Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Bachelor Project submitted for the degree of Bachelor of Science HES in International Business Management by Kerim TRUNIGER Bachelor Project Mentor: Christophe COURBAGE, Professor Geneva, 21st August 2020 Haute école de gestion de Genève (HEG-GE) International Business Management
62
Embed
Legal and ethical constraints in the use of Artificial Intelligence ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Legal and ethical constraints in the use of
Artificial Intelligence (AI) by Swiss health
insurance companies
Bachelor Project submitted for the degree of
Bachelor of Science HES in International Business Management
by
Kerim TRUNIGER
Bachelor Project Mentor:
Christophe COURBAGE, Professor
Geneva, 21st August 2020
Haute école de gestion de Genève (HEG-GE)
International Business Management
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER i
Disclaimer
This report is submitted as part of the final examination requirements of the Haute école
de gestion de Genève, for the Bachelor of Science HES-SO in International Business
Management. The use of any conclusions or recommendations made in or based upon
this report, with no prejudice to their value, engages the responsibility neither of the
author, nor the author’s mentor, nor the jury members nor the HEG or any of its
employees.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER ii
Acknowledgements
This project has been completed with the support of various persons. I, therefore, wish
to thank them for their assistance and encouragement.
First, I would like to thank my mentor, Mr. Christophe COURBAGE, for his helpful
supervision, guidance, and advices throughout the entire preparation of this work and for
providing me with helpful sources and contacts.
Then I am also grateful to Ms. FLUECKIGER, Director New Technologies & Data at the
Geneva Association, Mr. IRIA, Head of Digital Transformation at La Mobilière and to the
three other persons working in the insurance field for accepting to provide me with
interesting and helpful answers to my interviews.
Finally, I also would like to thank Mr. KOCH, the Innovation Scout Outpost Silicon Valley
and the current Head of services, Underwriting and Legal Department at La Mobilière for
the discussions we had and for putting me into contact with various people of the
insurance industry, and Ms. MORALES, legal counsel at Barclays for her guidance
throughout the legal framework of this project.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER iii
Executive Summary
Experts in new technologies and economists seem to agree: robotics and deep learning
by machines will change most of the economic activities in the coming decades, at a
very fast pace. If in the past, technological changes mainly brought about gradual
increases of efficiency in human work; robotics and the availability of big data, used with
Artificial Intelligence (AI) will bring about exponential change to the working world of
humanity, in the very near future. It seems almost impossible to predict with certainty
what the social, economic, and psychological effect on humankind will be. Simply
learning from the experience of past technological evolutions is not enough. Many
experts foresee more of a disruption rather than a further, gradual development.
Among various other economic actors, health insurance companies may find many
benefits by introducing and developing digital technologies such as AI. However, they
must consider different limits and risks, which can restrain them in the development of
AI. One of those limitations is the complex legal framework that regulates data use and
protection, particularly when the available massive data and sources include the
personal data of individuals. Furthermore, in order to preserve their trustworthiness and
reputation, they must comply with ethical boundaries to retain their current customers
and acquire new ones.
Only then will health insurance companies be able to take full advantage of the benefits
of these new technologies and obtain valuable return for their sizeable investments in
such new tools.
This research has the ambition to produce recommendations to Swiss health insurance
companies on how to best achieve balanced, profitable returns through the gradual
development and use of AI, within the legal and ethical limits. It further considers the
important hurdles and barriers which the introduction and expansion of such new
technologies will meet, both internally, with their customers and in the large public, and
provide some recommendations which need to be taken into account in the change
management process.
A thorough background study and examination of empirical data obtained from insurance
experts and staff will provide a solid basis to formulate some main recommendations in
how to deal with the trust of customers, the education of the staff, with legal and ethical
limits and the necessary required adjustment of governance processes.
Legal and ethical constraints of the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER iv
Contents
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss
health insurance companies ...................................................................................... 1
Disclaimer ..................................................................................................................... i
Acknowledgements ..................................................................................................... ii
Executive Summary ................................................................................................... iii
Contents ..................................................................................................................... iv
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 15
companies from realizing secondary uses of data. Imagine that a company has collected
user data and later discovers a new and valuable use for it. Under the current legal
regulations, the company would need to seek approval from every user before adopting
the data for this new purpose. While the intent of the legislation is sensible, its application
in the Big data world may greatly hinder the benefits that could be realized. Additionally,
the greater details of big data may allow users to be reidentified from anonymized data,
potentially revealing sensitive information in the process. The current tools, either legal
or technical, are already proving to be ineffective, and as we move further down the road,
they may become obsolete. (Mayer-Schönberger, Cukier 2013)
2.1.4 Ethical limits
Technological advances are strongly shifting the function of data in the insurance
industry and the general role of insurance itself. They are promising to bring major socio-
economic advantages. However, they are also bringing challenges regarding
confidentiality, privacy of personal information, protection of data, and individualization
of insurance. (Keller, Eling, Schmeiser, Christen, Loi 2018)
Intelligent systems are omnipresent. Concerns about the ethical implications in the
privacy and data protection areas are, therefore, increasingly flourishing among analysts,
policymakers, and the society at large. Most of the concerns refer to the capacity or the
lack of ability of intelligent machines to make ethical decisions by nature. Ethical debates
relate to the assessment of what is good and bad, acceptable, or right and wrong, it also
relates to lawfulness, fairness, integrity, and social accountability. Ethical interpretations
are therefore defined as having the capacity to determine, evaluate, and establish ethical
assertions based on various viewpoints. Due to their rising competencies, autonomy,
and communication abilities, intelligent systems are increasingly expected to act as
moral agents, meaning that they can judge what is right and wrong and can be held
responsible for their performances. When interacting with intelligent systems, users
expect them to have identical obligations and responsibilities as when interacting with
human beings. This brings up concerns relating to accountability and questions the ability
of AI systems to perform in accordance with human principles and to adhere to human
rights. (Dignum 2019)
As mentioned earlier, AI works only if provided with large amounts of data. When
applying AI, insurance companies, therefore, use the personal data of their customers.
This may lead to concerns related to confidentiality and protection of data. The first
issues linked to these concerns relate to fairness and discrimination. Fairness can be
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 16
compromised in insurance due to the profiling of the insured and cause discrimination.
Here, the term discrimination refers to a situation in which the insured that should be
treated equally, are not. It also refers to a situation in which the insured that should have
a different treatment are treated the same. To avoid such discrimination, specific
characteristics such as gender, ethnic group, sexual preferences, etc. should be
disregarded. In the insurance context, the meaning of discrimination is an underlying
dilemma. When the insured may receive a treatment based on its personal risk, this may
result in disadvantages for protected groups, when having a greater risk than the
average. Contrarily, if the insured are not treated based on their personal risk, the
classification of risk may be seen as unfair since it treats all the insured equally, even if
their risk is not comparable or irrelevant as the personal factors are not taken into
consideration. As the World Health Organization states, health is a fundamental right.
(WHO 2017) This dilemma cannot be resolved just by excluding discriminatory data
characteristics such as gender or ethnic group etc., because any intelligent system would
quickly gather those characteristics from other sources such as online media, that freely
permits access to personal information. A further issue linked to the concerns about
privacy and data protection is about intrusiveness into the right of informational
independence. Privacy is generally seen as a fundamental right such as in Article 8 of
the European Convention on Human Rights. (Council of Europe 2013) Insurance
companies that massively process data and use computerized decision-making may be
intrusive in the right of self-determination of persons. The usual corporate models that
are based on the monitoring of computer science, reward various behaviors that are
considered good and penalize those considered bad according to the insurance
corporation. They, therefore, can decrease risks by provoking a change in behaviors or
choices in ways of living of individuals. However, they could be seen as intrusive and
impeding an independent decision-taking of persons. This may cause issues if persons
cannot bear the cost of insurance-coverage for high-risk profiles and are, therefore,
limited in their life choices. (Keller, Eling, Schmeiser, Christen, Loi 2018) Manipulation of
behaviors using personal information is a further ethical issue. The important amount of
exchanges and the profound knowledge about persons provided by AI and Big data
results in having the insured exposed to misleading and behavioral manipulations.
(Müller 2020)
Other ethical concerns are arising such as the individualization of insurance. With the
shifting function of data, the premium of the customers is now established based on their
individual risk profile and no longer upon their classification in a particular risk group.
