2. Basic definitions in the world of AI
Agent: system that perceives environment and takes actions to maximize its chance of success.
• Can perceive it’s environment and has a current state, a goal, action repertoire and utility function.
• Agent can be placed in any kind of device, e.g. robot, computer etc.
• With today’s algorithms, is likely to struggle in situations where multiple goals may exist together.
Artificial intelligence: intelligence exhibited by machines and/or software.
• Goal can be to either ‘model humanity’ or ‘surpass humanity’
• Artificial General Intelligence (AGI) is complete brain modeling, Narrow AI (NAI) is per functional column
• Also the word for a multi-expertise field that includes coding, psychology, mathematics, philosophy etc.
Intelligence: the ability to learn and solve problems.
• Often means holding a hierarchical model of the world that allows predicting future states, rewards etc.
• Given complexity of the world only imperfect models can exist *
• Multiple forms of intelligence might exist (computing based, memory based etc.)
Machine learning: when computer learns from ‘experience’ without explicitly being programmed.
• Computer set up to learn from examples via huge datasets (10M examples needed for human-level P)
• Two main functions apparent: feature identification and prediction (classification/regression)
• Many forms of machine learning exist (supervised, unsupervised, reinforcement learning)
Learning: the acquisition of knowledge/skills and long term change in human disposition/capabilities.
• Consists of updates to the model of the world we hold internally
• In biological systems: neurons that grow/strengthen/prune connections
• In artificial systems: mathematical weights between ‘nodes’ in neural networks
Sources used: Superintelligence (Bostrom, 2016); www.neuralnetworksanddeeplearning.com, Prediction Machines (2018);
Artificial Intelligence (Russell & Norvig, 2016); Deep Learning (Goodfellow, Bengio; 2018); Wissner-Gross, Equation for Intelligence
* not having a model seems possible, as shown in behavior of certain animals
3. The AI startup market generally looks bright for the coming years...
3,222
4,819
7,345
11,284
17,268
25,996
37,987
53,231
70,972
89,847
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
2016 2017* 2018* 2019* 2020* 2021* 2022* 2023* 2024* 2025*
ExpectedrevenueinmillionsUSD
CAGR: 43%
Source: Statista
The total expected revenue in the AI market is expected
to reach ~90B in 2025...
1,739
3,477
4,569
6,255
15,242
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
2013 2014 2015 2016 2017
VenturefundinginmillionUSD
CAGR: 72%
...so there is a funding ‘boom’ happening at the
moment for AI firms
* Forecast
4. ...with computer vision, algorithmic trading, and healthcare revolutions
as the main focus areas for firms
8,098
7,541
7,366
4,680
4,201
3,714
3,656
3,567
3,170
2,473
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000
Static image recognition (classification, tagging)
Algorithmic trading strategies
Efficient scalable processing of patient data
Predictive maintenance
Object identification (detection, classification, tracking)
Text query of images
Automated geophysical feature detection
Content distribution on social media
Object detection and classification (navigation)
Better cybersecurity
Forecasted cumulative value, in millions of USD, 2016-2025
Source: Tractica
5. Utilizing the power of AI in existing organizations can create trillions of
USD in value in marketing/sales and supply-chain management
Customer service management
Next product to buy
(individual offering)
Price and promotion
Customer
acquisition/lead
generation
Churn
reduction
Channel
management
Marketing
budget
allocation
Predictive maintenance Yield optimization
Procurement
and spend
analytics
Inventory and
parts
optimization
Logistics
network
and
warehous
e
optimizati
on
Sales and
demand
forecast
Workforce
productivity and
efficiencty
Predictive
service/intervention
Task automation
Product
developmen
t cycle
Fraud and
debt
analytics
Risk
Product
feature
optimization
Smart capex
Analytics-driven
accounting and IT
Analytics-
driven HR
Total value potential of enhanced analytics
in marketing/sales is 3,3-6,0T USD...
...and 3,6-5,6T USD for supply-chain
management and manufacturing...
...where other areas will see
smaller gains
Source: McKinsey MGI
6. China and the US are currently leading in the AI race, fragmented
Europe is behind
London, 211
Paris, 73
Tel Aviv, 189
Shanghai, 77
Beijing, 150
Tokyo, 99San Francisco, 596
Los Angeles, 73
New York, 180
Boston, 102
Number of AI startups, top-10 cities
Source: Roland Berger & Asgard: Artificial Intelligence – A strategy for European startups (2018)
7. This disbalance becomes even more apparent when we look at startup
funding
Top-50 AI firms when it comes to funding (2017)
0
100
200
300
400
500
600
700
SenseTime
Face++
Upstart
Affirm
UBTECHRobotics
FlatironHealth
Zoox
CrowdStrike
ZestFinance
InsideSales,com
Mobvoi
Cybereason
NAUTO
Darktrace
Anki
Zymergen
C3IoT
CloudMinds
DataRobot
RecursionPharmaceuticals
VicariousSystems
SoundHound
Braincorporation
PreferredNetworks
Graphcore
Petuum
ShapeSecurity
ElementAI
Cambricon
Liulishuo
Endgame
babylon
CerebrasSystems
Appier
Afiniti
Freenome
OrbitalInsight
Drive,ai
Trifacta
Algolia
WorkFusion
TempusLabs
Invoca
Onfido
Versive
Conversica
CrowdFlower
MOOGsoft
Captricity
NarrativeScience
MillionsUSDfunding
United States
China
United Kingdom
Japan
Taiwan
Canada
Source: CB Insights; Next Generation Artificial Intelligence Plan (Chine’s Ministry of Industry and
Information Technology, 2017); Roland Berger & Asgard: Artificial Intelligence – A strategy for European
startups (2018)