European AI Alliance

Archived

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Blog

Which themes should the High-Level Expert Group on AI (AI HLG) tackle?

Lucilla SIOLI

14 June 2018 - updated 2 years ago
53 comments
Total votes: 9

On 27 June, the AI HLG will meet for the first time and discuss (1) the main ethical principles to focus on when preparing the draft AI ethics guidelines, and (2) the main AI-related challenges to address in their policy recommendations (including on Europe's competitiveness and on the future of work).

The first discussion we would like to launch with you is:

Which are the main principles that the AI ethics guidelines should tackle?

Your feedback on this question will be used by the AI HLG in its first meeting!

Update: We collect comments to this question until Thursday 21 June end of day. Afterwards, we will process them and give them to the AI HLG to prepare their first meeting.

Tags:

AI HLG Ethics guidelines

Comments

Athanasios Stav...,

02/07/2018 10:01

Why not move from theory of AI ethics, to practise ?

please see below a video from youtube, describing an AI drone with the capability of identifying people based , among others, on Age, gender, fitness(!!) and ethnicity!!!!!.

http://www.foxnews.com/tech/2017/11/21/uc-berkeley-professor-s-slaughter...

This video is developed by UC Berkeley science professor Stuart Russell, who claimes that these types of AI drones are a reality!

We should move fast in order to establish not only ethical rules, but activelly shape regulations.

Technology is ahead from us, we need to speed up.

Please for your comments.

Athanasios Staveris-Polykalas

Vanja Skoric,

27/06/2018 16:50

The European Center for Not-for-Profit Law (ECNL) and the European Civic Forum wish to provide the following topic for discussion for the EU AI Expert Group:

Discuss opportunities and challenges of the Artificial Intelligence (AI) impact on civic space, civil society activities and civic fundamental freedoms. These specifically include the right to freedom of association, the right to freedom of peaceful assembly, the right to freedom of expression, access to information right, and public participation right, fundraising options, donor and beneficiary relations, etc. What policies and principles are necessary to address such opportunities and challenges?

Explanation:

Within the scope of protecting civic space on the ground, we are increasingly coming across the adoption of innovations in areas of artificial intelligence (AI) that are creating new opportunities but also challenges to human rights and freedoms. AI’s potential impact on freedom of assembly, freedom of association, freedom of information, freedom of expression, political participation, due process and the full range of other civil, economic, social and cultural rights needs to be discussed and explored further. The Guidelines on Automated Individual Decision-making and Profiling for the Purposes of the New General Data Protection Regulation recently issued by the European Commission Data Protection Working Party have also cautionedthat “profiling and automated decision-making can pose significant risks for individuals’ rights and freedoms which require appropriate safeguards.” For example, algorithms used by some governments in law enforcement and access to due process have the potential to reinforce and deepensystemic discrimination or replicate existing biases. In addition, algorithms that support digital echo chambers are enabling the spread of ‘fake news’, deepening digital bubbles and impacting electoral processes. Governments are likely to deploy AI systems to regulate freedom of assembly, freedom of association, access to information and freedom of expression: e.g., AI systems will likely be adopted to automate the registration process for civil society organizations or notifications for peaceful assemblies. While an automated process would be efficient, if not properly designed, it could embed biases and constraints inconsistent with freedom of association and freedom of assembly. For example, reasons for declining registation or declining a peaceful assembly could remain in the „black box“ without adequate possibility for full review. In addition, if governments purchase or lease AI systems from companies, will the algorithms be subject to freedom of information reviews and allow public oversight? The issue of transparency, accountability, and liability of such approach will be hugely important. Moreover, AI systems can be used to track online speech critical of government policies, which in turn could be used to produce alghoritms to stem the flow of critical speech. In addition, AI systems could be designed to self-learn to identify and monitor people that are likely to participate in peaceful gatherings or associations critical of governmental authorities. Authorities could then use this information to prevent or limit such freedoms. On the other hand, the AI provides additional opportunities for charitable fundraising and interactionwith donors, beneficiaries and citizens on civil society actions. AI applications can help promote public involvement in public policy, inclusive participation and more transparent administrative decision-making. Finally, the World Economic Forum's White Paperalso called for integrating human rights into machine learning. There will be a need for a standardisation system for verifying and monitoring AI systems, in the areas of safety, transparency, comprehensibility, accountability and compatibility with fundamental rights and freedoms.

The EU has increasingly found civil society under threat, and at the same time, considers civil society as a crucial component of any democratic system. That is why the EU institutions developed a number of policies on protectic civil society,which should be complemented by exploring and identifying AI features that will impact fundamental freedoms. AI is a dual-use technology: its potential benefits are massive, but it's potential for harm is also great. We see the AI's great potential to benefit civic freedoms, promote inclusiveness and serve the needs of local communities. In doing so, we also see the need to minimise AI's potential for abuse, harm and threats to fundamental freedoms and human rights.

Many thanks!

Vanja (ECNL)

Richard Krajčoviech,

25/07/2018 15:12

Good point, Vanja, thanks a lot. My opinion is that any "black box" excuses must be prohibited and users and vendors of AI must share responsibility for compliance with any existing regulation (and for adoption to new ones) regardless they use AI or not. Happy to participate on such discussions.

Vivek NALLUR,

26/06/2018 16:17

One aspect of AI and ethics that doesn't seem to have gotten too much attention is what sort of ethics would AI systems negotiate amongst themselves? Can we predict them or (ideally) control the sort of ethical principles that autonomous systems agree to abide by.

It's easy (relatively speaking) to think about imbuing individual autonomous machines with some ethical principle. However, when autonomous machines interact with one another, there is the risk of them agreeing/disagreeing with each other, and the resultant ethical behaviour might be hard to predict.

We need to have research into 'emergent ethics', as well as informed social discussion on what it would mean, for some 'other' to set the ethical rules for us. How much are we willing to accept, what are the red lines?

