artificial intelligence or the brainization of the economy
TRANSCRIPT
ARTIFICIAL INTELLIGENCE OR THE BRAINIZITATION OF THE ECONOMYPaul Bazin & Pierre-Eric Leibovici
“I think we should think of AI as the intellectual equivalent of a backhoe. It will be much better than us at a lot of things.”
G e o ff r e y H i n t o n
“With art ific ia l intel l igence we are summoning the demon.”
E l o n M u s k
"Once you start to make machines that are r ival l ing and surpassing humans with intel l igence, i t 's going to be very difficult for us to survive, i t 's just an inevitabi l i ty ."
C l i v e S i n c l a i r
“With more advances in art ific ia l intel l i -gence ahead, the need for human labor wi l l fa l l further”
L a r r y F i n k
“These k inds of amazing things that just 10 years ago were sc ience fict ion, are going to be very helpful everywhere.”
J e ff B e z o s
"F irst the machines wi l l do a lot of jobs for us and not be super intel l igent .That should be posit ive i f we manage i t wel l . A few decades after that though the intel l igence is strong enough to be a concern. I agree with E lon Musk and some others on this and don't under-stand why some people are not con -cerned."
B i l l G a t e s
" I f we succeed, we wi l l have turnedthe most awful paradigmthat we know on i ts head.The inevitabi l i ty of death."
L a r r y P a g e
“So the biggest thing that we ’re focused onwith art ific ia l intel l igence is bui ld ingcomputer serv ices that have betterpercept ion than people,so the basic human senses l ike seeing,hear ing, language, core things that we do.”
M a r k Z u c k e r b e r g
“We wi l l t ranscend al l of the l imitat ionsof our biology”
R a y m o n d K u r z w e i l
WHAT IS ARTIFICIAL INTELLIGENCE?
WHAT THEY THINK ABOUT AI
B r u n o M a i s o n n i e r is the founder of Aldebaran robot ics . After g iv ing birth of Nao, Pepper & Romeo, three companion robots , he stepped back from operat ion and decided to think about the future of robot ics and the next step.
What we call AI today is in fact a succession of predetermined rules. We call it AI because it does things faster than our brain.
60 years ago, John McCarthy used for the first time the term “Artificial Intelligence”. What does it mean and how has it evolved since 1956?
John McCarthy first defined it by “the science and engineering of making intelligent machines. A more precise explanation was made by Nils J. Nilsson in his book Quest for Artificial Intelligence “Artificial intelligence is that activity devoted to making machines intelligent, and intelligence is that quality that enables an entity to function appropriately and with foresight in its environment.”
He added that intelligence can be measured by the abilities “to reason, achieve goals, understand and generate language, perceive and respond to sensory inputs, prove mathematical theorems, play challenging games, synthesize and summarize information, create art and music, and even write histories.” (ibid)
However, the very nature of what we call Artificial Intelligence makes it nearly impossible to find a more technical and immutable definition. Indeed, as the AI100 (a committee whose mission was to study the evolution of AI during 100 years) explained, as soon as a new AI technology becomes a practice in itself, it stops being called AI. According to Kevin Kelly, co-founder of Wired magazine, “most of the AI won’t be very exciting […] AI will be everywhere, cheap, utilitarian, boring, like electricity today.“
How would you define artificial intelligence?
“By the fact we say Artificial intelligence it means that it is NOT intelligent. When Nao is walking, he is not artificially walking. He just walks.AI hasn’t become much more intelligent since the victory of Deepblue against Kasparof. The parameter that has evolved is that we don't use directly calcula-tion power, but huge calculation power to create a dumb neural network which seems intelligent.“ What are the main challenges for the next five years?
“To become intelligent, the computer needs to take decision by himself, he has to be autonomous. For now, he only executes a code that human implement-ed via neural networks. We should try to imitate animal’s intelligence as a start. And then make it evolves.“ Will AI look like Sci-Fi in the next 50 years?
“AI is an incremental curve. Nowadays, this curve flatten itself but it just needs a good idea to really start the exponential. A human brain (1000Hz for basic processes) is millions of times slower than a computer(TeraHz), thus we can imagine that when we'll succeed to make it truly intelligent the field of possi-bilities will wildly open.“
AI IS MOSTLY A LOT OF BIG WORDS
Internet Of Things (IoT): A network that allows every connectable device to communicate and interact. The key challenges for IoT are: (1) to have only one communication protocol, (2) improve batteries and (3) spray the network.
Algorithmic Game Theory is a multi-agent algorithm that calculates the equilibrium between the goal of the different agents, the global goal and set the optimal role of every agent.
Virtual Personal Assistants is a software that performs autonomously the tasks of a personal assistant.
Recommendation Engines and Collaborative Filtering is an algorithm that predict the task of costumers by knowing its previous experience. For instance, the Netflix algorithm is a Recommendation Engines and Collaborative Filtering
Supervised learning is like a black box where the algorithms are trained by giving them a lot of inputs and linked outputs in order for them to learn the rule.
Reinforcement Learning is a supervised learning but with only one example given. Like a child who burns itself, he only has to do it once to know that fire burns. Research are currently working on it.
Unsupervised learning is a black box at which we give only the inputs and the algorithm adapt itself to understand and give the good output without previous example. It is the next big step for AI but it has not been done yet.
Machine Learning is a tool for AI. It’s a type of algorithms that learn from data sets, improve and refine their results over uses.
Deep Learning is a type of machine learning algorithm that uses classifier neural network. Those algorithms are often used in video, audio and speech recognition.
Computer vision, through deep learning algorithms, is able to classify object, people, movement, situation, etc.
Robotic gives a robot the ability to learn to interact, in a most natural way, with its environment and people using its hardware capabilities.
Natural Language Processing (NLP) is used to interact with human. It’s born with the Turing test in 1950. The goal is to know if the user is speaking to a computer or another human. It is used for bot, voice recognition and translation.
