Вы находитесь на странице: 1из 107

Daniele Gianotta

Lena-Marie Pätzmann
Timo Andre Gemkow
Philipp Hachmann

Management in the Digital Economy


Artificial Intelligence Enterprise Use Cases
- Lecture of December 6th 2018-
Artificial Intelligence Enterprise Use Cases
Team – Group of Master Students
Daniele Giannotta Timo Gemkow

«I was interested in how the basics of «It is fascinating to see how Artificial
Artificial Intelligence work and how Intelligence radically transforms
companies are trying to implement this traditional businesses in Switzerland.»
new technology.»

Lena-Marie Pätzmann Philipp Hachmann

«Uncovering the power of AI in real life «Speaking with people that deal with AI
use cases was definitely my favourite in their everyday work life was eye-
part of the research.» opening and enriching.»

1
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

2
Artificial Intelligence Enterprise Use Cases
Goals for Today: After this Lecture, you will…
…know the most important terms and functions of applied
Artificial Intelligence.

…gain insights in some real life use cases of applied Artificial


Intelligence.

…be able to identify the potential difficulties and issues of


implementation of AI.

3
Artificial Intelligence Enterprise Use Cases
Warm Up

What comes up to your mind when


you think about Artificial Intelligence?

4
Artificial Intelligence Enterprise Use Cases
Google Duplex

https://www.youtube.com/watch?v=znNe4pMCsD4

Source: Google, 2018


5
Artificial Intelligence Enterprise Use Cases
Google Duplex

Source: Google, 2018


6
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

7
Artificial Intelligence Enterprise Use Cases
Theory: Basics and Important Terms

Artificial Intelligence

Machine
Learning

Deep
Learning

Neural Networks

Source: Own Work


8
Artificial Intelligence Enterprise Use Cases
Theory: Basics and Important Terms

Artificial Intelligence

LEVEL OF STRENGTHS Machine Learning


Agents

Deep Learning TYPES OF LEARNING TYPES OF ANALYTICS


Narrow AI General AI Super AI

Artificial Neural Networks Unsupervised


Supervised learning Reinforced learning Predictive Prescriptive
learning

Convolutional Neural Network Recurrent Neural Network

Source: Own Work


9
Artificial Intelligence Enterprise Use Cases
Agents

Agents are software engines


instances that are able to
operate independently of
direct human input, as
opposed to regular software,
which needs human input.

Source: Fraenklin & Graesser, 1997


10
Artificial Intelligence Enterprise Use Cases
Machine Learning
Machine-learning algorithms detect patterns and learn how to make
predictions and recommendations by processing data and experiences,
rather than by receiving explicit programming instruction.
The algorithms also adapt in response to new data and experiences to
improve efficacy over time.

Source: McKinsey, 2018


11
Artificial Intelligence Enterprise Use Cases
Types of Analytics

TYPES OF ANALYTICS

Predictive Prescriptive

Source: Own Work


12
Artificial Intelligence Enterprise Use Cases
Types of Analytics

Source: McKinsey 2018


13
Artificial Intelligence Enterprise Use Cases
Types of Learning

TYPES OF LEARNING

Supervised learning Unsupervised learning Reinforced learning

Source: Own Work


14
Artificial Intelligence Enterprise Use Cases
Supervised Learning
An algorithm uses training data and
feedback from humans to learn the
relationship of given inputs to a given output
(e.g., how the inputs “time of year” and
“interest” predict housing prices).

You know how to classify the input and the


type of behaviour you want to predict, but
you need the algorithm to calculate it for
you on new data.

Source: McKinsey, 2018


15
Artificial Intelligence Enterprise Use Cases
Supervised Learning - Example
A mail inbox is able to identify an e-mail and classify them into regular and
junk e-mails.
“Regular e-mail” and “junk e-mail” are the two given outputs (categories)
that need to be classified.
Incoming e-mails Regular e-mails
Classifier
E-Mail 1 E-Mail 2

E-Mail 2 E-Mail 4
Junk e-mails
E-Mail 3 E-Mail 1

E-Mail 4 E-Mail 3

Source: Own Work


16
Artificial Intelligence Enterprise Use Cases
Unsupervised Learning

An algorithm explores input data


without being given an explicit output
variable (e.g., explores customer
demographic data to identify patterns).
You do not know how to classify the
data, and you want the algorithm to
find and classify the data for you.

