Ethical Frameworks for Trustworthy AI – Opportunities for Researchers in Human and Social Sciences

Karim Baïna, 3rd Edition of the Spring School for Doctoral Students of the Human and Social Sciences, Rabat, May 28-30, 2025, Morocco 1 / 56
Ethical Frameworks for
Trustworthy AI –
Opportunities for Researchers
in Human and Social Sciences
Prof. Karim Baïna
karim.baina@ensias.um5.ac.ma
Full Professor
ENSIAS, Mohammed V University in Rabat, Morocco
Leader of Alqualsadi Researh Group on Digital Innovation and Enterprise Architectures
Rabat IT Center, ENSIAS

2 / 56
Karim Baïna, 3rd Edition of the Spring School for Doctoral Students of the Human and Social Sciences, Rabat, May 28-30, 2025, Morocco
Outline
1. Introduction
2. Terminology
3. Ethics in AI, Ethics for AI or Ethics of AI ?
4. Ethics in AI Frameworks
5. Critical Perspectives on AI Ethics
Frameworks (Ethics of Ethics in AI)
6. Current Landscape of AI & Ethics in Morocco
7. Ethics and Trust in AI: Which Opportunities
for Researchers in the Human and Social
Sciences ?

Introduction
Prof. Karim Baïna
Full Professor

4 / 56
« Ethics without justice is just white
supremacy. » Timnit Gebru
ex-Google Ethical AI

5 / 56
« We cannot delegate moral
responsibility to machines. It is always
humans who are accountable. » Virginia
Dignum
Member oh EU HLEG

6 / 56
« Ethics becomes a tool for self-
regulation, and even worse, for
protecting companies against criticism.»
AI Ethics (book, MIT Press, 2020), Mark
Coeckelbergh
Prof. of Philosophy of
Technology, University of
Vienna
Ethics washing
https://www.cursor.tue.nl/en/news/2023/juni/week-3/beware-of-ethics-washing-warns-the-young-academy/

7 / 56
AI Incident Database (AIID) tracks instances of
ethical misuse of AI, such as autonomous cars
causing pedestrian fatalities or facial recognition
systems leading to wrongful arrests.
Ethical Incidents

8 / 56
https://www.sap.com/resources/what-is-ai-bias
Terminology – AI Biases
• AI bias, refers to systematic discrimination embedded within
AI systems that can reinforce existing biases, and amplify
discrimination, prejudice, and stereotyping.
• AI Bias typically arises from two sources:
●
the training data used by AI models embedding patterns and
correlations within this data to make predictions and decisions.
●
the design of AI models themselves : reflect the assumptions
of the developers coding them which causes them to favor
certain outcomes.
https://link.springer.com/article/10.1007/s44206-024-00142-x

9 / 56
https://research.google/blog/introducing-the-inclusive-images-competition/
AI Biases (Cutural Bias)

10 / 56
https://www.scalablepath.com/machine-learning/bias-machine-learning
AI Biases (Gender Bias)

11 / 56
https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing
AI Biases (Racial Bias)
Robert Julian-Borchak Williams was
wrongfully accused and arrested
due to the use of AI facial
recognition.
https://www.nytimes.com/2020/06/24/technology/facial-recognition-arrest.html

12 / 56
https://www.youtube.com/watch?v=6b9t8LOk-Lg
before after
AI Biases (Socio-economic Bias)

Terminology
Prof. Karim Baïna
Full Professor

14 / 56
Terminology – Cyberattacks
Manipulating AI Systems Behavior
●
Poisoning attacks occur in the training phase by introducing
corrupted data.
●
Evasion attacks occur after an AI system is deployed, attempt to
alter an input to change how the system responds to it.
https://www.nist.gov/news-events/news/2024/01/nist-identifies-types-cyberattacks-manipulate-behavior-ai-systems

15 / 56
Poisoning attack

16 / 56
NOISE
Evasion attack

17 / 56
●
Privacy attacks occur during deployment, are attempts to learn sensitive information
about the AI or the data it was trained on in order to misuse it (to reverse engineer the model
so as to find its weak spots — or guess at its sources)
●
Abuse attacks involve the insertion of incorrect information into a raw source, such as a
webpage or online document, that an AI then absorbs
– Unlike the aforementioned poisoning attacks, abuse attacks attempt to give the AI incorrect
pieces of information from a legitimate but compromised source, while poisoning targets
training data itself)
https://www.nist.gov/news-events/news/2024/01/nist-identifies-types-cyberattacks-manipulate-behavior-ai-systems