Collaborative Systems is still at a research level, but the goal is to build autonomous systems that can work collaboratively between them and with human.
Crowdsourcing and Human Computation, close to Collaborative Systems, is an area where the computer should ask for human expertise if it is conscious that he cannot solve properly the problem.
Virtual Assistants
Speech recognition
Recommendation engines
Deep Learning
Reinforcement learning
Context aware computing
Gesture control
Video recognition
“Machine Learning focuses on the quest ion of how to get computers to program themselves. ”
T o m M . M i t c h e l l ,C a r n e g i e M e l l o n
COULD R2D2 BE MY BEST FRIEND?
Let’s imagine Daphni, a personal robot. When
you go to bed you will say to Daphni “Please,
wake me up at 7am” (NLP). When the alarm
rings, Daphni will turn on the coffee machine
(Internet Of Things), because it has learnt
that you like to take your coffee half an hour
after you wake up (Machine Learning).
When you are showering, Daphni will ask the
soap provider to give you the amount you
need, which is not the same amount needed
for your 10-year-old daughter (Collaborative
Systems). Then, Daphni will go to the kitchen
(Robotic), grab the cup full of coffee and give
it to you (Computer Vision) without adding
sugar, however, you used to put sugar in
your coffee but three days ago you asked
Daphni not to add sugar (Reinforcement
Learning). While you drink your coffee,
Daphni tells you about your daily meeting
(Virtual Personal Assistant) that he sched-
uled for you based on the free time slot you
have (Recommendation Engines and
Collaborative Filtering). But he hesitates for
the 5:00pm slot because you have the
possibility to both attend the football game
of your son and meet with your CEO.
Daphni then asks for your advice to make the
decision (Crowdsourcing and Human
Computation).
Before leaving your home, Daphni advises
you to take a different ride than every other
day because of the traffic to make you save
10 minutes (Algorithmic Game Theory).
Your autonomous car will follow Daphni’s
advice and take you safely to work (Deep
learning).
Even if this is what we picture when we think
of AI, it is not for an immediate future! It is
still impossible to contextualize every life
scene. We are not able to make a human
smart computer! To have a sufficient power
of calculation, the robot needs to be huge
and equipped of a big and noisy fan. Also,
the overall capacities of robots are still
limited. For instance, they are still not able to
climb stairs or open doors unless the house
is equipped.
“Now computer v is ion and speech recognit ion just work. They ’re not perfect , but they work. And that enables a lot of appl icat ions, which is why you see a l l th is exci tement around deep learning and A. I . Because now that i t works, there are so many doors that are open al l of a sudden.”
Y a n n L e C u n , D i r e c t o r o f F a c e b o o k A I R e s e a r c h ( F A I R ) , N o v . 2 0 1 5
THE POWER OF OPEN SOURCE IN AI
AI has become an unavoidable topic of discussion. Famous CEOs (Bill Gate, Elon Musk, Larry Page…), well-known scientists (Stephen Hawking, Ray Kurzweil, Tim Berners-Lee…) and politicians (White house, European Parliament…) are concerned about how Artificial Intelligence will evolve. In order to find some answer and prepare the future, several associations avec been launched all over the world by public & private researchers.
Amazon, Facebook, Google, IBM and Microsoft combine forces by sharing their work in the Partnership on AI.Elon Musk and Sam Altman founded OpenAI, a non-profit association for AI research. Stanford University has asked leading thinkers to study the AI evolution for the next 100 years through the AI100 annual report.
… TO THE BIG BANG AROUND THE COMBINATION OF BRAINIZATION AND SOLIDIFICATION
After the digitalization of the economy we will be the witnesses or the actors of its brainization. At this time, we will observe the solidification of the economy with the development of service robots. Their mission is to replicate mechanical functions realized by human being in a moving environment (not to be confused with industrialized robots that are in a closed environment). The autonomous cars are service robots. The service robots sector has not yet exploded since robots are without brains. AI enable the brainization of robots, and ultimately change our lives drastically as well as the organisation of work.
We have seen a lot of corporations that have missed the digitalization and are now struggling to survive. Brainization and solidification of the economy will have a similar or even more powerful impact for the ones that don’t see and anticipate its potential.
AI will become a commodity like electricity and Internet are today. The question which is difficult to answer is when will it be the case? When it appears, the value will rely on the services developed around AI and not on the technology itself
"The internet and the platforms that i t makes possible a l low very smal l groups of indiv iduals to make enormous profits whi le employing very few people. This is inevi table , i t is progress , but i t is a lso socia l ly destruct ive."
S t e p h e n H a w k i n g
How would you define artificial intelligence?AI is the art of teaching machines how to reproduce human tasks. Technically, we use machine learning to build AI systems. Right now, supervised learning is very successful: we train a neural net with lots of examples -- know inputs and outputs, for instances pictures of objects and their description -- and then the machine is capable to reproduce the task with inputs that were never seen before.
What are the main challenges for the next five years?The main challenge is to give machines some sort of common sense. For instance, if you want to do a perfect translation from Chinese to French, you really need to understand the cultures of both source and target languages. Babies learn common sense by exploring the world, trying different actions, observing how these actions impact the world. This is what we call "unsuper-vised learning".A step towards unsupervised learning could be reinforcement learning (RL), an approach that's very hot this year. With RL we try to take into account reward or punishment feedback in real time. For instance, a child does not need to get burnt 1000 times before he or she understands boiling water is hot -- but our classical supervised learning model do need that many examples before they can get it! Will AI look like Sci-Fi in the next 50 years?Probably not. Expectations created by Hollywood (from 2001 A Space Odyssey to Her) are incredibly high. We need to solve unsupervised learning and other issues to get there, and we don't even have a clear path to that as of 2017.I don't think true intelligence can be learnt from datasets. Humans "ground" their mental concepts on their own life experiences. A truly intelligent AI would need a physical body, in other words a robot, to do this -- but then the speed of learning would be too slow. We could give the AI a virtual body in a virtual world in order to accelerate the process (for instance, some teams initially train their autonomous driving models in the Grand Theft Auto game), but the virtual world will not be as realistic as the real one. You need truly intelligent AI to do simulate this word... that's a chicken and egg problem.