This shift is raising different kinds of concerns. One of them is that high-risk individuals
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 17
might be rejected from insurances or may have to pay unaffordable premiums. Another
concern is related to the fundamental principle of insurance which is solidarity. This
principle may be deteriorated due to an increasing individualization of insurance that
might also jeopardize the main objective of an insurance which is risk pooling. (Keller,
Eling, Schmeiser, Christen, Loi 2018)
Last January, digitalswitzerland, a Swiss initiative and association that aims to enhance
the position of Switzerland as leader in innovation, launched the Swiss Digital Initiative
(SDI) in Davos. With the fast evolution of computer science and intelligent data-based
systems, the objective of this initiative is to assist the development and application of
guidelines, facilities, and procedures for the implementation of ethical principles for
companies on a worldwide scale. (DFF 2020) In Switzerland there are no legally binding
ethical limits to be respected. However, for the good of their label, companies that use
private data from their customers may maintain and promote a good reputation by
respecting an ethical code of conduct provided by external experts when processing
data, as for example at La Mobilière with the Ethical Code for Data-Based Business by
the Swiss Alliance for Data-Intensive Services. 12 (Ethics - Innovation - data-service-
alliance.ch 2020)
12 Interview with José Iria, Head of Digital Transformation at La Mobilière, Geneva, 08 July 2020
(cf appendix 2)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 18
3. Analysis
3.1 Methodology
This research used diverse approaches to gather sufficient data to reach the set
objectives. It used both primary and secondary data.
First, the current and future AI potential in this sector, as well as it’s legal and ethical
limits, have been analysed and compiled based on extensive review of available, second
source literature. Secondary data have been further gathered from literature and
guidelines produced by personal data protection specialists and institutions.
This compilation provided a general understanding of the different current knowledge
around this topic and enabled the structuring of the background that highlights the
existing use of AI in the insurance industry and the advantages, risks, legal and ethical
limits related to the use of such a technology.
In the second part of the research, primary qualitative data has been gathered by means
of a questionnaire, from five managers working in major Swiss insurance groups. These
written interviews included Isabelle Flueckiger, Director New Technologies & Data at the
Geneva Association, José Iria, the Head of Digital Transformation at La Mobilière, the
Regional Business Manager of a large Swiss healthcare organization, an Account
Manager of a Swiss insurance Group and the Head of Investment Management in a
Swiss major health partner group.The answers provided allow to have a snapshot of the
current understanding, knowledge and use of AI in the Swiss insurance industry.
The legal framework has been reviewed and discussed with Carla Morales, a legal
counsel working at Barclays. This meeting allowed confirmation of the completeness of
the information found about the subject. Additionally, several findings of this research
have further been discussed with Stefan Koch, the Innovation Scout Outpost Silicon
Valley and the current Head of Services, Underwriting and Legal Department at La
Mobilière.
3.2 Interviews
To obtain relevant primary data, a survey has been conducted by means of a standard
questionnaire. It has been circulated among major health insurance companies in
Switzerland, to assess current attitude, planned investments, legal and ethical limits. The
written interview has been done and sent by email due to sanitary restrictions during the
current pandemic crisis. The persons interviewed preferred to reply in writing, to avoid
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 19
being pushed into questions to which they would not want to, or were not be able to,
reply. Several questions were not answered in the written interviews, either because of
confidentiality or because the question was not applicable.
We may also assume that knowledge and statements from insurance staff about AI has
been limited and probably also simply absent, because development of such systems is
company secret, for various reasons such as:
• Companies are developing such toolboxes secretly because of their sensitivity and the legal aspects and limitations
• They develop them with the aim to reduce staff costs, in marketing, back office, loss adjustment and fraud prevention and detection. But as the technology still is in an exploratory stage, communications about such developments are kept confidential
• Costs of AI introduction are very considerable – currently hidden mostly in the admin and information technology budget lines
As several of the questions are sensitive in terms of business secrecy, it has been
necessary to conduct the research in a way that allowed full protection of the sources
who accepted to contribute. To do so, at the beginning of the questionnaire, the person
interviewed was asked if he/she was willing to have his/her name and the name of his/her
current employer kept confidential or he/she did not mind to have them mentioned. The
answers of those willing to keep confidentiality were anonymised, but a description of
their position and company has been added at the beginning of the interview to enable
the reader to have an idea of where the answers came from.
The written interview has been formed based on the current trends of the use of AI in the
health insurance industry in Switzerland through the analysis of secondary data,
researches, articles, and books. The questionnaire was then split into four different parts
in order simplify the understanding of the answers. The first part is about the integration
of Artificial Intelligence in the health insurance and aims at establishing in which area of
the health insurance AI is being the most used currently and where it is expected to be
increasingly used. The second part is about the advantages and constraints (apart from
legal and ethical limits) of the use of AI in this industry and aims at determining if the use
of such a technology is seen as an advantage from both the insured and the insurer. The
third part of the questionnaire refers to where the company of the interviewee is currently
focusing the introduction and use of AI. The objective of the questions in this part is to
assess in which operations the company currently focuses the use AI, if they are
developing and deploying this technology through external consulting firms or through
inhouse data scientists and what are the benefits and challenges of their current use of
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 20
AI. It also attempts to determine what the strategy of the company related to the
introduction of AI in the next 2-5 years is, and what the constraints are that prevent the
company from developing the use of AI even further. The fourth and final part is about
the legal and ethical limits to the use of AI. The objective in this part is to see what the
legal and ethical limits of the use of AI are, according to the interviewee and in the eyes
of the insured. Moreover, the final questions are aimed to identify how the company is
dealing with those limits and if they expect them to rather harden or soften in the coming
years.
3.3 Results
In this part of the analysis, the answers provided by different members of the insurance
industry are described. Each answer is reported in one of the four different parts as in
the questioners found in the appendixes.
Integration of Artificial Intelligence (AI) in the health insurance sector at large
In this first part of the interview, the majority of the interviewees have answered that
nowadays, the area of Swiss health insurance where AI is most often used, is in sales
and marketing and in the detection and prevention of fraud. Most of them expect an
increase of the use of AI in marketing, but also in direct customer interaction (such as
the use of chatbots), in the detection and prevention of fraud and in pricing. Regarding
the question on whether there are differences of trends in the use of AI between smaller
and larger groups, the answers were quite similar. Most of them replied that a
development of the use of AI is more likely to happen in large insurance groups because
the use of this technology requires significant amounts of data and resources which
might be difficult for smaller insurance groups to gather. One answer however, provided
by Ms. Flueckiger, highlighted the fact that the development of AI did not necessarily
depend on the size of the health insurance company, but more on how advanced the
executive management and board of directors are in their familiarity with AI. She has
observed in Switzerland, to the one, small companies with many AI driven processes,
and to the other, leading Swiss health insurance companies that are hardly using AI in
their operations.13
13 Interview with Isabelle Flueckiger, Director New Technologies & Data at the Geneva
Association, Geneva, 29 April 2020 (cf appendix 1).
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 21
Advantages and constraints in the introduction of AI in the health insurance sector
(apart from legal and ethical limits)
The first question in this part is to know whether in the point of view of the interviewees,
the insured consider the integration of AI in the health insurance to their advantage. Here
the answers differed; two persons answered that it depends on the area where AI is
implemented. According to Ms. Flueckiger, the integration of AI is seen as an advantage
only in certain automated areas, such as customer services for instance, in chatbots or
medical advice. In all other areas where personal data is processed, such as risk
classification, pricing or fraud detection, she assumes that the insured are concerned
and fear unfair treatment and discrimination.16 Another answer provided from a
confidential interview states that the insured perceive the use of AI as an advantage only
if it is correctly used by the insurance company. As per this interviewee, it is important
that the client is informed that he is dealing with AI and that he must be provided with
reasons for this use of technology.14 A further answer states that generally, customers
are very skeptical regarding the use of personal data and that they do not fully trust the
company. The person also added that it is still a long way to go, and that the company
must first prove that it can add value through the use of the customer’s data.15 The
second question is whether the health industry view the use of AI by Health Insurance
companies to their advantage. Answers are mainly positive. It is stated that AI can add
value particularly in automation and in the interaction with customers15 and that it has a
great potential for the insurance industry but it also carries some risks.14 According to
Ms. Flueckiger, the health industry mainly does not see the use of AI by health
insurances to their advantage. She mentions that in general, health service providers
such as doctors, consider that health insurance companies do not trust their expertise
and give more trust to their own AI system and that they come up with their own
conclusions based on the data generated by their AI algorithms, regarding the
appropriate treatment to be provided. Moreover, it demands some increased
administrative efforts for health service providers, since they must document each detail
of their decisions, whereas AI algorithms are perceived as a “black-box”. She also states
that life sciences and pharma firms fear discrimination on certain drugs and treatments
and that health insurances might refuse to pay for them and put pressure to find
14 Interview with a Regional Business Manager of a large Swiss healthcare organization, Geneva,
12 April 2020 (cf appendix 3)
15 Interview with an Account Manager of a Swiss insurance Group, Geneva, 30 April 2020 (cf
appendix 4)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 22
alternative treatments and drugs based on their own analysis of the data.16 The third
question refers to the main constraints faced by health insurance companies when using
AI, apart from the legal and ethical limits. Ms. Flueckiger asserts that the main purpose
of an insurance is risk pooling. However, with the introduction of AI, health insurance
companies tend to be willing to have details on the risks of each individual, thus, adapting
individual risk premiums or not insuring high risk patients, which contradicts the main
purpose of insurance. She also added that with the use of AI, individual treatments are
increasingly being scrutinized and, therefore create a debate about who his rightful about
decisions. In her view, the right to take decisions is being removed from health service
providers while they still have the full medical responsibility for patients.16 Two further
short answers, related to customer trust17 and to the quality of data used in the AI
algorithm.18 Another answer stated that health is a very personal topic and that the
customer must not lose confidence because of AI. He should be able to decide for himself
when he wants his personal data to be processed into AI. “The art is to show the
customer where he can benefit from AI without giving him the feeling that the company
wants to save costs or that his data is not safe”19.