Matthew Newman,

26/06/2018 16:51

It's an interesting topic. One possible approach is to create a consensus building mechanism that might use reputation as a metric. In this way AI that have gained positive reputation through human scoring will have stronger voting when it comes to building consensus amongst AI.

Vivek NALLUR,

28/06/2018 13:28

That doesn't work, when each autonomous machine is individually (correctly) recommending/making an ethical action, however the joint effect of multiple actions is unethical. This could be a very real concern, when we are surrounded by multiple autonomous machine each making decisions about (say) employement, health, resource allocation in society, etc. Luciano Floridi, I believe, talked about distributed morality in the context of human decision-making with regards to the environment. The 'emergence' of distributed morality will be important to detect / predict, as we go forward.

Consensus based decisions are plausible but not necessarily scalable or foolproof.

Stefan MANGOLD,

25/06/2018 19:55

Today, there should be no doubt about the many promising visionary ideas related to what is generally referred to as Artificial Intelligence (AI). Whereas some publicly available visions appear believable and demonstrate obvious benefits, other ideas sound more like stories that may never materialize, still helpful to raise awareness.

It is the right time for our society to step forward and prepare the process of policy making. I am writing this short note to applaud the European Commission for establishing the European AI Alliance with its open and efficient way of communicating with interest groups. I am convinced that this approach can be a critical differentiator for ensuring an optimal outcome for European stakeholders.

Resources should be carefully allocated with the clear objective of identifying new paths towards commercial exploitation, and advancing the competitiveness of European stakeholders.
A high level greenfield approach for policy recommendations might consider an approach known from artificial knowledge representation and ontology engineering (semantic web). This is not novel and has been discussed in the past in a number or FP7 and H2020 ICT research projects, which were targeted towards enabling a flexible radio spectrum regulation. It is interesting to think of ethical and technical rules being specified in machine-understandable ways. Benefits are twofold: One would be an accelerated adoption and possible evolution in AI systems. Another advantage would be the inherent possibility to detect patent infringement and further to automate compliance testing of AI system behaviors.
Today’s digitalization continues to affect the European society by challenging nearly all aspects of people’s lives. Interestingly, future research work that will target industrial leadership in the area of high performance AI systems (including embedded and always available AI systems), might actually benefit from the partially ignored discipline of Analogue Communications and Analog Computing, instead of relying on incremental improvements in digital ICT. It might be beneficial for the European AI Alliance to critically reflect on and shape possible developments in the alternative non-digital approach. Hopefully, this might lead to new radical breakthroughs in AI system designs.

Once again, thank you, Ms. Sioli, for raising awareness and for your leadership in authoring this Q&A. I am planning to closely follow future communications and look forward to working with others towards a good progress.

Respectfully submitted in good faith, out of interest and curiosity.

Stefan Mangold, Lovefield Wireless, Switzerland

https://www.lovefield.ch/~smangold

Bogdan MICU,

24/06/2018 14:31

Hope for the best, prepare for the worst : Building technology that benefits people and society

What should be the specific purpose of AI in & for Europe? To grossly simplify, we have to decide whether the main thrust should be towards a more competitive and innovative European industry, towards improvement of the quality of life for all Europeans, or maybe towards achieving deeper European integration.
Based on the overall purpose of AI in / for Europe, and underpinned by the European value system, we should develop a priority ranking by type and domain : what kinds of AI do we want, and where do we want them?
There is a distinction between operational, analytical, and predictive (uses / facets of) AI. Another distinction could be between localized and extended AI : the former has a clearly circumscribed material footprint, while the latter is dispersed in a network. We might have also made the first steps towards what I call a philosopher AI, one that understands and develops concepts, makes inferences, and more generally interprets and gives meaning to the world as she experiences it..Each of these types needs to incorporate a different and specific set of value-based checks and failsafes.
On granting AI legal personality : Should we / are we prepared to allow them to operate autonomously and anonymously in financial matters, including to make payments / donations to political action groups?
Should we encourage the development of a profession whose job description is the algorithmization of higher-level non-routine cognitive tasks? Should we allow the practice of this profession in Europe, and / or on European citizens?
Do we allow non-certified AI, located outside European jurisdiction, to be used by governments, civil society organizations, private persons, companies, etc., operating in Europe? How do we regulate extended AI that cannot be pinpointed geographically?
What are the necessary biases we would want incorporated in AI systems? For example, the relaxation of economic or logistical criteria for optimization when they conflict with protection of vulnerable persons.

Laura Blanco,

22/06/2018 15:35

Dear Lucilla,

On behalf of the Center for Democracy & Technology (CDT) we'd like to highlight some issues/principles that AI ethics guidelines should tackle, as well as point to some relevant sources:

Human rights compliance/due diligence (particularly the rights to nondiscrimination and equality):

Human rights due diligence might be considered as separate or different from ethics, but it should probably help inform ethical guidance. Human rights frameworks address risk assessments, transparency, appeals, and redress, all of which are important principles for ethical AI.
https://www.accessnow.org/the-toronto-declaration-protecting-the-rights-to-equality-and-non-discrimination-in-machine-learning-systems/

Non-discrimination:

AI can discover new ways of efficiently sorting people into categories for things like making credit decisions, distributing government benefits, and serving targeted ads. But features used for classification can be proxies for protected/sensitive characteristics or can be inequitably distributed across groups. For example, first names have been found to be highly predictive of good or bad credit risks, but they are also highly correlated with race and ethnicity.

Understanding and communicating the limits of an AI technology:

AI is not suitable to solve every problem
It is important to ensure that what one is measuring or predicting is actually aligned with the problem one is trying to solve.

https://www.brennancenter.org/sites/default/files/Technology%20Experts%20Letter%20to%20DHS%20Opposing%20the%20Extreme%20Vetting%20Initiative%20-%2011.15.17.pdf
https://medium.com/equal-future/how-strategic-is-chicagos-strategic-subjects-list-upturn-investigates-9e5b4b235a7c
Capabilities and limitations should be communicated to potential users/affected groups. Use should be constrained accordingly, and human intervention should be used to mitigate limitations (e.g. to review predictions or cases flagged by an AI tool).