Alexandre Lebrun is a successful ser ia l entrepreneur and the former co-founder and CEO of Wit .a i that he sold to Facebook in 2015. He then jo ined the Facebook Art ific ia l Intel l igence Research (FAIR) group. He is passionate about language, and helping machines understand humans.
… TO THE BIG BANG AROUND THE COMBINATION OF BRAINIZATION AND SOLIDIFICATION
After the digitalization of the economy we will be the witnesses or the actors of its brainization. At this time, we will observe the solidification of the economy with the development of service robots. Their mission is to replicate mechanical functions realized by human being in a moving environment (not to be confused with industrialized robots that are in a closed environment). The autonomous cars are service robots. The service robots sector has not yet exploded since robots are without brains. AI enable the brainization of robots, and ultimately change our lives drastically as well as the organisation of work.
We have seen a lot of corporations that have missed the digitalization and are now struggling to survive. Brainization and solidification of the economy will have a similar or even more powerful impact for the ones that don’t see and anticipate its potential.
AI will become a commodity like electricity and Internet are today. The question which is difficult to answer is when will it be the case? When it appears, the value will rely on the services developed around AI and not on the technology itself
WHAT THEY THINK ABOUT AI
L u c J u l i a is the v ice president of innova-t ion at Samsung where he developed the ARTIK c loud. He is a lso the co-authored of Apple ’ S ir i . Luc is making real people 's l ives better us ing technologies.
How would you define artificial intelligence?
Ideally, AI is made to replace human intelligence, unfortunately we don’t know how a human brain works. Thus, we can’t make a modelling of it.Today, AI is mainly about calculation and task automation. It is doable to model a strategy for a particular type of reflexion in a particular domain such as Chess or the Go game which have limited combinations on a single board. What is not doable is building cross-domain algorithms that would understand without any ambiguity different situations in different contexts.
What are the main challenges for the next five years?
Beside the power of calculation that has increased tremendously in the next 25 years, we haven’t see any break through. Algorithms that are used today are the same of the ones used 25 years ago. Thus, I can’t be optimistic for the next 5 years.However, AI should focus on analysing data with so called deep learning algorithm. Deep learning is, once again, only due to computational power. 25 years ago you had to wait 1 day for your algorithm to run a 2 layers’ neural network. Nowadays, a 10 layers’ neural network gives its result in real time.Beside big data, research should focus on cross-domain by crossing two algorithms that work on separate domains.
Will AI look like Sci-Fi in the next 50 years?
No, or maybe in 5000 years. But I can’t imagine it coming in our era. Human has analytical abilities that computer has not and it’s not only a question of data. A Robot in a factory will be 10 times quicker than a human to do its tasks, a calculator will perform complex operations 1 million times faster, therefore, they look smarter in these areas of expertise but has no intelligence at all in other domains.If a robot has to take the human place it should know about many domains, and resolve cross-domains ambiguities and complementarity. That is not realistic.
WHAT COULD SLOWTHE EMERGENCE OF AIAS A STANDARD?
Science fictionThere is a disappointment caused by the state of current technologies. Everyone is imaging AI as the Sci-fi we see in Hollywood movies. And the excitement around it today makes the expectations grow. The reality is that we will not have Chappie as a best friend for a long time. As internet in the early 2000’s, the interest around AI will decrease.
Legal debateRegulation can cause complication for the AI growth in the next years. “Who is responsible when a self-driven car crashes or an intelligent medical device fails? How can AI applications be prevented from promul-gating racial discrimination or financial cheating? Who should reap the gains of efficiencies enabled by AI technologies and what protections should be afforded�to people whose skills are rendered obsolete?” (AI100)The legislation should learn to adapt itself and be reactive by anticipating the innovation. We can already measure the effects of a strong regula-tion. Fortunately, things are moving fast and the European Parliament is already debating on robots’ legal status.
“When there is an innovation America makes a business, China copies it and Europe regulates it.”Emma Marcegaglia president of ENI
It should not be seen as a threat but an opportunity. It is going to modify markets’ organization but the mutation will be progressive as it was with Internet and the digitalization of most of the industries. Some players will not have the mind-set to adapt but new players will become leaders in their domain of expertise. We are in the process of brainization.
A job thiefAI could backlash because of the fear of workers to be replaced by intelligent machines, according to Forester Research, 7% of US jobs will be replaced by AI by 2025. This replacement will surely lead to protests and thus political debates. In an historical point of view, innovation leads to productivity improve-ments so the jobs of tomorrow are still to be determined. As an example, the business of phone operators suddenly disappeared long ago. On the other hand, the automation and the adoption of the phone have increased the produc-tivity of other jobs.
“ I f an a l ien watched TV before invading Earth, i t would think that the world was fu l l of robots. In real i ty , today's robots are st i l l too stupid to be let loose.”
M a s s i m i l i a n o V e r s a c e C E O N e u r a l a
HOW FAST DOES IT GO?
WHO ARE THE ACTORS, WHERE ARE THEY
AND WHAT DO THEY DO?
In 2014, the US represented more than 50% of the AI investment.European investments represented only 10%, falling behind China at 15%.
Worth to mention is that the research was pursued by GAFA located in the US.
When asking AI companies what they do in terms of AI, one third of them answers machine learning. More than a fifth says that they use natural language processing and another fifth would say computer vision. The last one fourth is divided with virtual personal assistants, smart robots, context awareness…
In 2015, HealthTech represented 15% of all AI venture invest-ments, followed by adtech and business intelligence. Followed by well-being, virtual assistants, transport & robotics. The investors were less interested by education and agriculture.
