Pham: We are not interested to have investment where you
are looking for an exit in the next three years or four
years. We really want to partner with investors that
are first looking for impact. For me, money is the
consequence. If you are looking for the right impact,
money will be way more than what you think. Impact-
driven visionary people who can understand that the
20th century was about information, it's why you have
a search engine. The 21st century is about knowledge,
it's why you're going to have a reasoning engine and
we want to be a leader in that. So we are looking for a,
yeah, investor that can see how this can impact any
industry, because it's about problem solving.
Kirill Eremenko: Wow. That is very admirable, at the same time, when
you said that the 20th century has a search engine,
and the 21st century has, should have, or will have a
reasoning engine, everything came together. What you
were talking about before about creating this
knowledge or reasoning network. Basically, what
you're saying is that, you are effectively creating, or
your goal is to create a Google, but not one that just
searches through information, one that helps you
reason. Is that what you're creating?
Khai Pham: You just summarized that. Yeah.
Kirill Eremenko: That is so cool. That is something, and I can totally see
myself doing that. If I have a question, for instance,
right now I'd go on Google. I don't know, how to make
a vegan lasagna. Then I get all these recipes and I have
to go through all this information myself. If on the
other hand there was some sort of other engine that
was a reasoning engine and I put in that question, it
wouldn't just give me information, it would actually, I
guess, tailor some answer to me. It would say, "You
need to take these following steps," or, "Based on your
preferences, Kirill, and based on what you've told us
about yourself, this is what you're going to enjoy the
most. How many people are coming? This is what
you're going to need," and blah, blah, blah. Something
like that. In a very rough description, is that the
difference?
Khai Pham: Yeah, absolutely. The thing is, a lot of time people are
talking about, "AI is going to take job," right?
Kirill Eremenko: Mm-hmm (affirmative).
Khai Pham: And change, the answer is yes. However, the role of
human will be very different. For me, human, we are
not designed to work. We are very weak. Until now the
thinking, the reasoning, apathy is the main thing, but
machine start to get better and better. Each time that
humanity we build machine, it's end up always better
than us. What I mean is, in the future, human, we are
no more there to solve problem. We are there to ask
the right question.
Kirill Eremenko: Wow.
Khai Pham: This is going to be a big shift, because even in
education today, everything is designed based on the
good answer you give, but now who cares about the
good answer? You can already start to see that with
Google, Alexa and so on, but later on it's about
problem solving. What will be important is, what is the
question we ask to the system to solve that really
matter? This is how I envision the future, and by doing
so, we are going to democratize expertise, make it way
cheaper, because the biggest asset that humanity we
have is not knowledge, it's expertise. How to use it, but
it's extremely rare and expensive and not everybody
can benefit from that. The consequence of putting that
into the machine in a digital way, we can really share
and scale all this expertise to help way more people
and solve major problem with environment and so on.
This is the mission and the dream for the company.
Kirill Eremenko: What an amazing dream. I can totally get behind that.
Love the dream. How far away are we from this? How
far are we away from where machines are so good at
answering questions, that it's no longer an occupation
or even an advantage for a human to be able to answer
questions? It all boils down to ask, oh, sorry, yeah, it
all boils down to asking the questions rather than
answering. For now, humans are still better than
machines, in my view, at answering sophisticated
questions involving multiple domains. How far are we
away from machines becoming the go-to for the
answers to the questions?
Khai Pham: Yeah. It's not black and white. If we talk about a
situation where it becomes systematically the machine
does it better, yeah, a lot of people talk about the
singularity. However, the singularity for me, we will get
there not with the machine learning only, we need the
reasoning. This would be, yeah, would make sense at
that time. Now, the other thing is, people talk a lot
about narrow AI, general AI and super intelligence. I
believe that we don't need general AI to reach super
intelligence.