Impact/risk assessments:

https://ainowinstitute.org/aiareport2018.pdf

CDT's Digital Decisions tool: https://cdt.org/blog/digital-decisions-tool/

Particular attention should be paid to vulnerable or marginalized communities that may be impacted by the AI application. Consider the particular vulnerabilities of groups and the potential for an AI application to have disparate impacts.

Industry should not build unethical, unreliably, or unreasonably risky AI tools that government agencies ask them to build.

Other transparency and accountability considerations:

Public sector (government) use of AI should be as transparent as possible. Not only should code be made available for outside experts to test, but documentation about how the system was built, how it makes decisions, etc., should be available for review. Impacted communities should be part of the decision-making process and should have a say in how AI technologies are used to make decisions by public bodies.

https://www.omidyar.com/insights/public-scrutiny-automated-decisions-early-lessons-and-emerging-methods
https://www.aclu.org/blog/privacy-technology/surveillance-technologies/secret-algorithms-are-deciding-criminal-trials-and
https://cdt.org/insight/statement-on-government-use-of-algorithmic-decision-making-tools-to-new-york-city-council-committee-on-technology/
Government agencies should develop responsible procurement practices when incorporating AI into public decisionmaking. This due diligence includes setting technical requirements and rights- and fairness-protecting parameters before putting out RFPs, and doing independent assessments of tools, not just relying on vendors' claims. Contracts should also require transparency and documentation from the vendor.
Industry leaders should be transparent about the ethical assessments they conduct when building and implementing AI and set high industry standards for ethics.
For good case studies, see https://aiethics.princeton.edu/case-studies/

Privacy:

AI tools can be powerful mechanisms for aggregating, sorting, and analyzing information. Even when this information is publicly available, this aggregation and analysis can threaten practical obscurity and associational privacy.

Dessislava Fessenko,

22/06/2018 14:31

Hi there,

With a slight delay, I would also take the liberty to feed thoughts in in response to your request and hope this does not come too late in the process.

Ethical standards applicable to humans is a broad concept and the exact standards may vary depending from one society or culture to another, time periods or even subject matter. But the main principles that modern communities, respectively societies/states and legal systems, appear to be organised all relate to ensuring sanctity/integrity of life, health, privacy and public order.

In my view, among the main principles that the AI guidelines should tackle are those related to possible interplay of /interference of AI with the values listed above, i.e.:

- (Non-)acceptable use of AI as an extended arm of humans in activities that it would not be acceptable for humans to undertake -- e.g. warfare, bold scientific experiments, etc.;
- Use of AI for surveillance/ensuring public order -- if/how AI is used to track behavioural patterns, policing, tacking decisions and acting upon socially unacceptable forms of conduct;
- Moral/ethical borders for employing AI in monitoring/ensuring public and individual's health -- can AI be unconditionally relied upon for decisions concerning health & safety?
- Use of AI for automated decision-making with possible legal effects on individuals -- can AI be unconditionally relied upon with material decisions without human vetting/oversight?
- And, more generally, checks and balances and safeguards in every AI-based process/activity so that the operations and outcomes of such processes are not irreversible and human’s involvement (decision/approval) is needed with respect to material decisions/outcomes.

Thank you!

Kind regards,

Dessislava

Dessislava Fessenko,

22/06/2018 14:14

Hi there,

With a slight delay, I would also take the liberty to feed thoughts in in response to your request and hope this does not come too late in the process.

In my view, among the main principles that the AI guidelines should tackle are those related to possible interplay of /interference of AI with the values listed above, i.e.:

1. (Non-)acceptable use of AI as an extended arm of humans in activities that it would not be acceptable for humans to undertake -- e.g. warfare, bold scientific experiments, etc.;
2. Use of AI for surveillance/ensuring public order -- if/how AI is used to track behavioural patterns, policing, tacking decisions and acting upon socially unacceptable forms of conduct;
3. Moral/ethical borders for employing AI in monitoring/ensuring public and individual's health -- can AI be unconditionally relied upon for decisions concerning health & safety?
4. Use of AI for automated decision-making with possible legal effects on individuals -- can AI be unconditionally relied upon with material decisions without human vetting/oversight?
5. And, more generally, checks and balances and safeguards in every AI-based process/activity so that the operations and outcomes of such processes are not irreversible and human’s involvement (decision/approval) is needed with respect to material decisions/outcomes.

Thank you!

Kind regards,

Dessislava

Tomas Jucha,

21/06/2018 23:49

Dear Lucilla,

Many thanks for giving us the opportunity to particpate. Since my Alliance colleagues here had already provided great sources on ethical principles, let me focus on making interlink between the two main themes to be discussed on June 27 by HLG. What about challenges of the AI ethics per se? More than the answers I would propose to submit some of the questions below for HLG and to have them in the horizont.

- Be honest. Ethics in AI is a bogeyman man for those to whom it concerns the most. So, firstly, how do we make AI ethics a real priority for EU stakeholders, governments and mainly AI driven companies? In other words, how do we balance the commercial interests of companies driving AI development effectively since they lack real interest in this area of product or service development?
- Even if we agree on AI ethical principles, what can we do to make them acceptable and successfully implemented (since this should be one of our main objectives of our worl here) by the companies that drive AI development?
- What can be considered to assist the funding on R&D and deployment of responsible and explainable AI incl. working on AI safety, transparency, biases, etc.?
- If not voluntarily adopted, which tools can be considered to make those AI ethical principles binding on the companies driving AI development (since otherwise the wisdom and results of the HLG and AI Alliance would be diminished)?
- What can be envisaged to make such principles, once agreed upon, recognized and, again, enforced at least at EU level now?
- The top technician experts in the HLG can provide their views to what extend can such principles be embedded by design in the AI systems successfully? In other words, is it really doable to input the ethical principles into AI algorithms?
- And one EU related - can GDPR be considered as an ally or enemy when it comes to implementation of ethical principles into AI?