FRENCH STARTUPS ARE INIn France, we count more than 200 startups that claim to use AI in France. If we assume the number of AI startups given by Venture Scanner (1589) to be correct, that would mean that 12.5% of AI startups are French.
We could add to this several other companies such as Carmat, a French company that creates artificial heart, that went public in 2010 and all the startups founded by French entrepreneurs across the world, such as: • Sentient Technologies ($ 143m funding) • CustomerMatrix ($ 16M funding)• IQ engines (Acquired by Yahoo)• Madbits (Acquired by Twitter)• …
EDUCATION
POD
HOLBERTONS SCHOOL
OC
TRANSPORT
NAUYA
XEE
MISTERFOX
AUTOKAB
ADTECH
DATABERRIES
ADOMIK
S4M
CRITEOL
COLLECTIVE
SKOVEO
RECAST.AI
CLUSTREE
CP
WELL BEING
REMINIZ
WITHINGS
WANDERCRAFT
SOMFY
VIRTUAL ASSISTANT
WIIDII
OTTSPOTT
JULIE DESK
ROBOTIC
ALDEBARANFYBOTSBALYO
PARROTImmersive Robotics
PROCESS
SHIFT TECHNOLOGYSCORTEXALKEMICS
CYBELANGEL
WHAT IS THE MARKET AND WHO PAYS FOR IT?
Tractica Research
GII Reserach
Statista
BB Research
Accenture
Techcrunch
Merril Lynch
38,56 %
53,65 %
59,42 %
60,3 %
61,22 %
102,34 %
159,95 %
$ 3,01 B
$ 5,05 B
$ 6,08 B
$ 6,24 B
$ 6,42 B
$ 20 B
$ 70 B
Source CAGR Valuation 2020
Because of the difficulty to define exactly what AI is, it is not possible to give a clear expected valuation of what the market will be in 2020 and in the future.The expectation for 2020 goes from $5.05bn for GII Research to $70bn for Merrill Lynch. Accenture expects the worldwide AI market to be valued more than $ 13Tn in 2035 with the US market ($8.3Tn) way ahead of Japan ($2.1Tn), Germany ($1.1Tn) and UK ($0.8Tn).
The Compound annual growth rate (CAGR) of the AI market is estimated to be 53.65% from 2015 to 2020 for a market valuation of $5.05B in 2020. (“AI Market report” by market-sandmarkets.com)
Based on the “AI Market report”, the AI market was valued at $590M for 2015. At the same time, investors have injected $ 2.4B in 397 AI startups (CBInsights).
When comparing the numbers, we see that the investments represent more than 4 times the valuation of the market. In other words, they are definitely betting on the future. The question is: how long will the AI market need to mature?
Thus, we can think that AI is a very promising sector, far from being mature and with a bright future!
NAME
11
17
5
30 +
cleversense
Moodstock
DeepMind
Jetpac DNNresearch
emu
Granata Dark Blue Labs
Vision Factory TimefulApi.ai
apteligent
Orbital Insight Inc
Clarifai
THE CLIMATE CORPORATION Kensho Rocana
Anomali MindMeld Kindred Framed
Scalyr
Datanyze
Farmers Business Network
urban engines Granular Ionic
Recorded Future
Saffron nervana systems itseez IQ Engines Movidius
Indisys
Embodied Incoming
Smartrip
PERFORMANCELAB
Prelert
BODY LABS
swre Xevo Cognitive Scale
api.ai savioke
emotient Fortscale Gotlt !
Perfant DataRobot DataRobot
bluedata Rithmio Prism Skylabs
MindMeld
MAANA ninebot LU Lumiata Sigfex Prafly
ChronocamPrecisionHawkWHOKNOWS
Reflektion
The CAGR of the entire Venture Capital market is estimated to be 35.86% from 2016 to 2020 ("Global Venture Capital Investment Market 2016-2020" report)
Considering the estimated growth speed of AI and VC markets, we can be sure that AI investments will take a much bigger part on the global venture market.
Why such a growth?
The field is shifting from simply creating systems that are intelligent to building intelligent systems that are human-aware, trustworthy and decision takers. Take a security system: instead of just triggering the alarm when there is a movement detected at an inappropriate hour, the system should recognize who it is based on the company’s organigram or social network and then decide by itself if it is an abnormal situation and what to do for each case. For instance, it would call the police and close the door if it is categorized as a robbery with a high accuracy.
The ecosystem is understanding the added value of AI in each sector. Such as the automation of customer services. Gartner Inc. predicts that, by 2020, 85% of customer interactions will be managed by machines.Therefore, we have seen trends such as bot companies in 2016 and VCs invest-ments are going along with these trends.
These investments are led by big actors, mainly from the US.
WHAT IS THE PLACE OF AI INVESTMENTS INTO THE VENTURE MARKET?According to Artem Burachenok, VCs have invested $704M in 79 AI startups on Q1 & Q2 2016.
On the same period, the global VC market financed 3 894 startups with $53.9Bn
That means VC investments in AI represent only 1.30% of the venture market.
Evolution of AI market7
6
5
4
3
2
1
02015 2016 2017 2018 2019 2020
Tractica Research GII Research Statista
BCC Research Accenture
NUMBER OF DEALS
NAMETYPE
CVC 30 +
25 +
18 +
15 +
INTEL
DATA
COLLECTIVE
KHOSLA
VENTURES
GV
VC
VC
CVC
Indisys
MAANA
WHOKNOWS
api.ai
Rithmio
incoming
PERFORMANCELAB
MindMeld Skully
Xevo Smartrip Swre
Prelert FORTSCALE
Gotlt! Perfant Savioke emotient
bluedata BODY X LABSPrafly
PrecisionHawk Chronocam ninebot
Sigfox DataRobot Reflektion
Embodied Cognitive Scale Lu lumiata
SIGOPT
BLUE RIVER
carsabi
DroneDeplay CloudMedx Nervana
We
TIMEFUL SI Vicarious Atomwise Zymergen
CAPPELA SPACE i l a fliptop LiftIgniter
Citrine Informatics kaggle Misocline
Verdigris Technologies
apteligent
clarifai
KINDRED
KENSHO Recorded Future
MindMeld Rocana THE CLIMATE CORPORATION
FRAMED SCALYR ANOMALI DATANYZE
Orbital Insight Inc
urban engines Granular
IONIC
Farmers Business Network
Blue River
MetaMind
Ayasdi Vectra Vectra Lookout Trueaccord
ThoughtSpot Vicarious CrowdMed
Timeful SI Kaggle Atomwise
Other players are as well investing a lot in AI, judging that it is the new technology shift: New Enterprise Associates, Plug and Play ventures, Horizons Ventures, Formation 8, Andreessen Horowitz, Accel Partners, GE ventures, Samsung Ventures, 500startups…
Name Investors N° rounds Total funding Last funding Location Industries
Kortschak Investments, L.P.