18 / 56
Trustworthy AI has three components, which should be
met throughout the system's entire life cycle:
1. Lawful AI : complying with all applicable laws and
regulations;
2. Ethical AI : ensuring adherence to ethical principles and
values; and
3. Robust AI : both from a technical and social perspective,
since, even with good intentions, AI systems can cause
unintentional harm.
Terminology –
Responsible/Trustworthy AI

19 / 56
Terminology – Explainability
• The question of why an AI model made a particular
decision (e.g. denying a loan application). AI-powered
decisions should be accompanied by the reasons and
explanations for the decisions (User-level understanding)
https://www.darpa.mil/program/explainable-artificial-intelligence
https://www.ibm.com/impact/ai-ethics

20 / 56
Terminology – Fairness
•The issue of treating different groups equally,
such as different genders having similar loan
approval rates in a training data set or in the
recommendations of an AI model.
Robert Julian-Borchak Williams was
wrongfully accused and arrested
due to the use of AI facial recognition.

21 / 56
•People need to be able to understand and
evaluate how the systems work and know their
strengths and weaknesses (System-level
visibility).
www.holisticai.com
www.fca.org.uk/insight/ai-transparency-financial-services-why-what-who-and-when
Terminology – Transparency

22 / 56
• People deserve to have their sensitive data protected : Only
necessary data should be collected (proportionality principle), and
consumers should have clear access to controls how and when their
data is being used, and what it is being used for and by whom.
●
PII (Personal Indentifiable Information) : information that can identify an individual,
when used alone (directely) or with other relevant data (indirectely) : name, surname, e-
mail, phone number, photo, IP address, postal address, cookies, etc.
●
SPI (Sensitive Personal Information) : Personal data that reveals the racial or ethnic
origin, political opinions, religious or philosophical beliefs, or trade union membership of
the data subject, or that relates to their health, including their genetic data : government-
issued identifiers (SSNs, driver's license numbers, passport details, ID), medical analyses,
political opinions, sexual orientations, biometric data (digital print..), etc.
●
User Experience
Security
Transparency
Control
PII/SPI
Terminology – Privacy

23 / 56
•Resistance to adversarial attacks, that target
the weakness specific to AI systems, without
causing serious harm)
NOISE
hackernoon.imgix.net
Terminology – Robustness

24 / 56
•The degree to which an AI system’s processes
and decisions can be systematically
examined, traced, and verified, often after the
fact through logging, version control,
reproducibility, independent review, model cards,
traceable pipelines, explainability tools.
Terminology – Auditability

25 / 56
•The assignment of responsibility for the
behavior and impacts of AI systems, including
consequences and governance through legal
liability, responsibility chains, ethical ownership.
Accountablity can be moral, legal, social,
institutional, and technical.
Terminology – Accountability

Ethics in AI, Ethics for AI or
Ethics of AI ?
Prof. Karim Baïna
Full Professor

Ethics in AI, for AI, of AI
●
Ethics in AI « Computational Ethics »
« Ethics by Design » (ethics as embedded
values & Principles) : refers to the technical
and computational implementation of ethical
considerations within AI systems, including
the formalization of values like fairness,
privacy, or accountability.
– e.g. Embedding a fairness constraint in a
classification algorithm to avoid
discrimination against protected groups
"The Ethics of AI Ethics: An Evaluation of Guidelines", Thilo Hagendorff 2020

●
Ethics for AI « Machine Ethics » « Ethics
in Design » (ethics as reasoning
capability) : investigates how to construct
machines (especially autonomous agents
or robots) capable of making moral sound
decisions, reasoning about ethical
dilemmas (like moral agents)
– e.g. Programming a self-driving car to
weigh harms in an unavoidable crash
scenario (value-sensitive design).
The Nature, Importance, and Difficulty of Machine Ethics James H. Moor, Dartmouth College 2006