Current focus of your company regarding the introduction and use of AI
The first question is to find out on what processes the company of the interviewee, is
currently focusing the use of AI. The most common answers were related to customer
interactions. One person answered that the current main AI focus is on customer services
such as chatbots and recommendations in the company’s applications.19 Another person
answered that in her company, AI is currently used in marketing, risk assessment and
pricing.18 A further answer provided by Mr. Iria states that the current main use of AI at
La Mobilière is in the automation of back office processes, related to contract
management and customer services. He added that some AI components are also being
sprinkled into the development of applications in sales and loss adjustment processes.
This is empowering the company’s sales agents and claims experts. He finally added
that there is a development in the use of AI in some internal actuarial processes, but that
16 Interview with Isabelle Flueckiger, Director New Technologies & Data at the Geneva
Association, Geneva, 29 April 2020 (cf appendix 1)
17 Interview with an Account Manager of a Swiss insurance Group, Geneva, 30 April 2020 (cf
appendix 4)
18 Interview with the Head of Investment Management in a Swiss major health partner group,
Geneva, 20 April 2020 (cf appendix 5)
19 Interview with a Regional Business Manager of a large Swiss healthcare organization, Geneva,
12 April 2020 (cf appendix 3)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 23
it is still in the early days due to tight regulations.20 The next question intends to establish
whether the interviewee’s company is using consulting firms or inhouse data scientists
for the development and deployment of this technology. One answers that the company
has one internal Data Scientist who is also trying to undertake different projects with
external experts.17 Another answer is that they only use inhouse data scientists18 and
some others use both consulting firms and inhouse data scientists. The following
question intends to determine the benefits of the current use of AI in the company. Mr.
Iria answers that the main benefits are in the speeding up of the execution of tasks
through (semi-) automation, enabling their experts to avoid spending their time on
manual and repetitive tasks and, instead, can focus on applying their expertise. He adds
that it translates into an increased cost effectiveness and an increase in the quality of
services provided to customers.22 Another answer suggests that the company is
benefiting from the use of AI to target marketing.22 A further answer states that the
benefits are mainly resulting in a better understanding of the company’s data and
possibilities.23 The last answer states that the current benefits of using AI are that his
company can efficiently handle redundant customer requests, is now available 7/24, can
generate insights from customer inquiries and can play out tailored content to the
customer.21 The next question is about the main challenges encountered with AI
regarding the company’s processes and staff. A first person answers that one of the
challenges is that the results are still of limited quality and value.17 A further answer is
that the main challenges are accuracy, the promotion of trust regarding AI and its
integration into the existing products and services.21 Finally, Mr. Iria states that like any
change, the introduction of AI into the company’s operations requires change
management. The purpose of the next question is to understand the strategy of the
organization with regards to the introduction and use of AI in the coming 2-5 years. The
first person answered that the company is continuously developing the chat offer,
introducing AI in further applications, and using voice recognition.20 The second person
stated that the company’s strategy is to improve the quality of data.22 A further person
stated that their strategy is to invest into AI technologies and build more resources
internally, in order to be able to implement more cases with added value for the
20 Interview with José Iria, Head of Digital Transformation at La Mobilière, Geneva, 08 July 2020
(cf appendix 2)
21 Interview with a Regional Business Manager of a large Swiss healthcare organization, Geneva,
12 April 2020 (cf appendix 3)
22 Interview with the Head of Investment Management in a Swiss major health partner group,
Geneva, 20 April 2020 (cf appendix 5)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 24
customer.23 Finally, Mr. Iria stated that La Mobilière is reshaping governance around
Data and AI, educating their staff on the benefits and risks of using AI, and raising their
technical expertise in AI by hiring Data Scientists. The goal is to monitor the new AI
trends in the insurance industry, and make AI a part of their software development
process.24 The final question relates to the business constraints that may prevent the
company from developing the use of AI even further. The general answer to that question
is that the main constraints are legal compliance, the data protection regulations, data
quality, transparency, and the customer perception regarding the use of this technology.
Mr. Iria states that he believes it is currently not cost effective to use AI in some
processes, due to current limitations of AI when applied to specific domains. He adds
that for La Mobilière, being personal and close to the clients is a fundamental part of the
way they do business, therefore, in some key processes AI will never replace humans
(by choice).23
Legal and ethical limits to the use of AI
The first question in this part is about the legal limitations from the perspective of the
insured. As reported by Ms. Flueckiger, the first limit is transparency. For instance, how
can an insured person reject certain AI-based decisions and who is legally liable for it.
She added that further limits are data privacy protection and prohibited discrimination.
Regarding discrimination she questions how it could be proven, since AI is a “black box”,
and what would be enough evidence for the insured to reject potential discrimination.25
The next question relates to the legal limits of the health insurance industry. Ms.
Flueckiger replies that the health insurance industry would like to apply AI as much as
possible but that they are confronted by various limitations. The first one is the difficulty
to explain AI and the need to acknowledge it to regulators and customers. She adds that
other limits are the high cost of resources, people and infrastructure and that companies
do not have sufficient knowledge and experience and are, therefore, dependent on
external consultants. Another limit in her view is the data itself, missing data
management and data governance. She states that often, data is not available in the
quality needed, or it is not permitted to use it (e.g. data of genomic tests). Additionally,
the market share in Switzerland is so small that not enough data is available resulting to
23 Interview with an Account Manager of a Swiss insurance Group, Geneva, 30 April 2020
(cf appendix 4)
24 Interview with José Iria, Head of Digital Transformation at La Mobilière, Geneva, 08 July 2020
(cf appendix 2)
25 Interview with Isabelle Flueckiger, Director New Technologies & Data at the Geneva
Association, Geneva, 29 April 2020 (cf appendix 1)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 25
fragmented data and data systems. For instance, only about 20-25% of the data in a
company can be accessed within a useful timeframe and on a regular basis.24 Another
answer, states that the regulations from the Federal Office of Public Health (FOPH) and
the Swiss independent supervisory authority (FINMA) imposes guidelines that are
sometimes stricter than the law on data protection.26 The last answer relating to this
question is provided by a colleague of Mr. Iria. He states that he has no knowledge on
whether there are specific regulations related to the use of AI. However, since AI
generates and processes data, the activity should be compliant with the Federal Act on
Data Protection (FADP) and the General Data Protection Regulation (GDPR). He added
that the use of sensitive data such health, race, religious and political affiliation, is legally
restricted. If they want to process this kind of data, they need to obtain the prior consent
of the insured and must inform him about how the data will be used and why. However,
if the data is anonymized, there are no legal restrictions.27 The next question relates to
the current ethical limits linked to the use of AI. The first answer provided by Mr. Iria’s
colleague, mentions that La Mobilière is applying an ethical code of conduct regarding
the processing of data. This is based on the principles of a group of experts from the
Swiss Alliance for Data-Intensive Services.27 (Ethics - Innovation - data-service-
alliance.ch 2020) Another answer provided by Ms. Flueckiger states that the ethical
limitations are discrimination, unfairness, inexplicability and the use of AI to adverse the
patients and customers.28 A further answer states that customers are not confident about
the use of their data. They fear it might be misused and that they would then be provided
with an insurance that does not correspond to their needs. Additionally, they worry that
it contributes to growing a two-tier society, in which healthy and sick individuals are
differentiated, thus, getting away from the social principle of a health insurance.26 The
next question relates to how the company deals with the legal and ethical limits of AI.