Data collective, SoftBank
Tencent
Floodgate, GE, IVP, Khosla Ventures, KPCB, U.S. Department of Defense
J.P. Morgan, Andreessen HorowitzIndex Ventures and Two Sigm
Access industries, Horizons ventures Tata Communication
16
3
1
3
7
2
$US2 100 000000,00
$US174 140 000,00
$US150 000000,00
$US143 800 000,00
$US106 350 000,00
$US105 000000,00
2017-01
2016-10
2016-04
2014-11
2015-03
2014-09
USA
USA
CHN
USA
USA
USA
#BigData #PredictiveAnalytics
Google, Alibaba 3 $US1 390 000000,00 2016-02 USA #AR #ComputerVision
Lux capitalDraper Fisher Jurveston
3 $US290 000000,00 2016-10 USA #AutonomousCar
Blackstone, Insight venture, DFJ growthFairhaven Capital, Khosla ventures
4 $US177 000000,00 2016-06 USA #CyberSecurity
#Robotics #BioInformatics
#BigData
NEA, BessemerGeorgian Partners
8 $US163 690 000,00 2016-10 USA #Healthtech
#Bigdata #HealthTech
#ProcessAutomation #AIAAS
#BigData #PredictiveAnalytics
#entertainment #Robotic
TOP 10 FUNDING BY PRIVATE INVESTORS IN THE AI STARTUPS
There are already big players that have developed flourishing
businesses around AI. And AI investments have also a specu-
lative aspect. Take Magic Leap funding for instance: it is the
second biggest AI investment of all time however, no one have
seen their technology. Investors have started to doubt their
capacities to deliver what they have promised.
MNCS DON’T WANT TO MISS THE BRAINIZATION TURN.
By 2018, Gartner predicts that most of the 200 largest companies in the world will use AI to exploit data, improve processes and better serve customers. To do so, Multinational Corporations (MNCs) have started to invest massively in young AI startups.
It goes by the acquisition of promising AI startups. With 11 acquisitions between 2011 and 2016, Alphabet is by far the leader in this area.
MNCs also create dedicated venture funds to have a piece of the AI revolution. Intel Capital is the second VC in the world to invest in AI, Google Ventures is fourth. On the tenth position, we find GE Ventures. Google and Intel are, by far, the most pro-active buyers and investors in the different AI sectors. With its $400M acquisition of Deepmind in 2014, Alphabet (at the time Google) has confirmed its big role to play on the AI growth. Facebook has created an AI division and has recruited Yann LeCun, a French Deep Learning expert to lead this revolution.
More than that Microsoft, through Microsoft Ventures, has invested in the Element AI incubator, an accelerator & research lab for AI startups based in Montreal connected to the world's best academic ecosystems.
In addition, Amazon announced that they give its AI blocks Rekognition, Polly and Lex to anyone that asks. Lex is the technology that powers Amazon Alexa, and allows developers to integrate rich conversational experiences in their offerings. Polly is a state of the art text-to-speech service that has forty-seven life-like voices in twenty languages. Rekogni-tion is an image processing service, that can identify content in images.
Toyota wants to invest $1B for the development of an autonomous car and production chain. Baidu, the Chinese giant is following the movement by launching a $200M corporate venture fund specialized in AI.
The governments of different states also want to join the race of AI. For instance, the government of South Korea government announced an investment plan of $800M in AI. The US plan to invest $4B.
Thus, big players have truly understood that AI is a revolution that can’t be missed. All the investments committed will make AI the standard of tomorrow.
GAFA collects data and Intel has the power of
calculation. By their investments they are
taking a tremendous advantage of the
resources that AI needs and it will be very
difficult for other players to catch up.
Most pessimists say that with the computer intelligence on their sides, GAFAs (and Chinese BATX) would be powerful enough to reverse the world political order!
ARTIFICIAL INTELLIGENCE: MOST ACTIVE CORPORATE BUYERS (1/2) 2011-2016YTD
NAME
IBM
11
17
5
30 +
5
4
0
3
1
cleversense
Moodstock
DeepMind
Jetpac DNNresearch
emu
Granata Dark Blue Labs
Vision Factory TimefulApi.ai
apteligent
Orbital Insight Inc
Clarifai
THE CLIMATE CORPORATION Kensho Rocana
Anomali MindMeld Kindred Framed
Scalyr
Datanyze
Farmers Business Network
urban engines Granular Ionic
Recorded Future
Saffron nervana systems itseez IQ Engines Movidius
Indisys
Embodied Incoming
Smartrip
PERFORMANCELAB
Prelert
Prelert
BODY LABS
swre Xevo Cognitive Scale
api.ai savioke
emotient Fortscale Gotlt !