Khai Pham: What I mean by that is, who cares about a system that
know how to go to the restaurant and understand the
menu and so on? Maybe what machine is better is to
have a network of very high skill expertise, connected
all these different expertise together to solve extremely
complex problem. I think that, if we talk about a
general way to solve any kind of problem, yeah,
singularity makes sense, but already today we can
apply in a number of application to solve very complex
problem, which sometimes people call that narrow AI.
But what if we make a network of narrow AI?
Kirill Eremenko: Okay. Very interesting. Could you summarize, what's
the difference between general AI and super
intelligence?
Khai Pham: Yeah. Usually, when people talk about general AI is a
machine that can understand at the human level and
solve problem at the human level. Super intelligence is
way beyond human level. The thing is, like I said, I
don't-
Kirill Eremenko: We don't really need general AI to get through super
human level.
Khai Pham: Yes, for a number of domain and then, why don't we
just, for example, connect all these super intelligence
in medical, in biology, in aerospace, in environment, in
agriculture? We combine them together, and maybe
the system understand nothing about how to behave
in the restaurant, but to get-
Kirill Eremenko: Yeah, or how to go bowling or how to have a picnic, the
human things.
Khai Pham: Yeah. Maybe we don't care much.
Kirill Eremenko: Very interesting. Okay. The cliche question, are you
afraid that a system like that would take over the
world?
Khai Pham: I believe that it's going to change it and it's going to ...
For me, technology is part of nature. It's just nature
that found a faster way to accelerate evolution. We
used to think that evolution is based on biology, well,
now there is technology, because technology come
from us and we are part of nature. The way I see it is
that, at some point, we are going to have a branch like
between the apes and human. We're going to have a
branch between human and machine.
Khai Pham: Machine is going to have its own evolution, because it
will be able to build better and better machine by itself
and human, we will be free from solving either "stupid"
problem or even complex problem. We will be free from
that. We will be able to develop something that we
were not able to develop until now, because we were
busy with our brain to remember things or to solve
problem. Maybe we are going to be able to, like I said,
just spend our time to think about, what is the best
next question?
Kirill Eremenko: Okay. Don't you think that, if everybody is thinking,
what's the best next question, then a lot of people will
be bored or just have not much to do and become
restless in their minds?
Khai Pham: Yeah. I think that, in your life, there are a lot of things
that you realize that ... How many times people say,
"Oh, at the end of day of my life, my family was the
most important and I didn't spend enough time," and
so on and so on? I don't think that, because we
associate too much human with intelligence, but
intelligence is just part of us. We have a bunch of
other dimension that maybe we don't develop enough,
because our society is so demanding on us to solve
problem, and we don't have time to develop the other
part of the human part.
Khai Pham: The other thing is, when you spend time to think
about the question, of course, you don't do stupid
things. What I mean by that is, you think about it. You
are not doing things without thinking about the
consequence of it. I think it will allow human being to
be deeper, to be wiser and to have more time to
develop the human dimension, because a lot of time I
think we are not human yet. We are pre-human and
we just take the title of human when you see what's
going on in the world. Some behavior is difficult, I
mean, it's difficult to be compatible with the human
definition.
Kirill Eremenko: Like what, for example?
Khai Pham: Well, the lack of compassion amaze me, because I
think that it's one of the major feature of human
being. Without that, we would not exist, because at the
beginning we were so weak. We help each other to
grow and so on. Our society today is doing more and
more things to get us more isolated and compassion
agnostic. Compassion is something that, I think, is
very interesting to think about.
Kirill Eremenko: Yeah. I see what you mean. I was just going to say that
it feels like we've actually moved in that sense from
human back to pre-human. I think it hasn't been this
way always, but I think there's been compassion
before, as you said, for us to survive previously
without technology. Without all these bottom layers of
the Maslow's hierarchy of needs taken care of by
automation and economies of scale and things like
that. Before, we had to have compassion, but it feels
like, I agree with you, some of the things that we see
happening in the world demonstrate a severe lack of
compassion or some like the race going towards a lack
of compassion and that's a bit of a shame as well. It
looks like we're moving backwards in that sense.
Khai Pham: Yeah.