●
Ethics of AI « Ethics for Design » (ethics as external evaluation and
critique) : is Concerned with the external societal, political, and philosophical
evaluation of AI’s impact on individuals, institutions, and societies — and whether
their development and deployment are morally acceptable.
– Metaethics (Philosophy of Mind, Moral Ontology) *
●
Is AI capable of moral reasoning?
●
Can machines be moral agents or patients?
– Normative Ethics (Moral Philosophy (Kantian, Utilitarian, etc.)) **
●
Deontological: Are we obligated to ban Lethal Autonomous Weapons ?
●
Consequentialist: Will AI increase or decrease social welfare ?
●
Virtue Ethics: What kind of society does AI foster ?
– Applied Ethics / Technoethics / Data Ethics ***
●
Data ethics (e.g., surveillance capitalism)
●
Ethics of automation and labor
●
Ethics of decision delegation (e.g., autonomous cars, predictive policing)
* Turing (1950), Moor (2006), Floridi (2006)
**- Wallach & Allen (2008), Bostrom (2014)
*** Zuboff (2019), Jobin et al. (2019), Dignum (2018)

Ethics in AI Frameworks
Prof. Karim Baïna
Full Professor

31 / 56
UNESCO 10 core principles
of Ethics of AI -
a Human-Rights centred approach
●
1. Proportionality and Do No Harm : The use of AI systems must not go beyond
what is necessary to achieve a legitimate aim. Risk assessment should be used to
prevent harms which may result from such uses.
●
2. Safety and Security : Unwanted harms (safety risks) as well as vulnerabilities
to attack (security risks) should be avoided and addressed by AI actors.
●
3. Right to Privacy and Data Protection : Privacy must be protected and
promoted throughout the AI lifecycle. Adequate data protection frameworks should
also be established.
●
4. Multi-stakeholder and Adaptive Governance & Collaboration : International
law & national sovereignty must be respected in the use of data. Additionally,
participation of diverse stakeholders is necessary for inclusive approaches to AI
governance.
●
5. Responsibility and Accountability : AI systems should be auditable and
traceable. There should be oversight, impact assessment, audit and due diligence
mechanisms in place to avoid conflicts with human rights norms and threats to
environmental wellbeing.

32 / 56
UNESCO 10 core principles
of Ethics of AI -
a Human-Rights centred approach
●
6. Transparency and Explainability : The ethical deployment of AI systems
depends on their transparency & explainability (T&E). The level of T&E should
be appropriate to the context, as there may be tensions between T&E and
other principles such as privacy, safety and security.
●
7. Human Oversight and Determination : Member States should ensure that
AI systems do not displace ultimate human responsibility and accountability.
●
8. Sustainability : AI technologies should be assessed against their impacts
on ‘sustainability’, understood as a set of constantly evolving goals including
those set out in the UN’s Sustainable Development Goals.
●
9. Awareness & Literacy : Public understanding of AI and data should be
promoted through open & accessible education, civic engagement, digital skills
& AI ethics training, media & information literacy.
●
10. Fairness and Non-Discrimation : AI actors should promote social justice,
fairness, and non-discrimination while taking an inclusive approach to ensure
AI’s benefits are accessible to all.

33 / 56

34 / 56
EU AI Act
Risk-based approach
Connecting the dots in trustworthy Artificial Intelligence: From AI principles, ethics, and key requirements to responsible AI
systems and regulation, Natalia Díaz-Rodríguez et al. Information Fusion’2023

35 / 56
OECD 5 Values-based
AI principles
●
Principle 1.1. Inclusive growth, Sustainable development and Well-Being : Principle
highlights the potential for trustworthy AI to contribute to overall growth and prosperity for
all – individuals, society, and planet – and advance global development objectives.
●
Principle 1.2. Respect for the rule of law, human rights and democratic values,
including Fairness and Privacy : AI systems should be designed in a way that respects
the rule of law, human rights, democratic values and diversity, and should include
appropriate safeguards to ensure a fair and just society.
●
Principle 1.3. Transparency and Explainability : This principle is about transparency
and responsible disclosure around AI systems to ensure that people understand when they
are engaging with them and can challenge outcomes.
●
Principle 1.4. Robustness, Security and Safety : AI systems must function in a robust,
secure and safe way throughout their lifetimes, and potential risks should be continually
assessed and managed.
●
Principle 1.5. Accountability : Organisations and individuals developing, deploying or
operating AI systems should be held accountable for their proper functioning in line with
the OECD’s values-based principles for AI.