Perfant DataRobot DataRobot
bluedata Rithmio Prism Skylabs
MindMeld
MAANA ninebot LU Lumiata Sigfex Prafly
ChronocamPrecisionHawkWHOKNOWS
Emotient Vocalia
Cognea Explorys AlchemyAPI
Perception TuplejumpTuri
Madbits TellApart Whetlab Magic Pony Technology
Reflektion
ARTIFICIAL INTELLIGENCE: MOST ACTIVE CORPORATE BUYERS (2/2)2011-2016YTD
IBM
EBAY
AOL
NOKIA
AMAZON
NICE
ORACLE
0
3
3
3
3
6
3
1
3
1
3
3
2
2
2
2
0
0
3
9
0
1
6
Cognitive scale PCSapi WayBlazer
Comfy
DigitalGenius Evariant
Hortonworks
Moneytree Introhive 6sense Insidesales.com
Insidesales.com
Hello Heart
Baixing
babel Cedexis Indix Rapidminer Moovit
Rocketfuel
Parracel Rachio
ElementAI Crowdflower Cognitive Scale Neura
Cognea
Equivio
Prediction IO
Lookflow
Wit.AI
Expert Maker
Sociocast
DESTI
ANGEL.AI Orbeus
Causata
Crosswize Palerra
Nexidia
Medio
Convertro Gravity
Sales Predict Hunch
Face.com
Skyphrase Indisys
MetaMindTempo
NETBREEZE GENEE
Explorys AlchemyAPI
Workfusion Sensoro Netpulse
DAPHNI WON’T MISS IT EITHER
In 1996 it was considered disruptive to launch a project within the sector of Internet. In 2016 what is disruptive is not to create a company in the Internet sector but to launch an insuretech AI, an agritech AI …
We think AI is a buzzword that is not properly used. AI is a tool and not an end. It is a resource for startup to develop new services that will disrupt manyf industries. However, lot of entrepreneurs claim to have a technology that works whereas they back their algorithms with cheap labours. We don’t want entrepreneurs to promise a technology while having nothing.
The five pillars of our investment thesis: 1. No snob or show-off. AI is a buzzword. Too many projects declare they are developing AI assets while they are only working on basics algorithms 2. No Algorithm 20% better than the one developed by the competition. Indeed, AI research are open source and led by GAFA. Competitive advantage won’t last against the powerful open source consortium. 3. Automation of rebarbative tasks/ data mining tasks will mutate thanks to AI. For instance, the law firms have not really yet digitalized their business. It should happen rapidly with the emergence of AI. Part of their value is to search into data and former cases and find a solution adapt to the assumptions of the case they have to deal with. This task could be optimized thanks to AI. 4. Owning of proprietary data set and strong sector expertise. The belonging of tremendous amount of data (coming from internet and sensors that have starting to appear everywhere) as well as of a strong sector expertise on a vertical market enable the development of new disruptive services and new usages. 5. AI will give service robots a brain. Service robotics will revolutionize numbers of sectors. Logistic and transportation market are good example of this mutation.
We are convinced that AI will become a new standard and not the way Bill Gate, Elon Musk and the others have said. It will automate rebarbative tasks, personalize products, prevent energy waste, increase security, create new jobs and so on so forth…AI is not a threat but an opportunity to cease and we won’t miss the brainization turn of the economy. The similarity of AI today and Internet 20 years ago are too obvious for being ignored. The market is growing fast and we think it is a good time to invest in good technology that disrupt specific verticals.
WHAT THEY THINK ABOUT AI
How would you define artificial intelligence?“Artificial intelligence consists in computer programs that emulates aspects of animal behavior and competences in software or hardware. Among AI, Neural Networks and Deep Networks are a sub-field that is enjoying the best results among all AI techniques due to their ability to more closely emulate brain processes and robustness. “
What are the main challenges for the next five years?“AI and Deep Networks today are laser-focused on individual competences, such as visual perception, speech recognition, navigation, motor control, to name a few. But, real intelligence uses senses that work together. Today's mobile robots, drones, and self-driving cars need advanced and, more impor-tantly, coordinated capabilities in perception and mobility to be effectively 'put to work' in complex environments. To date, the best implementations of these capabilities in a "single package" come from biology. The challenge for AI is to recognize that the unit of AI is a “brain” rather than its individual competence. “
Will AI look like Sci-Fi in the next 50 years?“The feeling for humans is that they “won’t be alone anymore”: there will be powerful brains in every device that will render our environment richer and more interactive. And smarter. “
Massimi l iano Versace is the co-founder and CEO of Neurala Inc. a company that emu -lates the human brain funct ion in software. He founded the Boston Univers i ty Neuro -morphics Lab where he has pioneered the research of Deep Learning and Art ific ia l Neural networks.
L a r r y P a g e
AT DAPHNI WE BELIEVE IN THE EUROPEAN NATURAL COMPETITIVEADVANTAGE OVER THE OMNIPOTENT US
Focus where Europe has natural advantage
Different from those of the US
Collective Quality of life Inventive
Productiviy Empowerment Entertainment
100 STARTUPS THAT BRINGS AI TO LIFE: This tables Give you information on who are the startups, the investors, where is the money invested in AI and what are the industries. On these 100 startups there are 64 from the US, 12 from the UK and 9 from France. That reflects the predominance of USA on AI ecosystem. On the investment part we have 66 funding, 29 acquisition and only 3 IPO.