One person answers that in the company, the legal and external advisors are included
from the start in the projects and developments.29 Another person states that in his
company, new scenarios are discussed and examined to see if they comply with
regulations and if the company is ready to take a certain risk, to get a step further into
26 Interview with an Account Manager of a Swiss insurance Group, Geneva, 30 April 2020
(cf appendix 4)
27 Interview with José Iria, Head of Digital Transformation at La Mobilière, Geneva, 08 July 2020
(cf appendix 2)
28 Interview with Isabelle Flueckiger, Director New Technologies & Data at the Geneva
Association, Geneva, 29 April 2020 (cf appendix 1)
29 Interview with the Head of Investment Management in a Swiss major health partner group,
Geneva, 20 April 2020 (cf appendix 5)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 26
the future.29 Finally, Ms. Flueckiger says that besides the report, by the Geneva
Association about responsible AI (published in January 2020), they will be doing some
research in 2021 about Ethical Insurance Conduct when using new technologies.28 The
final question intends to determine if the person is expecting the current legal and ethical
limitations to rather harden or soften in the coming years. The answers differ
considerably. Ms. Flueckiger states that before Covid-19, she was expecting the
limitations to harden, and more limitations put in place by the regulators. But now, in the
aftermath of the Covid-19 shock, she rather thinks that the limitations will soften as
tracking and AI based diagnostics are becoming standard and even imposed by
governments, pushing for the use of more data and AI technologies.28 Another person
replies that the European General Data Protection Regulation (GDPR) is definitely the
first sign that the legal side will harden. However, the Covid-19 crisis will increase the
customers awareness about the positive aspects of sharing data and thus, potentially
softening the legal and ethical limitations.30 Another interviewee, states that the
limitations will stay the same in the coming years.31
30 Interview with an Account Manager of a Swiss insurance Group, Geneva, 30 April 2020
(cf appendix 4)
31 Interview with the Head of Investment Management in a Swiss major health partner group,
Geneva, 20 April 2020 (cf appendix 5)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 27
4. Discussion
The various answers to this questionnaire are a snapshot of the current knowledge,
trends and development of AI in Swiss major insurance and health insurance companies.
These are only answers from a few major insurance companies; the overall result cannot
be considered as fully representative for the entire sector. However, they indicate that
the use of AI is still at an early stage. Whether it is about the introduction of this
technology into the company’s processes or the knowledge about its potential and risks,
we can notice in the diversity of the answers, that the understanding of this technology
and its potential need to be further developed.
One important point that was identified comparing different interviews, is the difference
between the basic health insurance and the complementary health insurance. One
person states that the use of AI in the basic insurance is very limited or not possible.
Data linkage between basic and supplementary insurance is not allowed and the link to
medical data is not possible.32 Moreover, as Ms. Flueckiger clearly explains, the basic
health insurance is defined by the governmental Office Federal de la Santé Publique
(OFSP), where the key driver to get new customers in only the price. In this insurance
coverage, AI is not used in the marketing processes. The other point is the
complementary health insurance that is offered by private insurance companies. In this
business area, AI is heavily used for customer segmentation, customer targeting,
customer lifetime value calculations and marketing.33 Another important point highlighted
by Ms. Flueckiger, which confirms the findings in the background, is the negative effect
that AI has on one of the basic principles of insurance. Indeed, one of the primary
principles of a health insurance is risk pooling, implying that the combination of the
healthcare costs of all the insured in a specific segment allows to compute premiums
and to compensate higher or lower average costs of the healthier and less healthy
individuals within the same pool. (Williams 2017) But with the enhanced risk profile
provided by AI for each individual, insurances may adapt the risk premium of each
individual, or even decide not to insure a person, based on the level of risk. This is in
contradiction with the principle of solidarity and it renders risk pooling very difficult.
32 Interview with the Head of Investment Management in a Swiss major health partner group,
Geneva, 20 April 2020 (cf appendix 5)
33 Interview with Isabelle Flueckiger, Director New Technologies & Data at the Geneva
Association, Geneva, 29 April 2020 (cf appendix 1)
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 28
By comparing the different answers, we can notice that each person has a different point
of view and knowledge on AI, its current and future use and, its legal and ethical limits.
As seen in the interview with the Account Manager of a Swiss insurance Group, rules
and regulations are expected to be further refined, notably due to the revision of the
GDPR and the new ePrivacy regulation (ePR) and additionally, due to the revision of the
Swiss regulations such as the FADP, that have recently been adapted to the GDPR.
However, as seen in the answers of two different interviews, with the current Covid-19
pandemic, governments are pushing for an increased use of digital technologies such
as AI to contain and track as much as possible the virus. For health and economical
purposes, this may be of advantage for health insurances; however, the regulations will
not change. Health insurances companies must, therefore, be very cautious regarding
their compliance with regulations when using data and digital technologies.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 29
5. Recommendations for health insurance companies & conclusion
In the coming years, Artificial Intelligence is expected to be much more autonomous
and will have an increasingly important role in the decision-making process of more
and more sectors. (Keller 2020)
The deep changes being brought through the introduction of AI in various processes
will obviously result into an increasing number of stricter rules and regulations.
Moreover, confusion and misconception about Artificial Intelligence truly could prevent
such technologies to develop and benefit the overall society. (Tabate 2016)
The overall findings and information gathered in this thesis allows the listing of the
following recommendations for Swiss health insurance companies:
The development and integration of these new technologies is unavoidable
Big data has been available for many years already and is not a temporary phenomenon.
It is not likely to disappear naturally or to be stopped by legal or technical intervention.
To the contrary: big data will be increasingly available and the capacity to create
information by developing new algorithms will grow at a fast pace, or even exponentially.
Considerable investment of internal and external expertise and financial resources into
such technology by health insurance companies appears unavoidable, in order to remain
competitive in the market.
Adapt governance of the company
In the past, insurance companies were steering and overseeing their IT developments
through committees of mostly technical experts. Future expert groups which pilot
development of big data and AI require to be enhanced with crucial, additional expertise,
such as legal and compliance experts, data protection officers, ethical experts, social
and marketing specialists, and others. Such steering committees must be integrated and
become part of the overall risk management framework of the company.
Customer focus
AI and use of big data should be further developed, with the primary aim to offer a best
possible service to the insured customer. Customers therefore need to be sensitised to
their advantages, such as reduction of administrative costs, an increase of efficiency in
loss adjustment, the prevention of fraud which ensures fair and equal treatment of all
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 30
insured, premium savings, more convenience, transparency and reliability in the
customer’s dealings with the insurance company.
Expand knowledge about AI to all staff
Management staff of health insurance companies may not have enough knowledge
about AI, to understand how it could be implemented into their processes to create value,
and its legal and ethical limitations. Those health insurance companies which develop AI
in their systems should therefore invest in staff training in parallel, in order to increase
awareness and knowledge about and support for this technology within the organisation.
This will then facilitate the understanding of the potential of the use of AI that is currently
still developing in the Swiss health insurances.
External audit and certification
Insurance companies are currently obliged to publish their legal code on use and safety
of personal data, on their Websites. Based on the findings, and in view of the complexity
and technical sophistication, it is advisable that health insurers periodically task third
party experts to audit their use and safety of data and certify compliance with the general
legislation and their own legal code. Companies may also develop, publish, and get
audited on an ethical code of conduct regarding the use and protection of data. This is
not compulsory but promoting transparency and compliance with legal and ethical norms
is essential in creating and maintaining a solid, trusted reputation.