Source: McKinsey, 2018


17
Artificial Intelligence Enterprise Use Cases
Unsupervised Learning - Example
A mail inbox needs to classify a large amount of mails without any given
output. Therefore, the mail inbox is free to define the outcome-categories by
identifying the type of mail by itself.

Incoming e-mails Work


E-Mail 3
E-Mail 1 Private
Classifier
E-Mail 2
E-Mail 2 University
E-Mail 4
E-Mail 3 E-Mail 1

E-Mail 4
Junk e-mail
E-Mail 5 E-Mail 5

Source: Own Work


18
Artificial Intelligence Enterprise Use Cases
Reinforced Learning
An algorithm learns to perform a task
simply by trying to maximize rewards it
receives for its actions (e.g., maximizes
points it receives for increasing returns
of an investment portfolio).
You don’t have a lot of training data;
you cannot clearly define the ideal end
state; or the only way to learn about
the environment is to interact with it.

Source: McKinsey, 2018


19
Artificial Intelligence Enterprise Use Cases
Reinforced Learning - Example
Imagine a child that has to learn how to walk on a specific surface. If
the child learns to walk, e.g. learning how to stand up, how to stand
still and how to take a few steps, it will receive a reward (for example
chocolate). If the child doesn’t learn how to walk, the reward is
negative (no chocolate).

Source: Analytics Vidhya,2017


20
Artificial Intelligence Enterprise Use Cases
Deep Learning
Deep learning is a type of machine learning that can process a wider
range of data resources, requires less data preprocessing by humans,
and can often produce more accurate results than traditional
machine-learning approaches (although it requires a larger amount
of data to do so).
In deep learning, interconnected layers of software-based calculators
known as “neurons” form a neural network. The network can ingest
vast amounts of input data and process them through multiple layers
that learn increasingly complex features of the data at each layer.

Source: McKinsey, 2018


21
Artificial Intelligence Enterprise Use Cases
Deep Learning

Source: McKinsey, 2018


22
Artificial Intelligence Enterprise Use Cases
Deep Learning

Artificial Neural Networks

Convolutional Neural
Recurrent Neural Network
Network

Source: Own Work


23
Artificial Intelligence Enterprise Use Cases
Convolutional Neural Network (CNN)
A multilayered neural network with a special architecture designed
to extract increasingly complex features of the data at each layer
to determine the output.
It is used, when you have an unstructured data set (e.g., images)
and you need to infer information from it.

Source: McKinsey, 2018


24
Artificial Intelligence Enterprise Use Cases
Convolutional Neural Network (CNN)
The convolutional neural network receives an image (e.g. the
letter A) that it processes as a collection of pixels. In the hidden
layers, it identifies unique features – for example, the individual
lines that make up an “A”. At the End, the CNN can classify a
different image as the letter “A” if it finds in the unique features
previously identifies as making up the letter.

Source: McKinsey, 2018


25
Artificial Intelligence Enterprise Use Cases
Recurrent Neural Network (RNN)
A multilayered neural network that can store
information in context nodes, allowing it to
learn data sequences and output a number or
another sequence.

It is used, when you are working with time-


series data or sequences (eg, audio recordings
or text).

Source: McKinsey, 2018


26
Artificial Intelligence Enterprise Use Cases
Recurrent Neural Network (RNN)
A recurrent neural network neuron receives a command that indicates the start of
a sentence. The neuron receives several words and then outputs a vector of
numbers that feeds back into the neuron to help it “remember” that it received
them once before. After receiving the last word (“free”), the neuron assigns a
probability to every word in the English vocabulary that could complete the
sentence. If trained well, the RNN will assign the word “tomorrow” one of the
highest probabilities and will choose it to complete the sentence.