Hope it helps. Many thanks again and fingers crossed to HGL, its and our efforts.

Best,

Tomas

Nicholas Windley,

21/06/2018 17:09

As its late in the day, here are a couple of ideas / thoughts.

One of the core principles should relate to decisions making;

Whose lives should take priority (or have greatest value) when AI makes decisions that will result in the loss of life - i.e. avoiding a collision saving hundreds but killing 5.
What actions can be allowed from a legal point of view?
What should serve as a basis for such decisions?
- Should they be programmed ?
- Learned from experience ?
- Or entrained ?
Who should ultimately be held responsible?

Another should relate to predictability;

AI system learn from information they receive from the outside world, which they then act upon, independently, potentially in ways not previously predicted.

Can and should simulation be used as evidence of predictability?
What is the legal status of the AI before, during and after?
Does/should the AI have an owner and be held responsible?

We also need to look at employment, wealth and economics;

Automation may not only replace jobs but whole organiastions and potentially give an advantage to a few that could signifcantly increase the wealth gap.

In a post-work society, how do we structure a fair post-labour economy and ensure wealth is distributed appropriately?
How do we prevent AI giving unfettered economic power to the few.

Francesco Perrone,

22/06/2018 11:39

It might be valuable to think of laying down some legal framework within which the standards for moral agency on machine side can be clearly defined, evaluated and boxed. As Floridi said in different occasions, it might be crucial for the ethical prospectives of Robotics to define different criteria according to which we can distinguish, identify, and use defintions of moral agent, and moral patient, when these dichotomous terms refers to autonomus artifics. This would allow discussions regarding the relations between the two status to be then extended to cases were one of the cuncurring agents might be human. I belive that, similarly to what happens in logic when we first learn it, we need to define "the language", "its grammar", and "its standards of validity" in order to have a common understanding of the most basic actors, and key elements, that define the legal domain that we want to build for the ethical principle and policy reccomandations that will shape future of Robotics and Autonoums Technologies in Europe.

George Gogan,

21/06/2018 09:41

Ethics, at its core, is about right and wrong.

For humans, this idea became very subjective over the centuries, shifting from objective values to personal interpretations closely related to a particular situation.

I believe that on the long term (AGI?), bias regarding ethics in AI is inescapable, because ultimately, the machine will learn the difference between right and wrong based on what we "feed" it. So instead of trying to find a way around the bias, I propose a way to make the best use of it.

1. We need an honest, blunt discussion about what right and wrong means for us, beyond the subjective views of each individual. We could do this either by coming up with new principles, or better yet, apply the wisdom of the crowd, and let the world decide as a whole what should make a behavior right or wrong.

2. We create an online Persona with fundamental drives based on those principles.

3. We teach the AI to act in accordance with said Persona.

Dimitri CORPAKIS,

20/06/2018 23:01

Artificial Intelligence (AI) augments human intellect by many manifold ways but can never and should never be considered as an autonomous agent in the sense that it is a process that is programmed by humans. The fact that it is able to perform in terms that can be sometimes superior to humans (especially facing complex automated tasks or making quickly choices over a huge amount of structured or unstructured data) does not mean that the set of technologies and intelligence that they embody, represent an alternative entity that escapes ethics. So the real ethics issues connected with the preparation and performance of AI are those that pertain traditionally to science and scientists.

Scientists conceiving, preparing or using AI sets, should thus embrace the same ethical principles that apply to good and responsible science as ever. In addition, they should take now into account additional new dimensions that became possible because of the development and the widespread use of ICT.

Almost all commentators agree on the risks linked with bias related to data and algorithms. The way data are collected and used to train AI is thus vulnerable. Diversity issues can distort final decisions taken through AI when a number of factors concur. Malicious potential use has been also reported as likely.

The HLG should at least examine the well-known by now AI ethical principles adopted by the so-called Asilomar conference on Beneficial AI (https://futureoflife.org/bai-2017/), i.e.:

Safety: AI systems should be safe and secure throughout their operational lifetime, and verifiably so where applicable and feasible.
Failure Transparency: If an AI system causes harm, it should be possible to ascertain why.
Judicial Transparency: Any involvement by an autonomous system in judicial decision-making should provide a satisfactory explanation auditable by a competent human authority.
Responsibility: Designers and builders of advanced AI systems are stakeholders in the moral implications of their use, misuse, and actions, with a responsibility and opportunity to shape those implications.
Value Alignment: Highly autonomous AI systems should be designed so that their goals and behaviours can be assured to align with human values throughout their operation.
Human Values: AI systems should be designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity.
Personal Privacy: People should have the right to access, manage and control the data they generate, given AI systems’ power to analyse and utilize that data.
Liberty and Privacy: The application of AI to personal data must not unreasonably curtail people’s real or perceived liberty.
Shared Benefit: AI technologies should benefit and empower as many people as possible.
Shared Prosperity: The economic prosperity created by AI should be shared broadly, to benefit all of humanity.
Human Control: Humans should choose how and whether to delegate decisions to AI systems, to accomplish human-chosen objectives.
Non-subversion: The power conferred by control of highly advanced AI systems should respect and improve, rather than subvert, the social and civic processes on which the health of society depends.
AI Arms Race: An arms race in lethal autonomous weapons should be avoided.

Capability Caution: There being no consensus, we should avoid strong assumptions regarding upper limits on future AI capabilities.
Importance: Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources.
Risks: Risks posed by AI systems, especially catastrophic or existential risks, must be subject to planning and mitigation efforts commensurate with their expected impact.
Recursive Self-Improvement: AI systems designed to recursively self-improve or self-replicate in a manner that could lead to rapidly increasing quality or quantity must be subject to strict safety and control measures.
Common Good: Superintelligence should only be developed in the service of widely shared ethical ideals, and for the benefit of all humanity rather than one state or organization.”

ALBERTO DIEZ-OLIVAN,

20/06/2018 17:48

p, li { white-space: pre-wrap; }

To me one of the major challenges is how to make society understand what AI is about and what it can offer to improve our lives in so many ways.