In conclusion, the integration of a technology such as AI will have a great potential for
Swiss health insurances. It can benefit the overall society through various advantages,
but still also carries some risks. It will, therefore, require significant efforts, prerequisites,
and investments from both sides, the insurance companies, and the insured. To be
implemented and further developed, insurance companies should gain trust among
customers through the promotion of their responsible use of data and AI. They should
pay particular attention to their reputation by being transparent regarding the processing
of customers’ personal data and by promoting compliance to the legal regulations and to
ethical codes of conducts. Additionally, Swiss health insurance companies should raise
awareness among customers about this technology, by showing them the advantages
of its use, that they will remain fair, they won’t discriminate and they won’t use personal
data to adversely impact the customers.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 31
Bibliography
CC 101 Federal Constitution of 18 April 1999 of the Swiss Confederation, 2020. [online]. [Viewed 14 August 2020]. Available from: https://www.admin.ch/opc/en/classified-compilation/19995395/index.html
CC 235.1 Federal Act of 19 June 1992 on Data Protection (FADP), 2019. [online]. [Viewed 16 July 2020]. Available from: https://www.admin.ch/opc/en/classified-compilation/19920153/index.html
CC 235.11 Ordinance of 14 June 1993 to the Federal Act on Data Protection (OFADP), 2012. [online]. [Viewed 16 July 2020]. Available from: https://www.admin.ch/opc/en/classified-compilation/19930159/index.html
CC 235.13 Ordinance of 28 September 2007 on Data Protection Certification (DPCO), 2016. [online]. [Viewed 16 July 2020]. Available from: https://www.admin.ch/opc/en/classified-compilation/20071826/index.html
COMMISSIONER (FDPIC), Federal Data Protection and Information, 2015a. Data Protection - Switzerland. [online]. 2015. [Viewed 16 July 2020]. Available from: https://www.edoeb.admin.ch/edoeb/en/home/der-edoeb/links/datenschutz---schweiz.html
COMMISSIONER (FDPIC), Federal Data Protection and Information, 2015b. Task. [online]. 2015. [Viewed 16 July 2020]. Available from: https://www.edoeb.admin.ch/edoeb/en/home/der-edoeb/auftrag.html
COPELAND, B.J., 2019. Artificial intelligence | Definition, Examples, and Applications. Encyclopedia Britannica [online]. 19 November 2019. [Viewed 10 December 2019]. Available from: https://www.britannica.com/technology/artificial-intelligence
COUNCIL OF EUROPE, 2013. European Convention on Human Rights. COE [online]. 2013. [Viewed 11 August 2020]. Available from: https://edoc.coe.int/fr/convention-europenne-des-droits-de-l-homme/5579-european-convention-on-human-rights.html
DFF, Département fédéral des finances, 2020. Swiss Digital Initiative. [online]. 2020. [Viewed 11 August 2020]. Available from: https://www.efd.admin.ch/efd/fr/home/themen/Digitalisierung/swiss-digital-initiative.html
DIGNUM, Virginia, 2019. Responsible Artificial Intelligence: How to Develop and Use AI in a Responsible Way [online]. Cham: Springer International Publishing. [Viewed 24 June 2020]. Artificial Intelligence: Foundations, Theory, and Algorithms. ISBN 978-3-030-30370-9. Available from: http://link.springer.com/10.1007/978-3-030-30371-6
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 32
Ethics - Innovation - data-service-alliance.ch, 2020. [online]. [Viewed 29 July 2020]. Available from: https://data-service-alliance.ch/innovation/ethics
European General Data Protection Regulation GDPR, 2019. S-GE [online]. [Viewed 24 July 2020]. Available from: https://www.s-ge.com/en/european-data-protection-regulation
FINMA, Eidgenössische Finanzmarktaufsicht, 2020. Supervision in the insurance sector. Eidgenössische Finanzmarktaufsicht FINMA [online]. 2020. [Viewed 14 August 2020]. Available from: https://www.finma.ch/en/supervision/insurers/
GDPR ART.5, 2019. Art. 5 GDPR – Principles relating to processing of personal data. General Data Protection Regulation (GDPR) [online]. 2019. [Viewed 15 August 2020]. Available from: https://gdpr-info.eu/art-5-gdpr/
GDPR ART.15, 2019. Art. 15 GDPR – Right of access by the data subject. General Data Protection Regulation (GDPR) [online]. 2019. [Viewed 14 August 2020]. Available from: https://gdpr-info.eu/art-15-gdpr/
GORDON, Clara-Ann and GUROVITS, András, 2020. AI, Machine Learning & Big Data Laws and Regulations | Switzerland | GLI. GLI - Global Legal InsightsInternational legal business solutions [online]. 2020. [Viewed 13 July 2020]. Available from: https://www.globallegalinsights.com/practice-areas/ai-machine-learning-and-big-data-laws-and-regulations/switzerland United Kingdom
HARARI, Yuval Noah, 2018. 21 Lessons for the 21st Century. 1st ed. London: Jonathan Cape/libri. ISBN Trade paperback 9781787330870.
INRIX, 2020. About. Inrix [online]. 2020. [Viewed 18 August 2020]. Available from: https://inrix.com/about/
KELLER, Benno, ELING, Martin, SCHMEISER, Hato, CHRISTEN, Markus and LOI, Michele, 2018. Big Data and Insurance: Implications for Innovation, Competition and Privacy. [online]. March 2018. P. 48. Available from: https://www.genevaassociation.org/sites/default/files/research-topics-document-type/pdf_public/big_data_and_insurance_-_implications_for_innovation_competition_and_privacy.pdf
KELLER, Benno, 2020. Promoting Responsible Artificial Intelligence in Insurance. Artificial Intelligence [online]. 2020. P. 24. Available from: https://www.genevaassociation.org/sites/default/files/research-topics-document-type/pdf_public/ai_in_insurance_web_0.pdf
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 33
KRAFT-BUCHMAN, Caitlin and ARIAN, Renée, 2019. Deadly Data Gap: Gender & Data. Deadly Data Gap: Gender & Data [online]. 21 November 2019. [Viewed 17 August 2020]. Available from: https://www.womenatthetable.net/blog/deadly-data-gap-gender-data
LPGA, 2019. RS 830.1 Loi fédérale du 6 octobre 2000 sur la partie générale du droit des assurances sociales (LPGA). [online]. 1 October 2019. [Viewed 14 August 2020]. Available from: https://www.admin.ch/opc/fr/classified-compilation/20002163/index.html
MAYER-SCHÖNBERGER, Viktor and CUKIER, Kenneth, 2013. Big Data: A Revolution That Will Transform How We Live, Work, and Think. Reprint edition. Eamon Dolan/Houghton Mifflin Harcourt.
MICHAEL, 2019. How Artificial Intelligence Is Improving Insurance. [online]. 22 September 2019. [Viewed 8 November 2019]. Available from: https://hackernoon.com/how-artificial-intelligence-is-revolutionizing-insurance-9n5e230zp
MÜLLER, Vincent C., 2020. Ethics of Artificial Intelligence and Robotics. In: The Stanford Encyclopedia of Philosophy [online]. Fall 2020. Metaphysics Research Lab, Stanford University. [Viewed 7 July 2020]. Available from: https://plato.stanford.edu/archives/fall2020/entries/ethics-ai/
NAKONZ, Jonas, 2019. AI “principles” need refinement. digitalswitzerland [online]. 14 June 2019. [Viewed 8 November 2019]. Available from: https://digitalswitzerland.com/2019/06/14/ai-principles-need-refinement/
OFSP, Office fédéral de la santé publique, 2020. Assurance-maladie : L’essentiel en bref. [online]. 2020. [Viewed 14 August 2020]. Available from: https://www.bag.admin.ch/bag/fr/home/versicherungen/krankenversicherung/krankenversicherung-das-wichtigste-in-kuerze.html
PANCHAL, Shubham, 2019. Types of Artificial Intelligence and examples. Medium [online]. 1 November 2019. [Viewed 10 December 2019]. Available from: https://medium.com/predict/types-of-artificial-intelligence-and-examples-4f586489c5de
PRICEWATERHOUSECOOPERS, 2020. New Swiss Federal Act on Data Protection: 5 changes to remember. PwC [online]. 2020. [Viewed 19 July 2020]. Available from: https://www.pwc.ch/en/insights/fs/data-protection-switzerland.html
SCHMIDT, Christian, 2018. Insurance in the Digital Age. [online]. September 2018. P. 20. Available from: https://www.genevaassociation.org/sites/default/files/research-topics-document-type/pdf_public/insurance_in_the_digital_age_01.pdf
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 34
SCHWARTZ, Oscar, 2019. In 2016, Microsoft’s Racist Chatbot Revealed the Dangers of Online Conversation - IEEE Spectrum. IEEE Spectrum: Technology, Engineering, and Science News [online]. 2019. [Viewed 4 May 2020]. Available from: https://spectrum.ieee.org/tech-talk/artificial-intelligence/machine-learning/in-2016-microsofts-racist-chatbot-revealed-the-dangers-of-online-conversation
SEFRI, Secrétariat d’État à la formation, à la recherche et à l’innovation, 2019. Intelligence artificielle. [online]. 13 December 2019. [Viewed 10 August 2020]. Available from: https://www.sbfi.admin.ch/sbfi/fr/home/bfi-politik/bfi-2021-2024/transversale-themen/digitalisierung-bfi/kuenstliche-intelligenz.html
SHAW, Jonathan, 2018. Artificial Intelligence and Ethics. Harvard Magazine [online]. 6 December 2018. [Viewed 8 November 2019]. Available from: https://harvardmagazine.com/2019/01/artificial-intelligence-limitations
SHROFF, Raj, 2019. How Are Insurance Companies Implementing Artificial Intelligence (AI)? Medium [online]. 5 October 2019. [Viewed 10 December 2019]. Available from: https://towardsdatascience.com/how-are-insurance-companies-implementing-artificial-intelligence-ai-aaf845fce6a7
SWISSMEDICALNETWORK, 2020. Swiss Medical Network - Data protection. [online]. 2020. [Viewed 14 August 2020]. Available from: https://www.swissmedical.net/en/data-protection/