Source: McKinsey, 2018


27
Artificial Intelligence Enterprise Use Cases
Level of Strengths

LEVEL OF STRENGTHS

Narrow AI General AI Super AI

Source: Own Work


28
Artificial Intelligence Enterprise Use Cases
Level of Strengths

Source: Zetaris, 2018


29
Artificial Intelligence Enterprise Use Cases
Level of Strengths
Speculation/ Goal
Intelligence

Current research
(Currently, there is no GAI) Super Artificial Intelligence
AI is ahead

Industrial use General Artificial Intelligence


(AI applicable for several issues)

Narrow Artificial Intelligence


(AI used for specific issues/ processes)

Time

Source: Buxmann & Schmidt, 2018


30
Artificial Intelligence Enterprise Use Cases
Rule Based Systems
Rule-based systems (or expert systems) are used in AI applications and
rely on a set of pre-defined rules to answer questions related to a
specific problem or topic, which means that their ability to learn is
limited.
They are are a simple kind of artificial intelligence, which for example
use a series of IF-THEN statements that guide a computer to reach a
conclusion or recommendation

Source: Buxmann & Schmidt, 2018


31
Artificial Intelligence Enterprise Use Cases
Some Applications of AI
Natural Language Processing NLP aims to analyse and understand natural (spoken or written) human language by
utilizing rules and algorithms. Due to the complexity of human language, large
amounts of data are needed to train the algorithm
Machine Vision Machine Vision is the term for training robots and machines to see things like humans
do
Image Recognition A subcategory of machine vision, that allows computer technologies to
recognize humans, animals etc. through the use of machine learning and
algorithms. This is frequently done with the help of convolutional neural
networks to filter images through various neuron layers.
Pattern Recognition Pattern recognition is the automated recognition of regularities in datasets.
There are different approaches to pattern recognition, this can be either be
done by machine learning or by hand-crafted/not-learned rules.
Classifier Classifiers are statistical functions that enable the classification of units into classes
based on specific parameters or characteristics.

Source: Buxmann & Schmidt, 2018


32
Artificial Intelligence Enterprise Use Cases
Natural Language Processing
Natural language processing includes many different
techniques for interpreting human language:
• Content categorization
• Topic discovery and modelling
• Contextual extraction
• Sentiment analysis
• Speech-to-text and text-to-speech conversation
• Document summarization
• Machine translation

Source: SAS, 2018


33
Artificial Intelligence Enterprise Use Cases
AI-News Anchor in China

https://www.youtube.com/watch?time_continue=1&v=Cjn_ZJQ-e0Y

Source: CNBC, 2018


34
Artificial Intelligence Enterprise Use Cases
AI-News Anchor in China

Source: CNBC, 2018


35
Artificial Intelligence Enterprise Use Cases
AI-News Anchor in China
China’s state-run Xinhua News Agency has
launched an artificial intelligence news anchor, a
move it claims to be a world first.
Modeled on the agency's Zhang Zhao presenter,
the new anchor learns from live videos and is able
to work 24 hours a day, reporting via social media
and on the Xinhua website.

Source: CNBC, 2018


36
Artificial Intelligence Enterprise Use Cases
Machine Vision
Machine vision (or computer vision), a rapidly growing branch of AI that aims
to give machines sight comparable to our own, has made massive strides
over the past few years thanks to researchers applying specialized neural
networks to help machines identify and understand images from the real
world.

Source: VentureBeat, 2015


37
Artificial Intelligence Enterprise Use Cases
Image Recognition

The way image recognition works,


typically, involves the creation of a
neural network that processes the
individual pixels of an image.
Researchers feed these networks as
many pre-labelled images as they
can, in order to “teach” them how to
recognize similar images.