B12 funding
Bit Stew Systems
Acquisition
Blippar funding
Vulcan Capital,
USVP Gannett CO7
Salesforce, Bain CapitalBattery, Venrock 3
Fenox Venture Capital, Horizons
KPCB, Myrian Capital4
daphni 1
IBM /
Softbank /
Mitsubishi, SMBC, D4V 3
Mayfield Fund, NorwestNVP 2
Amazon /
J.P. Morgan, Andreessen
HorowitzIndex Ventures and Two Sigm
2
Bain Capital, SSM Partners 5
KKR 1
Floodgate, GE, IVP, Khosla
Ventures, KPCB, U.S. Department of
Defense
7
General Catalyst Partners 1
GE /
Qualcomm ventures
Khazanah nasional
3
66 ,790,000 $ 2004 2014-08 USA
36,00,000$ 2013 2015-07 USA
33,720,000$ 2009 2016-05 USA
4,300,000$ 2015 2016-11 FRA
undisclosed 2005 2015-03 USA
100,000,000$ 2005 2012-03 FRA
2 ,800,000$ 2015 2017-01 JPN
13,160,000$ 2011 2016-11 USA
Undisclosed 2015 2016-09 USA
105,000,000$ 2010 2014-09 USA
36,090,000$ 2009 2016-05 USA
55,000,000$ 1995 2014-10 GER
106,350,000$ 2008 2015-03 USA
12,400,000$ 2015 2016-07 USA
153,000,000$ 2005 2016-11 USA
99,000,000$ 2011 2016-03 UK
#Adtech
#PredictiveAnalytics
#BigData #Analytics
#HealthTech
#Agriculture
#bot
#Robotics
#fintech #Trading
#bigData #HealthTech
#Bot
#entertainment #Robotic
#Healthtech
#ProcessAutomation
#AIAAS
#BigData
#PredictiveAnalytics
4info funding
6sense funding
Affectiva funding
Agricool funding
AlchemyAPI acquisition
Aldebaran
RoboticsAcquisition
Alpaca funding
analyticsMD funding
Angel.ai acquisition
Anki funding
Apixio funding
Arago funding
Ayasdi funding
#Aiaas
#IoT #SmartGrid
#Edtech
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
Butterfly
Networkfunding
Carmat IPO
ClarifAI funding
Clark funding
Cognea acquisition
Comma.ai funding
Conversica funding
Cortica funding
CouldMinds funding
Criteo IPO
Customer
Matrixfunding
Cylance funding
Aeris Capital
Jonathan M. Rothberg1
Public Valuation
USV, Menlo Ventures
Qualcomm2
Seven ventures
Axel Springer2
IBM /
A16Z, 1
Kennet Partners
Toba Capital
Horizons ventures
4
Softbank, Hon Hai
Precision Industry Co. Ltd.
Walden International
Keytone Ventures
2
Public Valuation
HSCB, Aster Capital 3
Blackstone, Insight
venture, DFJ growth
Fairhaven Capital, Khosla
ventures
4
100,000,000$ 2011 2014-11 USA
67,700,000$ 2008 2016-02 FRA
40,000,000$ 2013 2016-10 USA
14,750,000$ 2015 2016-08 GER
undisclosed 2013 2014-05 USA
3,100,000$ 2015 2016-04 USA
22,000,000$ 2007 2015-12 USA
37,900,000$ 2007 2014-03 USA
31,000,000$
$
2015 2016-01 USA
1,700,000,000 2005 2013-10 FRA
16,000,000$ 2013 2016-01 USA
177,000,000$ 2012 2016-06 USA
#HealthTech
#HealthTech
#ComputerVision
#Fintech #InsurTech
#bot
#AutonomousCar
#Bot
#ComputerVision
#Robotics
#Adtech
#BigData
#PredictiveAnalytics
#CyberSecurity
EBAY
AOL
NOKIA
AMAZON
NICE
ORACLE
3
1
3
1
3
3
2
2
2
2
0
0
3
9
0
1
Hortonworks
Hello Heart
Baixing
babel Cedexis Indix Rapidminer Moovit
Rocketfuel
Parracel Rachio
Lookflow
Wit.AI
Expert Maker
Sociocast
DESTI
ANGEL.AI Orbeus
Causata
Crosswize Palerra
Nexidia
Medio
Convertro Gravity
Sales Predict Hunch
Face.com
Skyphrase Indisys
Workfusion Sensoro Netpulse
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
Dark Blue
Labsacquisition
Darktrace funding
Datarobot funding
Deepgram funding
DeepMind Acquisition
Deepomatic funding
Google /
KKR & Co., Summit Partners 3
IA Ventures, Intel Capital
NEA,TechStars4
Ycombinator, Compound 2
Google /
Alven Capital 1
undisclosed 2014 2014-10 UK
104,500,000$ 2013 2016-07 UK
57,420,000$ 2012 2016-02 USA
1,920,000$ 2012 2016-09 USA
600,000,000$ 2012 2014-01 UK
1,000,000$ 2012 2015-09 FRA
#DeepLearning #bigdata
#CyberSecurity
#PredictiveAnalytics
#NLP #AudioRecognition
#AIAAS
#BigData
#PredictiveAnalytics
Defined
Crowdfunding
Digital
Reasoning
Systems
funding
DNN
ResearchAcquisition
Drawbridge funding
fundingDreamQuark
drive.ai funding
Amazon, Microsoft
Accelerator
Sony, Swan Venture
1
FinTech Innovation Lab,
Goldman Sachs,
In-Q-Tel, Lemhi Ventures,
6
Google /
Sequoia Capital
Northgate Capital 3
///
Northern Light VC
Oriza Venture 1
1,100,000$ 2012 2016-09 POR
73,960,000$ 2012
2012
2012
2016-05 USA
Undisclosed 2013-03 CAN
45,500,000$ 2016-05 USA
5,000,000$ 2012 / FRA
12,000,000$ 2012 2016-03 USA
#BigData
#PredictiveAnalytics
#BigData
#PredictiveAnalytics
#NLP
#ComputerVision
#VoiceReco
#Adtech
#BigData
#PredictiveAnalytics
#AutonomousCar
looking for
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
Eversight funding
Face.com Acquisition
Face++ funding
Graphcore funding
Gumgum funding
Emergence Capital
Sutter Hill Ventures2
Facebook /
China’s Innovation Works
Ignition Partners, Qiming
Venture
