it’s that we are currently in a phase
where business people and I guess this
is the right place to save things like
that are kind of outmaneuvered by tech
people and this is why I want to give
you a little glimpse in what is
happening in technology like a business
person normally has a goal and a tech
person has a problem that does not mean
we’re all depressed it just means that
we’re look at things with a distance and
the result of that stuff is that a
business person normally executes a plan
as we’re a tech guy writes a program
executing a plan is very flexible right
you say you have a plan and then the
executes but it keeps changing all the
time
a program does not change so much but it
can be executed millions and millions of
times and the interesting thing about AI
is it introduces flexibility into this
kind of solution and that’s what makes
it very powerful very powerful in terms
of economy we do have two different
types of economy we have the new economy
in the old economy unfortunately they
are set to clash extremely because the
modern platform companies make all their
money with 20 percent of the revenue
then some analyst tells them to have
some profit and margin and the rest is
left over normally more than 50 percent
which means that a platform company has
the freedom to try stuff no problem
if Alibaba for example chooses to try
and build a cancer drug and wastes a
billion dollars and then says it doesn’t
work what is going to happen to Alibaba
stock price nothing people are going to
congratulate mr. ma that he tried if on
the other hand Pfizer tries the same
thing which is the company where we know
that there is very little room from
nuvaring spends a billion dollar on a
cancer drug even though they’re the
bigger experts in cancer drugs then
Alibaba would be the drug does not work
the CEO has to step in front of the
crowd and say like huh we have to cancel
this billion dollar program there is a
vacancy on the CEO seat the next morning
and the stock price has collapsed by 30%
the problem ladies and gentlemen is that
this means the established companies
that have all these knowledge can’t even
try to build a good cancer drug because
they have to make very small steps the
next part and we call this disruption
the next is disintermediation right in
the old style of economy we have kind of
this model here on this side where a lot
of people who produce goods and services
are competing on multiple channels to
screw you the consumer they have to make
money with one single transaction which
means they will do anything so you buy
on the other hand if between you and the
person who produces something there is
not a lot of marketing channels but
there is a digital assistant who knows
you quite well and who only needs to
make money in your lifetime on you
you’re gonna get way better decisions it
is not that these new models are anti
consumer or whatever as a consumer
you’re getting the better results from a
disintermediated economy these two
models mean that a the old companies
don’t even see the new companies coming
until they replace them and B if you
want to take this to the extreme there
are currently eight companies in the
world that can produce a digital
assistant the world has eight customers
to the rest of the industry so if you
want to survive in that kind of a world
if you just want to survive as a company
you need to do three things you need to
hmm have a strong brand so people asked
for you you need to do innovation so you
do things that others don’t do and you
probably also need to offer services
because in the end people like to have a
person-to-person interaction if that’s
not happening at the point of sales it
is services my prediction to you the
services part that we have worked the
last 50 70 years to
rid off replaced the bank teller with
the ATM replaced the post guy with the
postbox and so on right so all that
stuff is coming back for survival but if
you want to compete if you actually want
to thrive there’s only one way to do
that we just change your business models
so all you guys studying here have a
great fail because you get to write the
new economics of exponential models and
can you create a market from this or is
it in dangerous all I kept all lalala
what a lot of people fear I think this
will end up in a fantastic market but we
have to get people on board with this
and this is why a AI is important
artificial intelligence and business go
together because in order to do any of
this you need money and people and here
comes AI what is happening in artificial
intense intelligence right now means
that any process can and will be run by
an AI today we can do this with today’s
technology there are three types of AI
there’s the one that you read about in
the paper which is the robot that tails
you I love you and actually means it and
knows what it’s saying sorry I just have
to tell you that’s not real
no one can build this none of us knows
even how to do that on the other hand
you have these narrow eyes where people
just download one algorithm off the
internet feed it with a lot of data
that’s the programmers answer to
McKinsey because what does that do it
creates efficiency it looks into a
single topic and creates a lot of
efficiency for my favorite narrow AI is
a program that has learned how to grow
tomatoes in a greenhouse you connect the
lights you connect the fertilizer you
connect the water proof 20% more
tomatoes unfortunately it’s an English
company that has this they’d find out
the tomatoes in England don’t taste so
great they try to grow beans obviously
this narrow AI kills all the beans
because some things don’t turn red
efficiency nothing we have to be afraid
of right because our companies have been
doing this all the time we are this is
what we do every day only.this these
technologies in AI mainly machine
learning allows us to do it with many
more free variables
the center of this is generally I things
that can learn from experience and then
execute this experience to take
completely individual solutions so in
order to get this we have to separate
the height from the science the reason
why I do talks at all is you know I’m
this nerd
I like basements and screens like not
like lights and stuff but it needs to be
separated right so we kind of separated
three things that AI is not number one
computers don’t understand anything so
if someone tells you my machine
understands blah blah blah you are
listening to the marketing record what
computers can do is they can compare
understanding structures that are
already created with an incoming data
stream and the great things that
computers can do is they can do this
24/7 without ever going tired but they
don’t understand a thing of the data
that you’re presenting them number two
to say that a human brain is the model
for a computer or a computer is model
after the human brain ladies and
gentleman is roughly like saying an