Source: TheNextWeb, 2018; Buxmann & Schmidt, 2018


38
Artificial Intelligence Enterprise Use Cases
What do you see Here?

Visual-semantic alignment

Alignment model learns to


associate images and
snippets of text using a
recurrent neural network.

Source: Stanford Computer Science, 2018


39
Artificial Intelligence Enterprise Use Cases
Pattern Recognition

Pattern recognition is the ability


to detect arrangements of
characteristics or data that yield
information about a given system
or data set.

Source: TechTarget, 2018


40
Artificial Intelligence Enterprise Use Cases
Classifier
Classifiers are statistical functions that enable the classification of units
into classes based on specific parameters or characteristics.

Source: TechTarget, 2018


41
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

42
Artificial Intelligence Enterprise Use Cases
Group Work

Now it’s your turn!


Please analyse the artificial intelligence
use cases

43
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

44
Artificial Intelligence Enterprise Use Cases
Why does AI matter?
Banking Sector

Source: McKinsey, 2018


45
Artificial Intelligence Enterprise Use Cases
Gartner Technology Hype Cycle

Source: Gartner, 2018


46
Artificial Intelligence Enterprise Use Cases
Deployment of AI Initiatives in 2018
Deployment of AI Initiatives in 2018
No Interest;
14%
Already have
No action invested and
planned; 35% deployed; 4%
Artificial
Intelligence

In short-term
planning; 25%
In medium- or
long-term
planning; 21%
Source: Gartner, 2018
47
Artificial Intelligence Enterprise Use Cases
The Status Quo of AI
Survey Results among German
Companies
100%
90% 14% Companies make use of
80% 33% Artificial Intelligence
70% 60%
60%
50%
40% 86%
30% 67%
20% 40%
Companies that do not
10% use Artificial Intelligence
0%
< €100 mn € 100mn - €1bn > €1bn
Annual Revenue

Do not use Use AI

Source: Technische Universität München, 2017


48
Artificial Intelligence Enterprise Use Cases
Microsoft AI Report

Significant increase in
European AI investment over
the course of the last 10
years.
Graph shows the number of
AI companies invested into
and total transaction volume
of $10.5bn.

Source: Microsoft, 2018


49
Artificial Intelligence Enterprise Use Cases
Where AI is Being Used Right Now Globally

Source: McKinsey, 2018


50
Artificial Intelligence Enterprise Use Cases
AI in Switzerland at a Glance

Only 4% 61%
of the companies are Percentage of companies
actively using AI in many that are still only in the
processes and to enable planning or piloting stages
advanced tasks

57% 55%
Expect AI to have a high Respond that AI is
impact on ‘business areas considered ‘an important
that are entirely unknown topic’ on the executive
today’ management level

Source: Microsoft, 2018


51
Artificial Intelligence Enterprise Use Cases
AI is an Important Topic on Management Level

Source: Microsoft, 2018


52
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

53
Artificial Intelligence Enterprise Use Cases
AI Impacts 4 Business Domains

Engage your Enable employees Transform your Optimize your


customers products & services operations
E.g. provide customers E.g. increase employee E.g. speed up product E.g. improve planning and
advice, shorten efficiency through innovation cycles, reduce costs through
conversion cycles, and predictions, enabled enable new value add intelligent prediction,
reduce time to support, and services, and provide operational efficiency, and
resolution automation of real time support deep insights, predictive
repetitive tasks maintenance