TABATE, Tom, 2016. History | One Hundred Year Study on Artificial Intelligence (AI100). [online]. 2016. [Viewed 10 August 2020]. Available from: https://ai100.stanford.edu/history-1
THESWISSAUTHORITIESONLINE, 2020. Health insurance in Switzerland: basic insurance, health insurance companies, supplementary insurance, coverage - www.ch.ch. [online]. 2020. [Viewed 18 August 2020]. Available from: https://www.ch.ch/en/health-insurance/
TISCHHAUSER, Stephanie and GEORGIEVA, Evelina, 2019. GDPR, Swiss DPA & ePrivacy – what Swiss companies should know. Pryv [online]. 20 November 2019. [Viewed 26 July 2020]. Available from: https://www.pryv.com/2019/11/20/gdpr-swiss-dpa-e-privacy/
WELLENS, Patrick, 2020. Ethics and Compliance Switzerland » Revision of the Swiss Federal Act on Data Protection: Lawmakers establish first cornerstones of the new Data Protection Act. [online]. 27 February 2020. [Viewed 19 July 2020]. Available from: https://www.ethics-compliance.ch/2020/02/27/revision-of-the-swiss-federal-act-on-data-protection-lawmakers-establish-first-cornerstones-of-the-new-data-protection-act/
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 35
WHO, 2017. WHO | Health is a fundamental human right. WHO [online]. 10 December 2017. [Viewed 21 August 2020]. Available from: http://www.who.int/mediacentre/news/statements/fundamental-human-right/en/
WILLIAMS, Joeff, 2017. Risk Pooling: How Health Insurance in the Individual Market Works | American Academy of Actuaries. [online]. 2017. [Viewed 30 July 2020]. Available from: https://www.actuary.org/content/risk-pooling-how-health-insurance-individual-market-works-0
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 36
Appendix 1: Interview with Isabelle Flueckiger
(Director New Technologies & Data at the Geneva Association)
Questionnaire on the use of Artificial Intelligence (AI) by Swiss health insurance companies
I ensure that this interview will be conducted in a way that allows full protection of the sources.
O I want my name and the name of my current employer to be kept
confidential
● I do not mind if my name and the name of my current employer is
mentioned in the thesis
1. Integration of Artificial Intelligence (AI) in the health insurance
sector at large
1.1 In which areas of Swiss health insurance is AI most often used, and
how is it used?
▪ Marketing
One has to differentiate between the basic health insurance (KVG,
en français: Loi fédérale sur l'assurance-maladie (LAMal)), and
supplemental health insurance according VVG (en français: Loi sur le
contrat d’assurance, LCA)
In KVG, the insurance benefit is defined by the government / BAG
(en français: office federal de la santé OFSP), and the key driver for
getting new customers is only the price. There typically, no AI is used
in marketing.
It is different in the supplemental health insurance that can be offered
by health insurance companies as well as private insurance
companies. On one side, several different insurance benefit contracts
exists, and one has a free market. There AI is heavily used for
customer segmentation, customer targeting, customer lifetime value
calculations.
▪ Risk assessment & pricing
Most products in the Swiss market are still based on and are
assessed by classical actuarial procedures. There are few direct
insurer that use AI, but still struggle with the lack of transparency and
explainability and therefore, they are very cautious to apply that only
on a stand-alone basis.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 37
▪ Loss adjustment
See also others. As part of the automated background processing of
bills, AI is used. When it comes to loss adjustments, they are either
made by an expert, or if there are serious and complex cases, even
with the help of layers. The AI is used for either: no large deviations
from the expectations, and thus payment, or deviations and then an
expert has to do it. The reason behind that are again regulatory
requirements as well as it need to be bullet-proved because
otherwise there will be a lawsuit.
▪ Detection and prevention of fraud
See also other.
AI is used in the automated background checking of bills for finding
inconsistencies, as well as for looking patterns over certain time
periods, taking into account bills from the same insured person, or
bills from the same health benefit provider e.g. if certain benefits are
billed twice on different bills and so on.
▪ Product development
This is only valid for the supplemental health insurance. Currently, still the classical actuarial product development methods
are used and not necessarily AI. The reason is that is more difficult to
find a product type that people will buy because they see a need for it
than finding individual demand by AI.
▪ Other?
In the health insurance industry, one has automated background
processes for checking the incoming bills on consistency (“shadow
processing”). If no inconsistency is found, the bill is just paid. Only
bills that seems to be inconsistent e.g. strange SwissDRG or Tarmed
codes, or quantities, will then go to an expert, sometimes to medical
experts, where the bill is reappraised. This is one of the most
important process in a health insurance company and thus, most
automated and with most sophisticated AI. I.e. 100% of the incoming
bills are assessed in that way.
Personalization of the insurance pricing based on wearable data e.g.
fitbits, fitness trackers, genetic tests and so on. There is a lot of
ethical and data privacy concerns but the trend is there to use these
data with AI for personalized insurance and finally, the determination
of the premium.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 38
1.2 In which of the above areas do you expect an increase of use of AI?
Marketing, pricing, fraud incl. automated background checking, and in
pricing based on wearable data e.g. fitbits and so on.
1.3 Do you expect a difference of these trends between smaller and
larger insurance groups?
It is not necessarily dependent of the size of the health insurance
company, or insurance company.
It depends how advanced the executive management and board of
directors are in the familiarity with AI. I have seen very small
companies with a lot of AI driven processes, and leading Swiss health
insurance companies, that are doing nearly nothing with AI.
The ones that implement have different approaches. Larger ones
build it up in-house, whereas smaller ones have standardized
products for AIaaS (AI as a service).
Direct insurers that have a supplemental health insurance line of
business, are typically leveraging the knowhow of the other business
and building it up by themselves.
2. Advantages and constraints in the introduction of AI in the health
insurance sector (apart from legal and ethical limits)
2.1 Do the insured view the integration of AI to their advantage?
It depends on the area where it is integrated.
If it is integrated in customer services e.g. chatbots, medical advice,
then it is considered as an advantage. But in all other areas, if data
are used for risk classification, pricing, fraud detection, the insured
have concerns and are expecting unfairness and discrimination.
2.2 Does the health industry view the use of AI by Health Insurance
companies to their advantage?
Mainly not. Hospitals or benefit providers (e.g. doctors) in general have
concerns because of the analyses and results based on AI and that
there are much more rejections for paying bills on the health insurer
side. Or because health insurer come up with their own conclusion of
what could be the best treatment based on the data and generated by
AI algorithms. Benefit providers say that the health insurance
companies do in general not trust to them and trust more to their own
AI than to the experts.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 39
All of this also leads to more administration effort of the benefit provider
because they have to document each detail of their decision, whereas
the AI algorithms are a black-box.
Life sciences and pharma firms fear that certain treatments and drugs
will be replaced by the health insurance companies because they will
not pay them and put pressure to alternative treatments. They fear a
discrimination of certain drug and treatments.
2.3 What are the main constraints which the health insurance industry
faces when using Artificial Intelligence apart from the legal and ethical
limits?
Medical treatment and advice is not black nor white and individual
treatments become much more scrutinized. And if there is a deviation
from the average there is a debate who is now right with the decision,
AI on the health insurance side or a doctor. There is more decision-
rights taken away from the benefit providers but they still have the full
medical responsibility for a patient.
Further, the purpose of insurance is risk pooling, but health insurance
companies wants to know the detailed individual risk and thus, details
individual risk premium or not insure the patient at all which is in
contradiction of the purpose of insurance.
3. Current focus of your company regarding the introduction and use
of AI
3.1 If any, on what processes is your company currently focusing AI
technology?
We are a global think tank, and we are doing research on the topic
but not implementing it.
(At the former company, a global consulting firm, we implemented it in
internal processes, like HR, IT support, and so on.)