Kirill Eremenko: What you're saying is, by having technology or AI take
over further of the answering of the questions, we'll
have more time for compassion and more time to
spend with our loved ones and families and actually be
humans, not pre-humans.
Khai Pham: Well, I am a extreme optimistic person. It's only my
personal opinion. Yes, I think that at least machine is
going to help us to not spend our time for things that
are not worth. When you think about, what is the
probability for you to exist? It's ridiculous. We
apparently have one life and we are going to spend our
life to go in the morning to work and to come back
doing things that we don't even like it, that take time
from our family or the loved ones or doing ... I think,
yeah, machine can help us to be more human.
Kirill Eremenko: Yeah. Totally agree. Do you happen to know Naval
Ravikant, who's the founder of AngelList?
Khai Pham: No. No, I don't know.
Kirill Eremenko: I think he will be very cool for you guys. I don't know
him personally, but if you ever get a chance to meet
him, he's really cool. I was listening to a podcast
recently and he's got interesting views as well on
technology and how things are going to progress, but
he gave this quote, he just said, "A man has," or a
person, "has one life." There's a quote by Confucius,
which I heard Naval quote that, every man or woman,
has two lives and the second one begins when he or
she realizes that they have just the one life. It's pretty
cool quote, yeah?
Khai Pham: Interesting.
Kirill Eremenko: Yeah. I love that personally.
Khai Pham: Thank you for sharing that with me.
Kirill Eremenko: No problem. No problem. I was very deeply inspired by
that. Once you realize you have one life, your attitude
towards life changes, and your second life starts. It's
pretty cool, cool meaningful thing. Khai, you
mentioned at [inaudible 00:41:49] podcast, I think we
talked about this a bit before, singularity. What is
singularity and how does it relate to general AI and
super intelligence? Just quickly, what do you
understand under or what should we see under
singularity, under that term?
Khai Pham: Yeah. I guess there're different definition of it, but I
guess the most common is when machines start to be
better than us in term of solving problem and so on.
For me, it has a different meaning, because this way of
saying singularity is mainly technology view of it, but
for me, singularity is the moment that really nature
will be able to use technology to accelerate evolution,
as I said. Then, it's maybe the beginning of the branch
that I was talking about between machine and human.
Now, how it's connected with super intelligence and so
on, so yeah, usually sometime people, singularity and
super intelligence are synonym and people use to put
in term of chronology, narrow AI, general AI and super
intelligence.
Khai Pham: Like I said to you, I'm not sure we need general AI to
get to super intelligence. It depend what we put into
this term. The other thing too is, even though if we
follow the same logic, the soon as we reach the general
AI, we have the super intelligence. Why? Because just
the machine can process more than five to nine
concepts at the same time. What I mean is, let's
suppose that today you have a doctor, biologist or
finance or whatever, that has the capacity to tap into
all available knowledge in his or her domain and be
able to process thousand and thousand and thousand
of criteria at the same time, don't you think that this
person would be a super intelligent person? What I
mean is, the intelligence is not based on how much
knowledge you have, it's based on how much
knowledge you can combine.
Kirill Eremenko: I see. Interesting. Okay. Got you. That's the whole part
where you were talking about the reasoning. That's
what it is.
Khai Pham: Exactly. It's why, in my presentation, I always talk
about the lady or tiger just to show that it's about
combining knowledge that we solve problem, not just
how much knowledge we have. Today, the world is
looking for to have more and more knowledge, which is
great, and it's why machine learning is there. We have
more and more knowledge, but it's not enough. It's
about how much knowledge we can combine together.
Kirill Eremenko: Fantastic. Fantastic. I love how all this came together.
Khai, you mentioned in your presentations that you
talk about a certain thing, that is a great segue. I want
to give a quick very exciting news just for a second,
news for our listeners that you are coming to
DataScienceGO to present in 2019, that you were in
2018 as a guest and we got to catch up and hang out.