Source: Microsoft, 2018


54
Artificial Intelligence Enterprise Use Cases
Methodology/Research – How did we do it (1)

Identifying Use Contacting Conducting Extracting &


Cases Company Reps Interviews Analyzing Content

Desk research LinkedIn In person Recordings


Articles E-Mail Via phone Slides
Studies Notes

Source: Own work


55
Artificial Intelligence Enterprise Use Cases
Methodology/Research – How did we do it (2)

56
Artificial Intelligence Enterprise Use Cases
Methodology/Research – How did we do it (3)

57
Artificial Intelligence Enterprise Use Cases
Use Case - NZZ

58
Artificial Intelligence Enterprise Use Cases
Use Case - NZZ

Source: Presentation NZZ, 23.10.2018


59
Artificial Intelligence Enterprise Use Cases
Media Consumption Has Drastically Shifted
Towards Digital

Source: Presentation NZZ, 23.10.2018


60
Artificial Intelligence Enterprise Use Cases
Revenue Distribution

Source: Presentation NZZ, 23.10.2018


61
Artificial Intelligence Enterprise Use Cases
Focus on Conversion

Reg-Gate Pay-Gate

Source: Presentation NZZ, 23.10.2018


62
Artificial Intelligence Enterprise Use Cases
Conversion Increase: Paywall
Classic Metered Paywall/Regwall (2012 -2014):

Source: Presentation NZZ, 23.10.2018


63
Artificial Intelligence Enterprise Use Cases
Conversion Increase: Personal Greetings
Personalization (2015 -2017):

Source: Presentation NZZ, 23.10.2018


64
Artificial Intelligence Enterprise Use Cases
AI Solution: Personalized User Journeys (1)
Version 1: Dynamic Paygate (2nd half of 2017)

User moves on site


Anonymous user User receives
User registers
comes to site smart paywall

Paywall is placed
The users
movements are
tracked and
analyzed

Source: Presentation NZZ, 23.10.2018


65
Artificial Intelligence Enterprise Use Cases
AI Solution: Personalized User Journeys (2)
Version 1: Dynamic Paygate – improving rules engine

Data is
collected
only when
user is on
website!

Source: Presentation NZZ, 23.10.2018


66
Artificial Intelligence Enterprise Use Cases
AI Solution: Personalized User Journeys (3)
Example 1: Pattern recognition for pricing preferences

Source: Presentation NZZ, 23.10.2018


67
Artificial Intelligence Enterprise Use Cases
AI Solution: Personalized User Journeys (4)
Example 2: Pattern recognition for call-to-action

Source: Presentation NZZ, 23.10.2018


68
Artificial Intelligence Enterprise Use Cases
AI Solution: Personalized User Journeys (6)
Version 2: Dynamic Paygate– propensity scoring (2018)

Source: Presentation NZZ, 23.10.2018


69
Artificial Intelligence Enterprise Use Cases
Example: Propensity Scoring with ML

Source: Presentation NZZ, 23.10.2018


70
Artificial Intelligence Enterprise Use Cases
AI Solution: Future Outlook

Version 3: Paygate – Omnichannel and Retention (2019)

Source: Presentation NZZ, 23.10.2018


71
Artificial Intelligence Enterprise Use Cases
Results (1)
1) More than doubled the share of 2) Increased Conversion Rate from 0,5%
digitally to over 2,5%
active subscribers

Conversion Conversion
rate: rate:
0,5% > 2,5%

Source: Presentation NZZ, 23.10.2018


72
Artificial Intelligence Enterprise Use Cases
Results (2)
3) Consistently beating peer group in paid circulation

Source: Presentation NZZ, 23.10.2018


73
Artificial Intelligence Enterprise Use Cases
General Thoughts on AI from Interview

Media Industry
Media industry is relatively far advanced when it comes to new technologies (AI, etc.)

NZZ is frontrunner in DACH-region


Large US media companies like the Wall Street Journal or NYT are global front runners
in terms of AI implementation, but within the DACH-region, NZZ is one of the most
advanced AI users

Source: Interview NZZ, 23.11.2018


74
Artificial Intelligence Enterprise Use Cases
AI Solution: Summary
Industry Media & Newspaper

Goal Increase Customer Conversion Rate

Context Placement of a Paywall for potential customers to buy articles

AI Approach Personalized placement of the paywall based on the users movement and
actions

Type of AI Supervised Machine Learning; Customer Segmentation

Source: Interview NZZ, 23.1


23.11.2018
75
Artificial Intelligence Enterprise Use Cases
Use Case - Geberit

76
Artificial Intelligence Enterprise Use Cases
Use Case - Geberit

Products:
→ Sanitary Appliances

→ Installation Systems

→ Drainage Systems

Source: VR Room, 2017


77
Artificial Intelligence Enterprise Use Cases
Context

Geberit covers warranty for 30 years


for any product and has to provide
spare parts

The plumbers that install and replace


these spare parts have to be able to
visually identify the currently built-in
part to be able to order the correct
replacement.