3.2 Is your company using consulting firms or inhouse data scientists for
the development and deployment of this technology?
N/A – we are a think tank for the insurance industry
3.3 What are the benefits of the use of AI currently?
N/A – we are a think tank for the insurance industry
3.4 What are the main challenges you encounter with AI regarding your
processes and staff?
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 40
N/A – we are a think tank for the insurance industry
3.5 What strategy does your company have with regards to the
introduction and use of AI in the coming 2-5 years?
N/A – we are a think tank for the insurance industry
3.6 What are the business constraints that may prevent your company
from developing even further the use of AI?
N/A – we are a think tank for the insurance industry
4. Legal and ethical limits to the use of AI
4.1 What are the legal limitations in the use of AI?
4.1.1 From the perspective of the insured
Transparency, e.g. how can an insured person veto against
certain AI-based decision? Who is responsible?
Data privacy protection, and not permitted discrimination, but
how to prove it, if AI is a black box? And what are enough
evidence that an insured person can veto against potential
discrimination?
4.1.2 From the perspective of the health insurance industry
The health insurance industry would like to apply AI as much
as possible. Limitations are:
• Non-explainability of AI, but needed to explain it towards
the regulators, customers, and so on
• Talents, skills and capacities to build the knowledge and
tools to apply it in a proper way – Google, IBM, or banks
are more attractive employers with higher salaries
• High costs of resources, people and infrastructure and
not sufficient knowledge and experience (or
dependence on external consultants)
• Data and data input: often data are not available in the
quality needed, or data are not permitted to use e.g.
data of genomic tests, or the market share in
Switzerland is to small that not enough data is available,
or very fragmented data and data systems (typically,
only about 20-25% of the data in a company can be
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 41
accessed within a useful timeframe and on a regular
basis (missing data management and data governance)
4.2 What are the ethical limitations linked to the use of AI?
Discrimination, non-fairness, no explainability, use to the adverse of
the patient / customer
4.3 How does your company deal with the legal and ethical limits of using
Artificial Intelligence?
N/A – we are a think tank for the insurance industry
We are researching these topics. Besides the report about
responsible AI (published in January 2020 by the Geneva
Assocation), we will do research in 2021 about Ethical Insurance
Conduct when using new technologies.
4.4 Do you expect the current legal and ethical limitations to rather harden
or soften in the coming years?
Before Covid-19 I was expecting the limitations to harden and more
limitation will be put in place by governments. But now, with Covid-19
and the aftermaths, I rather think it will be soften. Because tracking and
AI based diagnostics are becoming standard and imposed by the
government. Thus, the government pushes in using more data and AI
and softens limitations.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 42
Appendix 2: Interview with José Iria
Head of Digital Transformation at La Mobilière
Questionnaire on the use of Artificial Intelligence (AI) by Swiss health insurance companies
I ensure that this interview will be conducted in a way that allows full protection of the sources.
O I want my name and the name of my current employer to be kept
confidential
X I do not mind if my name and the name of my current employer is
mentioned in the thesis
1. Integration of Artificial Intelligence (AI) in the health insurance
sector at large
1.1 In which areas of Swiss health insurance is AI most often used, and
how is it used?
▪ Marketing
▪ Risk assessment & pricing
▪ Loss adjustment
▪ Detection and prevention of fraud
▪ Product development
▪ Other?
1.2 In which of the above areas do you expect an increase of use of AI?
I expect that insurance companies will mainly increase their use of AI
in "Risk assessment & pricing" and "Detection and prevention of fraud",
while benefiting from developments in the use of AI in "Marketing"
across all industries. There will be some use of AI in "Sales and
Distribution" and "Loss Adjustment" as well. Less so in "Product
Development"
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 43
1.3 Do you expect a difference of these trends between smaller and
larger insurance groups?
No reply.
2. Advantages and constraints in the introduction of AI in the health
insurance sector (apart from legal and ethical limits)
2.1 Do the insured view the integration of AI to their advantage?
No reply.
2.2 Does the health industry view the use of AI by Health Insurance
companies to their advantage?
No reply.
2.3 What are the main constraints which the health insurance industry
faces when using Artificial Intelligence apart from the legal and ethical
limits?
No reply.
3. Current focus of your company regarding the introduction and use
of AI
3.1 If any, on what processes is your company currently focusing AI
technology?
Currently, our main use of AI is for automating back office processes related to contract management and customer services.
There are also some AI components being sprinkled into the development of applications in sales and loss adjustment processes, empowering our sales agents and claims experts.
Finally, we start seeing the use of AI in some internal actuarial processes, but it is still early days due to tight regulation.
3.2 Is your company using consulting firms or inhouse data scientists for
the development and deployment of this technology?
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 44
Both
3.3 What are the benefits of the use of AI currently?
Mainly speeding up the execution of tasks through (semi-)automation, avoiding that our experts spend their time doing manual and repetitive work and can focus on applying their expertise.
This translates into increased cost effectiveness and increase in the quality of our service to the customer.
3.4 What are the main challenges you encounter with AI regarding your
processes and staff?
Like any change, it needs change management. Overall, I would say
that most collaborators are positive about AI.
3.5 What strategy does your company have with regards to the
introduction and use of AI in the coming 2-5 years?
We are reshaping governance around Data and AI, educating our staff
on the benefits and risks of using AI, raising our technical expertise in
AI by hiring Data Scientists, monitoring the new trends in AI in the
insurance industry, and making AI a part of our software development
process.
3.6 What are the business constraints that may prevent your company
from developing even further the use of AI?
Besides the constraints related to the usage of Data (see below), I
believe that it is currently not cost effective to use AI in some
processes, due to present limitations of AI when applied to certain
domains. Moreover, for La Mobilière being personal and close to their
clients is a fundamental part of the way we do business, therefore in
some key processes AI will never replace a human (by choice).
4. Legal and ethical limits to the use of AI
4.1 What are the legal limitations in the use of AI?
Je n'ai pas connaissance de limitations spécifiques sur l'utilisation de l'IA; cependant bien attendu, comme l'IA utilise et génère des données, l'activité doit respecter les lois suisse LPD (loi sur la protection des
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 45
données) et européenne RGPD (règlement sur la protection des données).
Les données pour lesquelles il y a des restrictions légales sont les données sensibles sur la personne (ex: santé, opinions religieuses / politiques, appartenance à une race, …). Si l'on veut travailler avec des données sensibles (récolter, communiquer, exploiter…), il faut obtenir l'accord de la personne et l'informer sur l'utilisation que l'on va en faire. Cependant, si les données sont anonymisées, il n'y a plus de restriction légale.
4.1.1 From the perspective of the insured
No reply.
4.1.2 From the perspective of the health insurance industry
No reply.
4.2 What are the ethical limitations linked to the use of AI?
La Mobilière applique aussi un code éthique sur l'utilisation des données; rien de particulier sur l'aspect IA; la Mobilière s'appuie sur les principes de ce groupe d'experts:
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 46
Appendix 3: Interview (Confidential)
Interview with a Regional Business Manager in a large Swiss Healthcare Organization providing health and accident insurance
Questionnaire on the use of Artificial Intelligence (AI) by Swiss health
insurance companies
I ensure that this interview will be conducted in a way that allows full protection of the sources.
x I want my name and the name of my current employer to be kept
confidential
O I do not mind if my name and the name of my current employer is
mentioned in the thesis
1. Integration of Artificial Intelligence (AI) in the health insurance
sector at large
1.1 In which areas of Swiss health insurance is AI most often used, and
how is it used?
▪ Marketing
▪ Risk assessment & pricing
▪ Loss adjustment
▪ Detection and prevention of fraud
▪ Product development
▪ Other?
Customer service: Answering or triage of customer enquiries
1.2 In which of the above areas do you expect an increase of use of AI?
Marketing
1.3 Do you expect a difference of these trends between smaller and
larger insurance groups?
More increase in larger insurance groups. They have more resource
to develop and invest.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 47
2. Advantages and constraints in the introduction of AI in the health
insurance sector (apart from legal and ethical limits)
2.1 Do the insured view the integration of AI to their advantage?
If AI is used correctly, it is perceived as an advantage by the insured.
It is important that the client understands that he is dealing with AI
and is given reasons why it is being used.
2.2 Does the health industry view the use of AI by Health Insurance
companies to their advantage?
AI offers a lot of potential for the insurance industry, but also carries
risks. It is important to assess where the use of AI makes sense.
2.3 What are the main constraints which the health insurance industry
faces when using Artificial Intelligence apart from the legal and ethical
limits?