It was really cool, we went to that dinner, it was a
fantastic time, but now in 2019, you're coming back to
be a presenter at DataScienceGO. Very excited. If
anybody doesn't know yet, it's end of September this
year in San Diego. Tell us a bit about that, how do you
feel of coming to DataScienceGO to present and what
will you be talking about?
Khai Pham: First of all, thank you very much for having me at your
event. Like I said before the podcast, I mean, I really
appreciate what you guys are doing, because you
really try to motivate and make people aware about
everything around data science, but like I said, data
science is just the beginning, but it's so important that
people understand how crucial is that. The goal of my
talk, and it's not just for data scientists, which of
course, is important, but it's for general public as well,
is to make people see that, like I said, data science is
just the beginning. You have to see the bigger picture.
Khai Pham: You have to see why we do data science. We do data
science for two main things. One is to have more
knowledge, and two, is to build predictive system,
pattern recognition, but to go to the next step, it's
about reasoning and problem solving. The talk is
about how these two things interact to each other so
both of them can benefit from each other, because if
you only think about data, you're going to miss the big
picture. The talk is about, is to understand which
based on the application the problem you try to solve,
then you know if you need only about machine
learning or you need only about system reasoning or
you will need both.
Kirill Eremenko: Very cool. I'm looking forward to that already. How to
combine, especially after listening to this first part of
the podcast where we learned about reasoning, how to
combine that and how these two pillars of data
science, more knowledge and building predictive
systems, how they can be combined, and reasoning,
what role reasoning plays in all that. Super excited
and I hope those of you who are listening and are
coming to DataScienceGO, are super pumped about
Khai's talk as well. I think you're going to have a whole
crowd of people attending your talk, Khai, very
pumped.
Kirill Eremenko: At this stage, I wanted to switch gears a little bit and
talk about something else that you're doing, which I
find very inspiring and very admirable. You are a
mentor. You are a part of this, I think is a network
called Connect or is it a group? I'd love for you to tell
us a bit more about that, but basically, you spend time
giving back to the community of entrepreneurs, things
that you have learned in your entrepreneurial journey.
Tell us a bit about that. Why do you do it and what are
some interesting highlights from there?
Khai Pham: Yeah. First of all, unfortunately, I have to slow that
down, because the company is in a very active mode
right now, so I had to stop for now. But the idea, as
you said, I mean, I learn a lot from, when I started I
really knew nothing about business. I even never
heard about business plan. A lot of people helped me
and give me advice, but advice, it's important you take
the advice that are positive advice, don't take advice
from experts that are telling you, "No, this you cannot.
No, this cannot." Only take the one that say you,
"Okay. Yeah, this you can." What I mean is, it was
helping me a lot.
Khai Pham: It's important for me to give it back and to see if I can
help some younger entrepreneur to go to the right
direction faster than have to experience things.
Connect is a very interesting organization. They've
been there for 30 years. The people over there are
fantastic. I have, actually today, a lot of people from
Connect working and ThinkingNode Life Science,
because as you know Connect now merge with SDVG
is another amazing organization for startup
community in San Diego. Sometime you just need to
ask the right question to help the entrepreneur to
realize something, and these can have some impact in
the way they see their business.
Kirill Eremenko: Helping somebody like mentoring or coaching is not
even about being smarter, it's about having a different
perspective, isn't it? It's like you see things from a
different way than they do and that might help them
open up their mind or see something new in their own
thinking or in their own product or process.
Khai Pham: Well, I think it's not just about throwing out there your
experience, because each of us, we have unique
experience and it's very important to take that into
account in term of context. I think the first thing is,
it's about really to understand the entrepreneur,
because each entrepreneur is different with the
personality, with the ambition, with the reason and so
on. So to help, first of all, the entrepreneur to ask the
right question, again, in this case.
Khai Pham: The second thing is to then try to put yourself into
their shoes and see, with the experience you have,
what would you do? It's not just about throwing to
them all your experience and that's it, it's more about
understanding who they are, in what situation they
are, and then try to think, "Okay, if I'm in your shoes,
this is what I would do, because of this and because of
that. It doesn't mean that it's the right way. It just
mean, based on my experience, this is what I would
do. Just think about it."