Source: Interview Geberit, 14.11.2018


78
Artificial Intelligence Enterprise Use Cases
AI Solution: Pro-App – Image Recognition

Source: Geberit, 2018


79
Artificial Intelligence Enterprise Use Cases
Image Recognition: Training

Source: Own work; Interview Geberit, 14.11.2018


80
Artificial Intelligence Enterprise Use Cases
Problem

After training the


algorithm, Geberit discovered that the
pictures that were sent in by plumbers
posed a challenge to the AI for a few
reasons.

The pictures were so diverse in terms


of lighting, the angle, color and
bathroom furniture, that the algorithm
struggled to correctly assign a picture to
a previously learned „product“.

Source: Own work; Interview Geberit, 14.11.2018


81
Artificial Intelligence Enterprise Use Cases
General Thoughts on AI from Interview

Lack of trust
Within customer service there is lacking confidence that a chatbot or a talking
"agent" can successfully satisfy customers.

Few disruptive AI solutions in the industry


For internal process optimizations, there are still very few disruptive AI solutions
on the market that justify themselves economically

Source: Interview Geberit, 14.11.2018


82
Artificial Intelligence Enterprise Use Cases
AI Solution: Summary
Industry Sanitary Appliances

Goal Ease the replacement process for plumpers

Context Geberit has to provide spare parts for sanitary appliances that are up to 30
years old. Not all plumpers working for Geberit are able to safely identify all
products of the past 30 years.

AI Approach Geberit’s „Pro-App“ is a self-programmed mobile application that enables


plumbers to identify a product by taking a photo with the app. The app passes
on the photo to an algorithm that is able to use image recognition to accurately
identify which part a plumber is looking at.
Type of AI Supervised Machine Learning, Image Recognition

Source: Interview Geberit, 14.11.2018


83
Artificial Intelligence Enterprise Use Cases
Use Case – Axa Winterthur

84
Artificial Intelligence Enterprise Use Cases
Use Case – Axa Winterthur

Source: AXA, 2018


85
Artificial Intelligence Enterprise Use Cases
Context
Strategy Problems

We are a data-driven company

Feedback of the employees/users:


“Dashboards are good and simple. But
We have a SPOT – single-point-of-truth analysis and deepdive are not really
used because I don’t feel comfortable
with the tools ... “

We have many analytic tools with


dashboards etc.

Source: Interview AXA, 21.11.2018


86
Artificial Intelligence Enterprise Use Cases
AI Solution: Siri-Bution – NLP

Translate your business challenges into a simple question

Which customers should I


visit in the next 4 weeks?”

Get your answers using


the best visualization
within seconds

Source: Interview AXA, 21.11.2018


87
Artificial Intelligence Enterprise Use Cases
AI Solution: How it Works

https://www.youtube.com/watch?v=UM8GnBgE2U8

88
Artificial Intelligence Enterprise Use Cases
AI Solution: How it Works

https://www.youtube.com/watch?v=UM8GnBgE2U8

89
Artificial Intelligence Enterprise Use Cases
User Feedback

Source: Interview AXA, 21.11.2018


90
Artificial Intelligence Enterprise Use Cases
Use Case – Axa Winterthur

91
Source: Interview AXA, 21.11.2018
Artificial Intelligence Enterprise Use Cases
General Thoughts on AI from Interview

Collaboration with start-ups


Collaboration with start-ups & rapid prototyping is key

AXA is advanced in some areas


AXA is extremely advanced in terms of AI in certain areas, but lags behind in
others