C4 ventures, Samsung, Robert Bosch VC 1
Upfront ventures, NEAFirst round, Morgan Stanley
5
24,200,000$
$
2012 2016-04 USA
60,000,000
$
2012 2012-06 ISR
48,000,000 2012 2015-05 CHN
30,000,000$
2016 2016-10 UK
36,830,000$ 2007 2015-05 USA
#BigData #Analytics
#Retail
#ImageRecognition
#ComputerVision
#Hardware
#Adtech
H2O.ai funding
Capital One Growth,
Nexus Venture
Paxion Capital,
Transamerica Ventures
4
33,600,000$
2011 2015-11 USA #BigData
#PredictiveAnalytics
icarbonx funding
Inbenta funding
Indisys Acquisition
IQ Engines Acquisition
Acquisition
Acquisition
Acquisition
Jetpac
LookFlow
Madbits
Magic Leap funding
Tencent1
Level Equity
InverSur Captital
3
Intel /
Yahoo /
Google /
Yahoo /
twitter /
Google, Alibaba 3
150,000,000$
2015 2016-04 CHN
13,370,657$
2005 2016-04 USA
26,000,000$
$
2005 2013-09 ESP
Undisclosed
Undisclosed
Undisclosed
2008 2013-08 USA
2011 2014-08 USA
2009 2013-10 USA
undisclosed 2013 2014-07 USA
1,390,000,000 2011 2016-02 USA
#Bigdata #HealthTech
#NLP
#NLP #bot
#AR #ComputerVision
#ComputerVision
#ComputerVision
#ComputerVision
#DataBase
#AR #ComputerVision
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
Mist Systems funding
Mobvoi funding
MoneyFarm funding
Movidius Acquisition
Nara Logics funding
Navya funding
Netra funding
Nexidia acquisition
Nutmeg funding
nuTonomy funding
Oben funding
Oculus acquisition
Orbeus acquisition
Ozlo funding
Palantir funding
GV, Cisco investments 2
Google, SIG China, Sequoia, Zhenfund 3
Allianz Ventures, Cabot Square Capital 3
Intel /
406 ventures, Peter de Roetth 1
360 Capital Partners 3
NXT ventures, Launchpad Venture group
1
NICE Systems /
Pentech ventures, Convoy
InvestmentTaipei Fubon bank
4
Samsung
Highland Capital Partners
2
CrestValue Capital 2
Facebook /
Amazon /
Greylock, AME cloud 1
Kortschak Investments, L.P. 15
42,400,000$
2014 2016-10 USA
71,620,000$
2012 2016-07 CHN
29,810,000$
2011 2016-09 UK
400,000,000$
2006 2016-09 USA
13,000,000$
2010 2014-10 USA
38,020,000$
2014 2016-10 FRA
2,470,000$
2013 2016-07 USA
135,000,000$
2000 2016-01 USA
89,830,000$
2010 2016-12 UK
19,600,000$
2013 2016-05 USA
7,700,000$
2014 2016-11 USA
USA2012 2014-03
undisclosed 2012 2015-12 USA
USA
USA
14,000,000$
$
2013 2016-05
1,990,000,000 2004 2015-02
#IoT #VoiceRecognition
#Analytics
#Fintech #WealthManagment
#ComputerVision
#AIAAS
#AutonomousCar
#ComputerVision
#AudioVideoRecognition
#Fintech
#WealthManagment
#AutonomousCar
#IoT #VR
#AR #ComputerVision
#ComputerVision
#Bot #BigData
#PredictiveAnalytics
Maluuba acquisition
Metamind acquisition
Microsoft /
Salesforce /
undisclosed 2011 2017-01 CAN
32,800,000$ 2014 2016-04 USA
#NLP
#ImageRecognition
$ 2,000,000,000
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING
PathwayGenomics
funding
Paxata funding
PredictionIO acquisition
Preferred Networks
funding
Ravelin funding
Recast.AI funding
Rocket Fuel IPO
Saffron technology
acquisition
Scaled
Inference
funding
Sentient Tech. funding
SkyPhrase acquisition
SkyTree funding
IBM 1
Accel Partner, Intel,
Microsoft EDB investments
4
Salesforce /
Toyota, Fanuc 3
Playfair Capital
Amadeus Capital 3
Kima ventures & Bas 1
Public Valuation
Intel /
Khosla ventures 2
Access industries, Horizons ventures Tata Communication
3
yahoo /
Scott McNealy, UPS, USVPJavelin Venture American Express
3
40,000,000$ 2008 2016-01 USA
60,990,000$
2012 2016-10 USA
undisclosed 2013 2016-02 USA
JPN
5,640,000$
2014 2016-09 UK
1,120,000$
2015 2016-06 FRA
942,320,000$
2008 2013-09 USA
undisclosed 1999 2015-10 USA
13,600,000$
2014 2014-10 USA
143,800,000$
2007 2014-11 USA
undisclosed 2011 2013-02 USA
20,500,000 $
2012 2013-04 USA
#Healthtech
#BigData
#PredictiveAnalytics
#BigData
#PredictiveAnalytics
#IoT
#Fintech #FraudDetection
#Bot
#Adtech
#ProcessAutomation #BigData
#AIAAS
#ProcessAutomation
#AIAAS
#NLP #BigData
#BigData #PredictiveAnalytics
17,300,000$ 2014 2015-12
Wit.ai Acquisition
X.AI funding
Zero zero
robotics
funding
Zoox funding
Zymergen funding
Facebook /
Two Sigma Ventures,
FirstMark Capital
IA ventures, SoftBank
Capital
Two Sigma Ventures,
FirstMark Capital
IA ventures, SoftBank
Capital
3
IDG, GSR Ventures
ZhenFund & ZUIG
2
Lux capital
Draper Fisher Jurveston
3
Data collective, SoftBank 3
undisclosed 2013 2015-05 USA
34,300,000$
2014 2016-04 USA
25,000,000$
2014 2016-04 CHN
290,000,000$ 2014 2016-10 USA
174,140,000$ 2013 2016-10 USA
#bot
#PersonalAssistant
#drone #ComputerVision
#AutonomousCar
#Robotics
#BioInformatics
#BigData
STARTUP TYPE INVESTORS TOTAL FUNDINGN° ROUNDS FOUNDED ON COUNTRY INDUSTRYLAST FUNDING