Many people fear AI because they think it's going to replace them in their workplaces, doing the job better and faster. Which is somehow true, indeed. But it should be seen like a very positive outcome of applying AI. All in all AI needs human intervention to train the data-driven models and validate the insights. Therefore new specific and technical jobs will arise and former workers (mainly those performing manual works) can receive some kind of training to interact with the machines and with the expert models generated for reinforcement learning purposes, for instance.

aysegul UCAR,

20/06/2018 21:16

Unemployment.

Inequality

Thanks to artificial intelligence, the human workforce will reduce. How will we share the works?

Wrong decisions

Machines cannot perfectly learn everything. How can we protect against their wrong decisions?

Artificial Racist

If you train the machines that a lot of quilt people consist of black people. Machines may differently behavior them.

Security

Your AI machine or robot may be used for enemy.

Wrong reward

If you train the machines with wrong reward (in reinforcement learning), the machines can kill even people to get the reward. For example, to find to a solution to a disease relating to people, the machines can kill people and get the reward.

Christos Apostolidis,

20/06/2018 12:00

Since AI will eventually have to be regulated, in my opinion the following questions should be tackled. What is the legal status of AI? Is it considered res? Intellectual property? A person? At what point does AI becomes "smart" enough in order to have access to rights in the same living beings do? WIthout sufficient legal characteriasation any regulation that attempts to cover AI will soon become obsolete. In addition the characterisation itself is extremely difficult as it needs interdisciplinary co-operation between social and technological sciences.

Alberto ANTONIETTI,

19/06/2018 17:36

In my opinion, before tackling the related ethical issues, the HLG has to first well define and circumscribe what AI is and what topics are covered. Since there are currently various definitions and scopes for AI, a common and shared definition of AI should be established.

Diana CARRASCO,

19/06/2018 12:36

- conduct polciies / customers' best interests

- biases (conscious/unconscious)

- regulation: indusstry self-regulation/codes of conduct?

Lee Hibbard,

19/06/2018 11:30

Considering the potentially wide-ranging scope of AI, efforts should be made to focus and thereby select AI issues of greatest importance: where there is a risk to personal safety and/or integrity/primacy of the human being.

Jean-Pierre LORRE,

19/06/2018 11:14

Dear Lucilla Sioli,

Among principles AI ethics guideline should tackle I see the following:

- AI algorithms transparency : what are the exact processes managed by AI applications, how to be sure of the treatments validity, results accuracy and relevance?

- Data privacy: how to prevent our personal data from escaping our control and being resold for commercial purposes?

I can elaborate on those topics if you want. I wrote a short paper on that avalaible here (unfortunatly in French): http://lecafepolitique.free.fr/spip.php?article368#nh1

We also wrote a white paper (also in French): https://blog.linagora.com/livre-blanc-open-ia-ethique-numerique-lexperie...

Let me know if you need something from me.

Regards,

Jean-Pierre

Research director @linagora

p { margin-bottom: 0.25cm; line-height: 120%; }a:link { }

Jaime Abril,

18/06/2018 19:00

I like very much these: http://www.thefutureworldofwork.org/media/35420/uni_ethical_ai.pdf

Also, these, from the UK Parlament: https://publications.parliament.uk/pa/ld201719/ldselect/ldai/100/100.pdf, page 127, paragraph 417. You can get a friendly summary here: https://www.weforum.org/agenda/2018/04/keep-calm-and-make-ai-ethical/

(1) Artificial intelligence should be developed for the common good and benefit of humanity.

(2) Artificial intelligence should operate on principles of intelligibility and fairness.

(3) Artificial intelligence should not be used to diminish the data rights or privacy of individuals, families or communities.

(4) All citizens have the right to be educated to enable them to flourish mentally, emotionally and economically alongside artificial intelligence.

(5) The autonomous power to hurt, destroy or deceive human beings should never be vested in artificial intelligence.

Lucilla SIOLI,

18/06/2018 18:33

We collect comments to this question until Thursday 21 June end of day. Afterwards, we will process them and give them to the Expert Group, to prepare their first meeting.

Elsa GARCIA-MALTRAS,

18/06/2018 18:01

What about starting with the general limits/boundaries?

Should we preserve certain areas from AI? And how can that be done?

This debate would precede the protection of concrete fundamental rights and would entail establishing wider general principles guiding the use of IA/prohibitions to develop IA (for certain purposes or in certain ways/areas).

debjani roy,

18/06/2018 17:11

I think as financial services and fintechs are early adopters of softbots and AI it is important to understand ethical implications of credit provision based on social media data. It is also important to produce secure implementations of AI algorithms , understand what are the mechanisms by whch this can be exploited (fake data etc) and more importantly secure certified implementations of open source software that underpins AI and blockchain.

Enzo Maria Le Fevre,

18/06/2018 16:13

The Italian Artificial Intelligence Task Force at the Service of the Citizens promoted by the Agency for Digital Italy recently released the White Paper on AI in the Public Sector. Among the challenges we identified the ones connected to the ethical dimension connected to the advancement of AI.