Source: Interview AXA, 21.11.2018


92
Artificial Intelligence Enterprise Use Cases
AI Solution: Summary
Industry Insurance

Goal Ease the use of data and empower employees to make deep dives

Context Sales people do not know how to access data and use it, except for the simple
dashboards. Deep dives are too complicated and so are the tools which are
available, which leads to people not using the dashboard at all.
AI Approach The idea behind the project was to create a chatbot that can be asked
questions about the underlying data. This means that a user can type in a
question in natural language into a query field, the AI will analyze the question
and transform it into a data access request and ultimately presents the user
with a visual result.
Type of AI Supervised Machine Learning, Chatbot w/ Data Access

Source: Interview AXA, 21.11.2018


93
Artificial Intelligence Enterprise Use Cases
Why Did We Show You these Use Cases?

Example of how a generic Example of how an AI Example of how a great AI


problem can lead to an AI Solution developed in a Solution is developed out
Use Case solution that is completly „lab“ might not be ready of a formal innovation
internally programmed. for „reality“. roadmap in collaboration
with external start-ups.
AI maturity at
entire company

Source: Own work


94
Artificial Intelligence Enterprise Use Cases
Use Cases – Key Takeaways

Many large enterprises in The solutions that are being Managers across all interviewed
Switzerland have deployed deployed create measurable companies believe that AI will
functional AI solutions and efficiency gains and added have a tremendous impact in
/ or are working on new ones value the next few years

Source: Own work


95
Artificial Intelligence Enterprise Use Cases
Where Do Our Interviewees Stand?

high

AI maturity
compared to peer
group

low

low AI maturity overall high

Source: Own work


96
Artificial Intelligence Enterprise Use Cases
A Swiss Frontrunner
In size, we are probably the second largest AI
250 employees company in Switzerland – right behind Google.
in a dedicated AI unit

Source: Interview Swisscom, 28.11.2018


97
Artificial Intelligence Enterprise Use Cases
AI Implementation – Framework

Source: Own work


98
Artificial Intelligence Enterprise Use Cases
Top 3 Business Risks Associated with AI in CH

1 Regulatory Requirements
Without clear regulations, investments in AI can be perceived as risky for companies because they may invest in
something allowed at the time that may not be later on.

2 Impact on Personnel
AI is not just about technology but equally about change management, including culture and mindset shifts. It
requires balancing employees’ fear of losing their jobs with the awareness that once AI frees them from menial
tasks, they will gain more time to help customers and develop products and services.

3 Information Overload
Information overload may occur when trying to figure out what is useful in vast amounts of data, identifying
insights or establishing predictive trends. Working with such large amounts of data requires consideration for the
users looking for understandable, actionable results.

Source: Microsoft, 2018


99
Artificial Intelligence Enterprise Use Cases
Agenda
1 Introduction

2 Theory and Basic Concepts of Artificial Intelligence

3 Interaction and Group Work

4 Current State of Artificial Intelligence

5 A Selection of Real Life Use Cases in Swiss Companies

6 Wrap Up, Outlook & Feedback

100
Artificial Intelligence Enterprise Use Cases
Food for Thought

“In some cases, there is too much hype, but paradoxically, the
potential opportunities and benefits of AI are still, if anything,
under-hyped. Often, the impact of new technologies is
overestimated in the short term and underestimated in the long
term, and while there is a lot of noise regarding AI, there’s been
a lack of in-depth discussion and analysis of how it’s
actually going to transform businesses.”
- Nigel Duffy, Global AI Innovation Leader, EY

Source: Microsoft, 2018


101
Artificial Intelligence Enterprise Use Cases
Outlook: How AI will Change the Business World

• Improved business intelligence


• Understanding, targeting and serving customers
• Changing the way social media is used
• Introduction of customer responsive products
• Enhanced fraud prevention abilities
• Constant customer support
• Improvements in efficiency & reduced

Source: Forbes, 2018


102
Artificial Intelligence Enterprise Use Cases
AI Experiment – Google Teachable Machine

Click here to start the experiment

Source: Google, 2018


103
Artificial Intelligence Enterprise Use Cases
Feedback

Give us a short and honest feedback about our lecture!