Kirill Eremenko: Yeah, yeah. No, I totally understand. You mentioned
you have a lot of people or quite a few people from
Connect working with you now. Are you at the moment
hiring for any more positions?
Khai Pham: Yeah, sure. We are hiring, even though we are in the
fundraising times, but what is important for me is to
know people. What I mean by that is, hiring is so
important. Having the right skill is one thing, but
having the right mindset is another thing. For
example, for me, human, we went to the moon, not
because of the technology, but because of the mindset.
Because at the time that Kennedy say, "Okay, we go to
the moon," we didn't have any idea how to get there.
Khai Pham: So, yes, we are looking forward to meet people, to
know these people, so when we get the full funding
then we can have the whole team together right away.
We start already the interview and meeting people. We
are looking for people who are really open minded,
people that are not afraid of trying something that they
don't know. You were talking about the quote of
Confucius. I have a quote that I really like from
Picasso. It say something like this, "I like to do things
that I don't know, so I have a chance to learn."
Kirill Eremenko: Very nice.
Khai Pham: "I have a chance to learn how to do it." Yeah, it's a
mindset that we are looking for, because what we're
doing, what we try to achieve, is ambitious, which
means that a lot of time, we are going to realize we are
wrong and we have to change it. It's not a problem. We
do it again and again and again. So persistent people,
of course, brilliant people with the lowest ego, if we
can, yeah. Yeah.
Kirill Eremenko: Yeah. Got you. A timeless approach. Persistent,
talented people with lowest ego. What are your
comments on, you've dealt with a lot of entrepreneurs.
You were and are an entrepreneur yourself. Any advice
for listeners who are into data science, who are data
scientists, and are considering maybe becoming
entrepreneurs? Does being a data scientist give you an
advantage at being an entrepreneur? What areas is
data science best positioned to disrupt in the coming
years?
Khai Pham: Interesting question. I think that the short answer is,
yes, it helps to be data scientist, not just because it's
about data science, but because, when you're a data
scientist, you have a certain way of thinking, which
means, okay, what do I have as a data? And based on
that, what can I deduce from there? If it's not right,
how I can improve it? It's a way of thinking that will
help you to build your company, because company, of
course, it's about ... You have different kind of
company. People always talk about in marketing the
red ocean or the blue ocean.
Khai Pham: The red ocean is where you try to do 10% better than
your competition and the blue ocean, when you create
a totally new market. Of course, it depends on your
personality, what you want to do, but still you need to
gather data, you need to then analyze them and think
about it and so on. Now, related to data science itself,
of course, today it's a very important skill. However,
it's important that people see that very rapidly a
number of tests that data scientists are doing will be
automated with more and more software, making it
easier and easier. So your value is not just about doing
data science, it's about thinking with data science. I
don't know if it makes sense what I'm saying.
Kirill Eremenko: Mm-hmm (affirmative).
Khai Pham: I try to say that, think about how to apply data
science, what is the consequence of applying it and
how you can apply it. Do you have enough data? What
kind of data, and so on. Does the competition can have
this data or not? The technique, as any technique,
evolve and become easier and easier, it will know more
the barrier of entry to entry. So don't take data science
just as an asset by itself, but use it as the way of
thinking and think about your business through it.
Kirill Eremenko: Very wise words. Couldn't agree with you more on
that. Data science, not just an asset. It's going to get
easier to do, therefore, it's going to become more
democratized.
Khai Pham: Yes.
Kirill Eremenko: Use the thinking approaches that you've developed,
the type of mindset, like you said, success is about
mindset as much as it is about mechanics. In fact,
Tony Robbins says that success is 80% psychology,
20% mechanics. It's all in your head, but having this
background in data science is a huge advantage,
specifically in terms of mindset, not just the doing.
Khai Pham: Well, no, absolutely. I would be even more extreme.
For me, everything is about mindset.