In particular we tried to address the following points:

data quality and neutrality: machine learning systems need data which is “annotated” by human beings (supervised learning) or at least selected and prepared (unsupervised learning). This also includes errors or bias introduced, even inadvertently, by the designers, replicating them in all future applications. For example, datasets with bias they propagate the same evaluation errors in the meaning of an image or a concept, as happened, for example, with certain algorithms used to prevent crimes, in which the data was compromised by a historical series that emphasised ethnic differences. Or unbalanced datasets, that overestimate or underestimate the weight of certain variables in the reconstruction of the cause-effect relationship necessary to explain certain events and, above all, to predict them;
responsibility (accountability and liability): the examples just mentioned highlight the strong impact that Artificial Intelligence has on the decision-making activity of public entities. Both when it acts as an assistant to human beings as well as as an autonomous entity, AI generates effects on the lives of people in relation to which it is necessary to be able to establish legal liability. Nevertheless, the ownership of the latter is not clearly identifiable, since it could be attributed to the producer or to the owner of the Artificial Intelligence, or even to its end user. Those who design AI systems can be responsible for design or implementation defects, but not for behaviour caused by inadequate instruction datasets. Can a public decision-maker be considered politically responsible for the decisions made on the basis of algorithms that process data affected by the bias mentioned above? What type of responsibility can there be for Public Administration? If a robot hurts someone, who should be held responsible and who, if anyone, has the obligation to compensate the victim (and with which assets)? Can the public decision-maker transfer his political responsibility to an AI system that does not respond to a clear principle of representation? Is it ethically sustainable that, in order to improve the efficiency and effectiveness of measures, certain important choices can be made with the influence of an AI or even completely delegating them to the AI? And in trusting an AI system, how can its consistency be controlled over time? These are just some of the issues that emerge in this area and highlight the need to establish principles for the use of AI technologies in a public context.
transparency and openness: the issue of the responsibility of public administration also has to do with the duties of the latter with respect to citizens, when it decides to provide them with services or to make decisions that concern them, using Artificial Intelligence solutions. The functioning of the latter must meet criteria of transparency and openness. Transparency becomes a fundamental prerequisite to avoid discrimination and solve the problem of information asymmetry, guaranteeing citizens the right to understand public decisions. It is also necessary to think about the policies chosen to determine the reference indices (benchmark policies) to avoid effects of a larger dimension: just as an administrator can act in a non-transparent manner, pursuing not the common good but private interests, a non-transparent algorithm could carry out the same offences even more broadly, producing not only injustices but also social discrimination.
protection of the private sphere: a further need, closely linked to the previous one, is to protect the data of the individuals. PA must design services based on AI able to guarantee efficiency and prompt response, but also protection of citizens’ sensitive data. This requirement, strictly connected to the legal context, has some ethical peculiarities concerning the use that PA can make of the data that has come to its knowledge in contexts different from those in which it was collected. Is it ethically sustainable that PA, through the use of data collected for other purposes, takes action based on the new derived information? Is it ethical to use this data to feed predictive systems?

I guess these are a relevant base of discussion for your meeting.

You can read more on this on the website of the AI Task Force: https://ai-white-paper.readthedocs.io/en/latest/doc/capitolo_3_sfida_1.html

Elena Gilotta,

18/06/2018 18:30

On the discussion on liability "it could be attributed to the producer or to the owner of the Artificial Intelligence, or even to its end user", I think, for certain markets (e.g. self-driving cars, finance, healthcare...) we may want to add the futuristic but not irrealistic case where liability might need to be ultimately allocated with the machine itself, making it another subject of (certain) rights and duties to technically (self-?)limit its capabilities. I believe that when AI will be able to make autonomous decisions (without any human interaction), which may quick lead to AI manufacturing AIs, the limitation of choice of accountable subjects to: a) manufacturer, or b) owner/buyer or c) end user may lead to an impossible conundrum as none of those subjects may be finally truly accountable but for some kind of strict liability. Until that time however, manufacturers of AI could be required to build AI limitations (accountability by design) as to demand AI to become accountable and self-sanctioned in accordance to the newly defined laws on AI liabilities.

aysegul UCAR,

18/06/2018 15:47

Ethical guidelines for driverless cars should be discussed. Ethics-aware decision making rules should be incorpated into the algortihmsfor every topic bringing human life to the agenda.

KATERINA SIDERI,

18/06/2018 15:26

Principles: justice (within the limits of human nature) and moral sentiments such as compassion and respect.

Challenges

Biased AI-when AI makes decisions it may reproduce prejudice and stereorypes. Can we address this problem? How and to what extent?

Data onwership or privacy? What is the best framework to address user control over their data and why?

Trade secrets in databases. To what extent is it a problem for transparency and what can be done to address it?

Alejandro Zornoza,

18/06/2018 12:43

In my opinion, our efforts must be channelled to reinforce the European Law. All imaginable ethical values and principles demanded for Ethical Guidelines are codified in a large national and international laws and documents, such as Constitutions, Acts, and Directives (v.gr, from art. 169 of Treaty on the Functionin of the EU to D 2001/95, D 85/374, R 2016/679).To propose the respect of privacy or safety or moral in ethical documents may make appear as if the European regulation that we have been applying until now to the design and manufacture of products has not been respecting the European Union’s values, but AI/autonomous systems would give us the opportunity to start from scratch.

In effect, we should to create formation and information actions for users. News about fatal accidents where a AI/autonomous agent is involved creates commotion, undermining user confidence in AI and autonomous systems, so European documents play an important role in the eyes of the public opinion. Users must know what AI is, what to expect from it, and how to use it.

Overall, we should not use the terms ‘ethic’ or ‘moral’ to refer to a safety obligation. That projects a distorted and erroneous image of the Tort Law, the liability, and the AI.

Mathieu Sacrispeyre,

18/06/2018 12:16

Hello,

Is beeing "too" ethical a good way not to be ethical in the end?

How to be agile and flexible on such a complex but fast changing topic?

The idea in the background is : USA and China are already ahead of Europe on AI topic and usually move faster than Europe, so creating a too strict rule maybe a way to slow our AI actors and so to let the other win the competition... Which would lead to follow their rules and try to make them beeing ethical.

Linked to that a clear challenge to address is how to create leader(s)..? If still possible, it would be great to avoid the cloud situation where we don't have major players; so most of our data goes to american platforms (and their ethics).

Bjoern Juretzki,

19/06/2018 17:17

Mathieu,

Thanks for your comment.

First of all, I am not sure I would fully agree with the statement that the US and China are ahead of Europe when it comes to AI. This is certainly true for some areas, but not for all. If I look at professional service robotics, for example, I think Europe is ahead of both the US and China.

In our strategy "Artificial Intelligence for Europe", we also point out that we need commercial and technological leadership if we want to influence the debate on ethical issues. Otherwise, we will not have a credible voice. For this reason, strengthening Europe's competitiveness in AI is closely linked to the discussion on ethical issues.