Client must not lose confidence because of AI. Health is a very
personal topic. Therefore, the customer must be able to decide for
himself if and when he wants to be confronted with AI.
The art is to show him where he can benefit without giving him the
feeling that you want to save costs or that his data might not be safe.
3. Current focus of your company regarding the introduction and use
of AI
3.1 If any, on what processes is your company currently focusing AI
technology?
Customer service / Chatbot → Recommendations in our applications
3.2 Is your company using consulting firms or inhouse data scientists for
the development and deployment of this technology?
Both
3.3 What are the benefits of the use of AI currently?
Efficient handling of redundant customer requests
7x24 availability
Generate insights from customer enquiries
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 48
Playing out content tailored to the customer
3.4 What are the main challenges you encounter with AI regarding your
processes and staff?
Accuracy
Promoting trust in AI
Integration into existing products and services
3.5 What strategy does your company have with regards to the
introduction and use of AI in the coming 2-5 years?
Continuous development of the chat offer, use of AI in further
applications, use of voice recognition
3.6 What are the business constraints that may prevent your company
from developing even further the use of AI?
Data protection Transparent customer
4. Legal and ethical limits to the use of AI
4.1 What are the legal limitations in the use of AI?
No reply.
4.1.1 From the perspective of the insured
No reply.
4.1.2 From the perspective of the health insurance industry
No reply.
4.2 What are the ethical limitations linked to the use of AI?
No reply.
4.3 How does your company deal with the legal and ethical limits of using Artificial Intelligence?
No reply.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 49
4.5 Do you expect the current legal and ethical limitations to rather harden
or soften in the coming years?
No reply.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 50
Appendix 4: Interview (Confidential)
Interview with an Account Manager in a Swiss Group made of the leading health insurers in Switzerland
Questionnaire on the use of Artificial Intelligence (AI) by Swiss health
insurance companies
I ensure that this interview will be conducted in a way that allows full protection of the sources.
X I want my name and the name of my current employer to be kept
confidential
O I do not mind if my name and the name of my current employer is
mentioned in the thesis
1. Integration of Artificial Intelligence (AI) in the health insurance
sector at large
1.1 In which areas of Swiss health insurance is AI most often used, and
how is it used?
▪ Marketing
▪ Risk assessment & pricing
▪ Loss adjustment
▪ Detection and prevention of fraud
▪ Product development
▪ Other?
Currently, the main usage is in sales & marketing (optimization of the sales process, personalization of services and interaction). Coming more and more is the usage in process automation (for example self-services like chatbot, automated E-Mail responses) and detection and prevention of fraud as another atomization case.
1.2 In which of the above areas do you expect an increase of use of AI?
In the direct customer interaction
1.3 Do you expect a difference of these trends between smaller and
larger insurance groups?
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 51
The amount of data you need for fraud detection or prevention has to
be quite big, not sure if small insurances will be able to build a model
on that. Smaller insurance groups will probably not invest as much in
AI and do more with simple management of data-based rules, getting
probably already 50-75% of the possibilities out of it.
2. Advantages and constraints in the introduction of AI in the health
insurance sector (apart from legal and ethical limits)
2.1 Do the insured view the integration of AI to their advantage?
No, generally, our customers do not really trust us, this is still a long
way to go. So, they view the usage of data very skeptical. We first have
to provide proof that we can add value with using the customer’s data
so they will become more open to sharing their data with us.
2.2 Does the health industry view the use of AI by Health Insurance
companies to their advantage?
Definitely, this is a field that can add value especially in automation and
in the interaction with the customer.
2.3 What are the main constraints which the health insurance industry
faces when using Artificial Intelligence apart from the legal and ethical
limits?
As mentioned under 2.1 this is trust.
3. Current focus of your company regarding the introduction and use
of AI
3.1 If any, on what processes is your company currently focusing AI
technology?
Customer interaction
3.2 Is your company using consulting firms or inhouse data scientists for
the development and deployment of this technology?
We have one Data Scientist internally and try to do projects in a
tandem with external experts as well.
3.3 What are the benefits of the use of AI currently?
A better understanding of our data and the possibilities.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 52
3.4 What are the main challenges you encounter with AI regarding your
processes and staff?
Everyone thinks they know what AI is, but in fact they have all different
opinions. Thankfully, we do not encounter a lot of skepticism.
3.5 What strategy does your company have with regards to the
introduction and use of AI in the coming 2-5 years?
Invest in it and build up more resources internally, to be able to
implement more cases with added value for the customer.
3.6 What are the business constraints that may prevent your company
from developing even further the use of AI?
Customer perception and definitely also legal and compliance.
4. Legal and ethical limits to the use of AI
4.1 What are the legal limitations in the use of AI?
4.1.1 From the perspective of the insured
As long as the insured allows us the usage of their data, there are no
limitations.
4.1.2 From the perspective of the health insurance industry
The BAG and FINMA give us some guidelines, which are sometimes
stricter than the data law.
4.2 What are the ethical limitations linked to the use of AI?
Customers are not sure if we could misuse their data and then not
provide them with the insurance they need.
Additionally, it gets more into a two-tier society, differentiating more
between healthy and sick, getting away from the social principle of
health insurance, as data makes everything transparent.
4.3 How does your company deal with the legal and ethical limits of
using Artificial Intelligence?
With new cases, they are discussed if they compile with regulations
and if our group as a company is ok with taking a certain risk if
necessary, to get a step further into the future.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 53
4.4 Do you expect the current legal and ethical limitations to rather
harden or soften in the coming years?
GDPR is definitely the first sign that the legal side will harden. However,
crisis like Corona will make customers more aware of the positive
aspects of sharing data, maybe softening their ethical view as well on
the other side.
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 54
Appendix 5: Interview (Confidential)
Interview with a Head of Investment Management in a major health partner group in Switzerland
Questionnaire on the use of Artificial Intelligence (AI) by Swiss health
insurance companies
I ensure that this interview will be conducted in a way that allows full protection of the sources.
X I want my name and the name of my current employer to be kept
confidential
O I do not mind if my name and the name of my current employer is
mentioned in the thesis
1. Integration of Artificial Intelligence (AI) in the health insurance
sector at large
1.1 In which areas of Swiss health insurance is AI most often used, and
how is it used?
▪ Marketing
▪ Risk assessment & pricing
▪ Loss adjustment
▪ Detection and prevention of fraud
▪ Product development
▪ Other?
1.2 In which of the above areas do you expect an increase of use of AI?
Marketing
1.3 Do you expect a difference of these trends between smaller and
larger insurance groups?
Yes
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 55
2. Advantages and constraints in the introduction of AI in the health
insurance sector (apart from legal and ethical limits)
2.1 Do the insured view the integration of AI to their advantage?
Depending on the area yes
2.2 Does the health industry view the use of AI by Health Insurance
companies to their advantage?
Yes
2.3 What are the main constraints which the health insurance industry
faces when using Artificial Intelligence apart from the legal and ethical
limits?
Data quality
3. Current focus of your company regarding the introduction and use
of AI
3.1 If any, on what processes is your company currently focusing AI
technology?
Marketing, Risk assessment & pricing
3.2 Is your company using consulting firms or inhouse data scientists for
the development and deployment of this technology?
Inhouse
3.3 What are the benefits of the use of AI currently?
Target Marketing
3.4 What are the main challenges you encounter with AI regarding your
processes and staff?
Results are still of limited quality and value
3.5 What strategy does your company have with regards to the
introduction and use of AI in the coming 2-5 years?
Improve data quality
Legal and ethical constraints in the use of Artificial Intelligence (AI) by Swiss health insurance companies Kerim TRUNIGER 56
3.6 What are the business constraints that may prevent your company
from developing even further the use of AI?
Regulatory, data quality
4. Legal and ethical limits to the use of AI
4.1 What are the legal limitations in the use of AI?
There are no “perspectives” with respect to legal limitations, but there
is a difference between basic insurance and supplementary
insurance. In basic insurance a lot of fields above are simply not
possible. Data linkage between basic and supplementary is not
allowed. Link to medical data is not possible.
4.1.1 From the perspective of the insured
No reply.
4.1.2 From the perspective of the health insurance industry
No reply.
4.2 What are the ethical limitations linked to the use of AI?
In the basic insurance, this is addressed by the legal context. In
complementary, the insured should not be in worse situation as if
there was no AI.
4.3 How does your company deal with the legal and ethical limits of
using Artificial Intelligence?
Legal department and external advisors are included from the start in
the projects and developments
4.4 Do you expect the current legal and ethical limitations to rather harden or soften in the coming years?