Kirill Eremenko: Totally, totally agree. Well, Khai, I just looked at the
clock. I cannot believe how fast this hour has gone by.
I feel like we're just getting started. We could keep
talking for at least another few hours about all of this,
but we need to wrap up.
Khai Pham: Sure.
Kirill Eremenko: We've approached the hour mark. I wanted, before I let
you go, please tell us, how can listeners find you and
follow you or learn more and get more of these
amazing knowledge bombs that you shared today on
the podcast?
Khai Pham: Well, first of all, I am on LinkedIn, so it's easy. Just
contact me there, and maybe putting like it's come
from the podcasts of DataScienceGO. Then I will
understand the context of it, because I try not to take
contact of people that I have no idea. They try to, just
marketing or something, but if the people mention that
it's from DataScienceGO, then it would be different. I
think this is the best way to contact me. Otherwise,
yeah, we have the website ThinkingNode Life
Science.ai, and you can find via email over there.
Kirill Eremenko: Fantastic. Of course, people can come and find you in
person at DataScienceGO in the 28th September of
this year.
Khai Pham: Sure.
Kirill Eremenko: I think that'd be really cool encounter. We'll share all
these links and URLs in the show notes for this
episode. One final question, Khai, for today. What's a
book that you can recommend to our listeners that
can impact their careers or their lives? Something that
you found useful for yourself.
Khai Pham: Yeah, this is a tough question and we talked about
that before the podcast. But I was thinking, there is a
recent book that can be interesting to start to think
about reasoning, is called, The Book Of Why, from
Judea Pearl. It's really explain very well the difference
between machine learning, reasoning, where you go
and so on. I would recommend this book.
Kirill Eremenko: Got you. Could you repeat the name please, again?
Khai Pham: The Book Of Why.
Kirill Eremenko: The Book Of Why, got you.
Khai Pham: From to Judea Pearl.
Kirill Eremenko: The Book of-
Khai Pham: Pearl, P-E-A-R-L, and Judea is J-U-D-E-A.
Kirill Eremenko: Thank you. The Book Of Why.
Khai Pham: Yes.
Kirill Eremenko: Well, on that note, it's thank you so much, Khai, for
joining me today for this chat and sharing these
amazing insights and philosophical things for people
to think on, and best of luck with your project. This
town's extremely exciting. The reasoning engine and if
that's going to be the new Google then that is going to
be so epic and is going to make so many lives easier
and more fun and can get some equal answers. Thank
you so much.
Khai Pham: Thank you very much, Kirill, for having me.
Kirill Eremenko: Thank you, dear friends, for tuning into the
SuperDataScience podcast and joining me and Khai
for this episode. What an amazing person Khai is and
what a fantastic conversation. All these insights that
he shared with us today. I am super pumped and
super humbled to have been part of this and to learn
these things. This whole idea about reasoning engines
and creating reasoning versus being just simply data-
driven or machine learning-driven. That is a brand
new idea, and you can see that it takes somebody who
really thinks about philosophy, who really considers
the future, has visions, has ideas, it really takes a
person like that to come up with something as
complex, and it takes a lot of courage to jump into
that, create a company around that and push the
world in that direction. Push the frontiers of
technology into the space of reasoning.
Kirill Eremenko: I really appreciated what Khai said about questions
versus answers. It'd be interesting to see if indeed
that's where the world will end up. It sounds like a
very exciting place to be in. On that note, you can get
all of the show notes for this episode at
www.superdatascience.com/277. As I mentioned on
the podcast, Khai will be joining us for DataScienceGO
2019, which is on the 27th, 28th and 29th of
September this year, in San Diego. So if you haven't
gotten your tickets yet, make sure to go get them
www.datasciencego.com. That's datasciencego.com, get
your tickets today while they're still on special
promotion, and you can meet Khai and many other
speakers and entrepreneurs and influencers and fellow
data scientists in person. We're looking forward to
hosting from 600 to 800 data scientists this year.
Can't wait to see you there and network with you
personally. Once again, that's datasciencego.com, and
I'll see you there.