Your feedback is anonymous.
Click here! Thank you very much!

104
Artificial Intelligence Enterprise Use Cases
Sources
Presentation NZZ, 23.10.2018. Artificial intelligence applied in real life – Personalized user journeys @NZZ.
Interview NZZ, 23.11.2018.
VR Room, 2017. Mit VR ins Badezimmer. http://vr-room.ch/2017/03/22/mit-vr-ins-badezimmer/
Interview Geberit, 14.11.2018.
Geberit, 2018. https://www.geberit.ch/dienstleistungen/geberit-proapp-smartphoneapp/
AXA, 2018. Homepage. https://axa.ch/de/privatkunden.html
Interview AXA, 21.11.2018.
Interview Swisscom, 28.11.2018.
Microsoft, 2018. Artificial Intelligence in Europe - Switzerland
Google, 2018. https://ai.googleblog.com/2018/05/duplex-ai-system-for-natural-conversation.html
Technische Universität München, 2017. Status Quo und Trends von künstlicher Intelligenz in der deutschen Industrie.
Gartner, 2018. https://www.gartner.com/smarterwithgartner/5-trends-emerge-in-gartner-hype-cycle-for-emerging-technologies-
2018/
Gartner, 2018. https://www.gartner.com/smarterwithgartner/2018-will-mark-the-beginning-of-ai-democratization/

105
Artificial Intelligence Enterprise Use Cases
Sources
McKinsey, 2018. An Executive‘s guide to AI. https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/an-executives-guide-to-ai
McKinsey, 2018. AI adoption advances, but foundational barriers remain. https://www.mckinsey.com/featured-insights/artificial-intelligence/ai-adoption-advances-but-
foundational-barriers-remain?cid=other-eml-alt-mip-mck-oth-1811&hlkid=4b798c19aba540ba9a782656e79e1c63&hctky=2147670&hdpid=bb08451a-dbdb-4103-
a88c-e9f1d270445e
McKinsey, 2018. The executive‘s guide to AI. ‘https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/the-executives-ai-playbook?cid=other-
eml-alt-mip-mck-oth-1811&hlkid=488f58c1fffa4d15a4767c74e3af91c3&hctky=2147670&hdpid=41d49d6e-b7fe-4c11-9fd2-254d7e9f2b1b
Stanford University, 2018. https://cs.stanford.edu/people/karpathy/deepimagesent/
CNBC, 2018. https://www.cnbc.com/2018/11/09/the-worlds-first-ai-news-anchor-has-gone-live-in-china.html
Google, 2018. Experiments with Google. https://experiments.withgoogle.com/teachable-machine
Analytics Vidhya, 2017. Simple Beginner’s guide to Reinforcement Learning & its implementation. https://www.analyticsvidhya.com/blog/2017/01/introduction-to-
reinforcement-learning-implementation/
Zetaris, 2018. AI & Data Preparation – Avoiding “Garbage-In, Garbage-Out”. https://zetaris.com/ai-data-preparation/
Buxmann & Schmidt, 2018. Künstliche Intelligenz – Mit Algorithmen zum wirtschaftlichen Erfolg. Berlin: Springer.
SAS, 2018. Natural Language Processing – What it is and why it matters. https://www.sas.com/en_id/insights/analytics/what-is-natural-language-processing-nlp.html
VentureBeat, 2015. Why machine vision is the next frontier for AI. https://venturebeat.com/2015/12/05/why-machine-vision-is-the-next-frontier-for-ai/
TheNextWeb, 2018. A beginner’s guide to AI: Computer Vision and image recognition. https://thenextweb.com/artificial-intelligence/2018/07/18/a-beginners-guide-to-
ai-computer-vision-and-image-recognition/
TechTarget, 2018. Pattern Recognition. https://whatis.techtarget.com/definition/pattern-recognition

106

Вам также может понравиться