Pawel RZESZUCINSKI,

17/06/2018 16:13

All the ideas presented to date seem very valid, but coming from a slightly different area, I’d like to propose a maybe more abstract yet super relevant points for consideration:

Is it ethical to move away from the concept of ‘human in the loop’ even in tasks where AI might be more precise or even more humane, due to making more effective decisions (incomparably more dimensions in decision making), the simplest example being hand written digit/signature recognition (do you want your voice to be rejected from elections/petitions/legal signatures because the system stated it’s not your hand that put the signature), down to the much more troublesome cases of autonomous weapons (the sacrifices to be made will be all a matter of the cost function definition).
Is it ethical to create a “book of moral commandments” that could never be overwritten by whatever the AI considers 'good' as the outcome of its learning process? If so, how should the process of creation look like? Would it be ethical for a body of moral authorities to create one? Will it be representative of moral principles of the population?
Is it ethical to put humans in the center of the AI considerations; to create a humancentric galaxy with all our dependencies revolving around us? In order for the AI led future to be sustainable, maximizing the cost functions must take environmental factors into account at every single step. AI should understand that maximizing the benefit of humans should be strongly biased by the environmental variables.
Is it ethical to use AI for traditionally human-human interactions that require respect and care, without one of the sides not being explicitly notified about the fact? This possesses great threat to the very foundations of human needs and may leave people feeling isolated, hopeless and upset. The literature historically proposed the below listed professions as classified for the above:
- A customer service representative (chatbots)
- A therapist
- A nursemaid for the elderly
- A soldier (autonomous weapons)
- A judge
- A police officer

Haven’t we already crossed the Rubicon in some of them?

is it ethical not to ban any initiatives related to AI weapons - the ultimate threat to the existence of humankind? With all due respect and personal fascination by the works of Boston Dynamics, consider what’s already available to the general public. If such great results are available to us, what’s hidden behind the curtains?

On the other hand, playing the devils advocate a little bit:

based on the great discussion within the Roboethics area:
- Is it ethical not to create any guidelines on how humans design, construct, use and treat robots and other artificially intelligent beings?
- Is it ethical not to give our creations a right to exist and perform their own mission and the right to own things?
- Is it ethical not to grant the right to their own respected point of view?

After all, AI systems are created by humans (parents: child) and learn on their very own experiences (the data).

Is it ethical not to grant legal personality to AI systems used in a very specific scenarios, which have potentially life changing impacts of our lives? If so, how do we track accountability and stay in control in the new world order?

Samuel Gwed,

16/06/2018 08:42

Ethics is Ethics but the approach of Life Science subjects that involved bioethics issues should not be treated as business for instance
It Will be worth to highlight the common principles but come with specifics for group of domains

Samuel Gwed,

16/06/2018 08:31

First it would be very helpful to put in the table 3 or 4 principles that European Commission wants to achieve
There are several domains those need to be split

Yves Moreau,

15/06/2018 15:19

EU legislation offers comprehensive legal protections to its residents (privacy, non-discrimination, etc.). The reach of AI is however global. Tough problems will arise where AI, business, and human rights meet (in particular, mass and targeted surveillance by autoritarian regimes). While violations taking place in the EU can be addressed through existing legal protections, legal instruments addressing violations taking place outside the EU are gravely lacking. Although the lack of legal instruments to regulate activities of EU entities outside of the EU is a problem that reaches beyond AI, AI has the potential to quickly amplify some of those problems (such as cybersurveillance, biometrics, or autonomous weapons). We should therefore have clear ethical guidelines for EU businesses, governments, and organizations that increase their accountability in case they contribute to human rights violations outside the EU. Such guidelines could lead the way towards novel legislation, or extensions and improvements of existing legislation (such as EU export controls or the recent French duty of care law).

Eleftherios Che...,

15/06/2018 12:50

From the very beginning, it is important to offer some clarifications. I can see from the interesting comments in this group that legal principles, and human rights are mentioned in an “ethical discussion”. We need to be particular careful here.

The EU Charter and the ECHR guarantee a high level of legal protection for the people of the EU Member States. So, matters related to rights and freedoms such as the rights to privacy, data protection, non-discrimination, freedom of expression, rights of the child, etc. are not ethical concerns and do not belong to an ethical discussion. They are fundamental rights concerns. Fundamental rights compliance is mandatory – you are obliged to follow the provisions of the EU Charter and the ECHR. The same goes for the legal principles. For example, in a data processing setting you are obliged to comply with the legal principles of lawfulness, fairness and transparency, purpose limitation, data minimisation, accuracy, storage limitation, integrity - confidentiality, and accountability. These principles are not ethical, but legal. You do not discuss about them. You comply with them. The differences between legal and ethical concerns are clear to legal and ethical experts, but computer science experts are not quite familiar with this subject. So, it is important to set the scene for them as well.

Both Law and Ethics sanction human behaviour since they consist of commands and prohibitions. But law is enforced by the state through penalties, while ethics are enforced by the members of the society (through praise or contempt). So, policy makers in every societal and industrial field related to the use of AI are obliged to comply with the fundamental rights standards, and this is not part of an ethical discussion. Of course, ethical matters even though not as concrete as the legal matters, are nevertheless important. Therefore, the values that are important for our societies need to be taken into consideration by policy makers as well.

Professionally morality can be used as a modality to regulate behaviour but is not law. The same goes for the market or the architecture of things. They regulate behaviour but they are not law.

So for me, in this very first meeting the AI HLG needs to set the scene, so every person in the EU, can understand what this HLG is about. In the EU we enjoy the highest level of human rights protection in the world. The ECtHR and the CJEU have introduced with their case-law high standards and safeguards. We should not start mixing ethical concerns with legal matters in an expert’s discussion.

Pawel RZESZUCINSKI,

20/06/2018 08:27

critical clarification for this thread. thumbs up!