amoeba is modeled after the human body
which it clearly is not just and they
are similar yes because they both have
cells only the one is one so and the
other is a few billion so um a large
neural network today has about a million
a million-two
nodes and it needs half a nuclear power
plant to run a human brain has 86
billion nodes and that’s after it killed
off around 30 billion in puberty this
really happens read it and it only runs
on 20 watts like I don’t get this but
you are all fit so you get this right so
um and in the brain there’s also a
chemical system so if you’re having the
party later and and drinking too much
alpha you’ll know you’re only doing
brain research and people also suspect
there is a quantum mechanical system
seriously the computers that we built
today are about as far away from a brain
as an amoeba and even if we can build
one with 86 billion nodes it still
doesn’t have the chemical or quantum
mechanical system
third is people believe that AI equals
machine learning machine learning is one
algorithmic family and then most people
think that AI is actually not just
machine learning it’s deep learning so
only one type of algorithm imagine you
can only imagine one thing with your
brain this is what we call mail write
mail or do you think one thing so there
are lots of other people in the world
who can imagine many more things machine
learning is not the single thing than AI
there are many more algorithms like
machine reasoning like genetic
algorithms and the love of stuff it’s
very important to understand yet they
they can run any process if you want to
build an AI there’s four things you need
you need algorithm is not just one this
is like many you need data you need
talent and you need compute power
compute power is available you just get
this from the cloud algorithms are
available you can download them from the
internet data that’s a difficult topic
we just need to talk about and talent is
why you’re here
so data data is very important because
people think that data has value like
data is the oil which no data is used to
describe the world because a machine
arrives in the world and it’s empty it’s
not like your baby who already has an
opinion and so on and so forth it comes
completely empty so you have to give it
something that it understands what you
want it to do and don’t have to explain
like everything that’s data data has
nothing to do with the oil of the 21st
century if data is the oil of the 21st
century we are the Indians who know only
setting fire to the oil on High Holidays
no cars no plastic data is there to
describe the world to a machine and you
can collect data in two ways to do that
you can do it inside out you basically
sit inside the economy you look at the
data that is being produced and thus you
try to interpret the world because if
you know how to make toasters and who
buys toasters and if you know how to
make bread and who buys bread you kind
of probably will understand sooner or
later that people
use bread put it in a toaster and then
eat it because somehow it doesn’t come
back right you can also learn this if
you sit on the consumers shoulder like
Google and Facebook do and they tell you
that they toasted their bread and they
eating it and where they bought it and
all this kind of stuff it’s all there
just to describe the world this is
important and before you get crazy in
Europe here is data and AI and data
protection are not contradicting this
really doesn’t matter because there is
so much data around there you don’t have
to like steal it from everybody if
there’s no contradiction there I just
need to put this in because there are so
many people say no but we can’t do AI
hear because bla bla bla not true it’s
more question of thinking if you want to
introduce AI into a company and if you
really say like I want these 80%
efficiency because I need to compete I
want to thrive in this new world you
need to find one important starting
point you need to want to do it if you
only want AI for marketing I have a very
good suggestion you take out your
telephone you take 20 pictures you have
used at least 10 different algorithms in
the AI field you can now write in your
marketing brochures you’re doing AI put
the phone back you’re done it’s very
cheap if you really want the efficiency
out of AI you will have to get your act
together and want to do it and you will
have to introduce it step by step like
an onion you use AI to automate
something that generates a little more
data than it needs to automate the next
thing and the next and the next and so
on and you do that part with the data so
that everything you automate next is
faster than the one you started with
because if everything takes 2 years and
you have a hundred processes you’re done
in two hundred years I don’t think the
company will exist that long and only
after you’re started this start thinking
about your new business model and how to
use data to analyze it and so on one of
the large CEO asked me just last year
said like mr. Bo’s if I give you all my
data will be telling me my new business
model and I said no if you give me all
your data I will run 80% of your company
and you have time to think about your
new business
and this is a very important message
because people are creative they can
create new business models they can even
motivate others to turn around machines
can only execute 80% of our processes or
all processes that don’t use human human
communication can and will be run by a
machine that is putting our economy
upside down AI is the way to do it and
it’s going to be great for us because as
humans we’re finally gonna get back to
something that we can do well which is
being creative and being nice to each
other we’re not so great in working like
machines this is why in highly developed
company countries we get depressed there
are two ways of learning you can learn
from observation this is when you start
with the machine learning part where you
look and experiment or you can learn
because someone who already knows
teaches you learning from observation is
typically what companies do that have a
lot of very broad data learning from
someone who already knows is what
established companies do because they
already know something I hope the
interesting part is the learning from
what you already know has a 95 percent
energy advantage it needs 95 percent
less energy so the next time you hear
someone telling you oh I’m so sad all
these Valley companies are gonna run
over me that means it like you just
start using your advantage yours you are
like you’re supposed to die if you’re
not using a 95 percent advantage this is
very important so if we really want to
fulfill the promise of artificial
intelligence in the economy is probably
the only thing that can save the
established economy it means making the
experience that is stored in all of us
in all our companies executable thank
you very much you