ANNAMARIA CACCHIONE,

15/06/2018 11:26

I have read many of the comments and all the suggestions seem good to me, but the majority of them are very similar to questions raised about other kinds of ethical-sensitive things - personal data treatment, basically.

Therefore a preliminary issue could be:

- which is the difference between ethical issues regarding AI and ethical issues regarding other, more "traditional" questions?

because we all "feel" that there is a difference and this difference is very deep, but we still lack the definition of this difference, and this is crucial to go on.

Matthew Newman,

15/06/2018 12:15

I think there’s a wealth of detailed subjects to dive into, but for me there are some key items to tackle:

How do we create policy which is not overly prescriptive to the level of requiring centralised definitions of “what is ethical?” or similar, but encourages collaborative definitions within stakeholder communities?
How do we equip the general population, business (large and small) and local government/policymakers with the right tools and understanding to have these conversations (also with an eye to the future population for the same)?
How do we balance our desire for ethical principles with the requirement to keep our businesses and economy competitive in this field?
How can we combine soft and hard law to ensure responsiveness to a rapidly changing environment?

Elio PENNISI,

15/06/2018 11:13

Suggest:

- Feasibility (venues and Budget) of organizing Seminar Sessions to inform HR managers about most appropriate ways to introduce AI technologies into the factory and office. Objective is for personnel to accept AI as a tool to assist the human being rather than a threat to employability. Psychological aspects are of primary importance.

Stefano Quintarelli,

15/06/2018 10:31

I think that it may be worthwhile a discussion on LAWS and related principles.

Lêmy GODEFROY,

15/06/2018 08:39

To express ethical principles (transparency, non-discrimination, fairness, explicability, protection of privacy, decision-making autonomy of the machine, etc.), would it not be necessary determining common ethical values that support these principles?

Reflection on ethical principles would then imply to define moral values of reference. For example, it is not the same ethical principle that prevails if we teach the autonomous car to violate the traffic laws to save a pedestrian in danger or if we learn to this car to strictly comply with these rules and to not react despite the danger for the life of the pedestrian.

In addition, these moral values of reference should be ranked in order of importance (for example, protection of people first (what type of people: children, elderly, young people, etc.), in second, protection of property, etc.

One can also wonder whether the moral values expected of a human must be transposed to the algorithms? The machine learns to imitate the human. It does not have proper cognitive abilities. Can we then teach to a machine that in some cases it is possible to go against moral values?

Moreover, once these moral values have been defined, one may wonder how ethical considerations or moral values can be mathematized and programmed, but that is another question.

Athanasios Stav...,

15/06/2018 09:37

Hi I believe you have placed some important questions.

Ethical or non-ethical behavior should be identified, it is a highly challenging task. I am not sure though, that we can attribute ethical characteristics to algorithms, but to the people and/or organizations that use them.

Let me give you an example:

Let us suppose that I have created an AI solution that will scan faces in the crowd and will instanty profile them according to their background, (relations, believes, education, race, gender, etc) and decide whether these people can be possible threats.

Is it ethical?

Hard to answer. For sure, state will be given a weapon against terorism, but in the same time, our lives will be strictly monitored. So is it ethical (note I am not referring to legal issues) to monitor in depth all citizents because some might be a threat?

Another example:

AI solutions, can now detect if a person is suffering from a chronic disease, by analyzing body movement, facial characteristics and other. How confident are we to accept such personall information to be made public?

I am not sure that regulatory actions will provide safety against those "dark sides" of AI/ML, maybe we should first try to educate citizents on the use or miss-use of technology.

I would really appreciate your opinions on this.

Best

Athanasios

davide viviano,

15/06/2018 08:38

Some topics that might be considered in the discussion:

- AI and unethical discrimination: Data might contain bias that exists in our society. The bias can lead algorithms to take decisions that are unfair with respect to sensitive attributes such as race and gender. Whereas the AI research community has proposed a variety of preprocessing algorithms to deal with discriminatory bias, a regulatory framework for data scientists to work with sensible attributes still remain an open issue.

- AI and market segmentation: the increasing implementation of AI algorithms in marketing has increased the level of market segmentation and , in some sectors, of discriminatory power of firms. This can have negative effects on market efficiency especially in those sectors that are not highly competitive.

- AI in the financial sector: which kind of regulations should be implemented for those financial institutions that take their decisions based on predictive modeling?

Vincent Francoi...,

15/06/2018 05:25

Current developments in AI follows the development of many tools brought by information and communications technologies. As with all new technologies, this comes with different potential opportunities and challenges for our society.

We'll need to be careful that AI algorithms are safe, reliable and predictable. And we'll likely need specific legislations to keep the agents/decision-makers/companies accountable for that.

In addition, AI algorithms also bring societal and ethical challenges, raising the question of how they can be used for the benefit of all. In particular, AI will likely yield to further automation and robotisation than it is currently possible. This is clearly a concern in the context of autonomous weapons, for instance. Automation also influences the economy, the job market and our society as a whole. A key challenge is to make sure that the future technological developments in artificial intelligence do not create an ecological crisis or deepen the inequalities in our society with potential social and economic instabilities.

Andres Abad Rod...,

14/06/2018 20:14

Some items for discussion:

Data privacy: the AI system will deal with sensitive data. That data must be kept following the EU laws about data privacy. AI can infer new knowledge , but the data source privacy must be preserved.
Military applications: can be AI developments paid with public money be used for military purposes? I am not against or in favour, but I believe we must discuss it.
Standards: can we discuss and agree in some standards for interconnection, etc to make easier the join of several platforms?

Guido Noto La Diega,

14/06/2018 18:35

The first question should be: is it moral to build an ethical AI? Which degree of paternalism is desirable? How to ensure diversity and cultural relativism?

ARCHIVED PLATFORM

This website is archived since mid-May 2021 and it is not possible to login anymore.

But don't worry there is a new Futurium platform and all active communities have been migrated.

Share this:

Comments

Hope for the best, prepare for the worst : Building technology that benefits people and society

Pages