36 / 56

37 / 56
Impact of AI Regulations on
Responsible AI decisions

38 / 56
Government AI Readiness
Index 2024 Structure
Pillars Dimensions Description
Government
Vision Existence of a national AI strategy.
Governance and
Ethics
Regulatory and ethical frameworks for AI use.
Digital Capacity
Digital skills and government's ability to adapt to
new technologies.
Technology
Sector
Innovation Capacity
Level of technological innovation and number of AI-
specialized startups.
Business
Environment
Ease of doing business and support for
technological entrepreneurship.
Human Capital
Availability of qualified AI professionals and
education level in STEM fields (science, technology,
engineering, and mathematics).
Data and
Infrastructure
Data Availability Access to quality data for training AI systems.
Data
Representativeness
The extent to which data accurately reflects the
diversity of the population.
Infrastructure
Quality and extent of digital infrastructure, including
Internet connectivity and computing power.
index measuring the level of effective and responsible integration of AI in public services
in 188 countries.

39 / 56
Government –
Governance & Ethics
Pillars Dimensions Description
Government Governance and Ethics
Regulatory and ethical frameworks for AI use,
that builds trust and legitimacy.
●
1. Data protection and privacy legislation.
– 100: Law in force ; 50: Draft law or pending approval ; 0: No evidence of law.
– Desk research (with support from the GovTech Maturity Index I-38 and IAPP Global Privacy Law and DPA
Directory)
●
2. Cybersecurity. Global Cybersecurity Index
●
3. Regulatory quality. Worldwide Governance Indicators (World Bank)
●
4. Ethical AI Principles.
– 100: Adoption of OECD AI Principles or independent frameworks aligned with
– OECD values, 50: Draft principles exist or clear evidence they are being developed in alignment with OECD ; 0:
No evidence of adoption.
●
5. Accountability. Worldwide Governance Indicators (World Bank)

40 / 56
Rank Country Score
1 United States 84.80
2 Singapore 81.97
3 United Kingdom 78.57
4 Finland 77.37
5 Canada 77.07
6 France 76.23
7 South Korea 75.85
8 Germany 75.42
9 Japan 74.89
10 Netherlands 74.56

Critical Perspectives on AI
Ethics Frameworks
(Ethics of Ethics in AI)
Prof. Karim Baïna
Full Professor

Ethics of Ethics in AI – critical
examination of ethical frameworks
●
Checkbox Principles : Guidelines Reduced to Checklist and AI Ethics cannot be reduced to checklist !!
●
Isolated principles : How AI can be Ethical if Education, and Industry are a-ethical missing out on broader or
more systemic issues ?
●
Meaningless, Toothless and Useless ethics principles : ethics is being used in place of regulation. Ethics is
being asked to do something it was never designed to do.
●
Gender Biased principles (How guidelines that are biased can detect biases ?) : Male-Way (address moral
problems primarily through a “calculating”, Cost-benefit decision, “rational”, “logic-oriented” ethics of justice).
Women do not, as men typically do, but rather, interpret them within a wider framework of an “empathic”,
“emotion-oriented” ethics of care, Value-based decision, Normative decision, Ethical decision, Moral decision,
Mission-driven decision, Identity-based decision, Principled decision, Integrity-driven decision.
●
AI Ethics are more Computational Ethics centered : Technical Solutions for technical problems
(Algorithmétique).
●
Lacks of mechanisms to reinforce/operationalize normative claims : Gap between abstract ethics values &
principles and technical discourses and implementations.
●
Limited Technical Solutions to operationalize normative ethics values & principles : Technical solutions of
enforcing Ethics principles adopt one particular understanding of ethics principles and& responds to it in one
particular way (e.g. insuring data privacy on the cloud = encryption so that the ability to identify subjects or de-
anonymize them is minimized). But such work adopts one particular understanding and responding to privacy.
●
Lacks of broader vision of Ethics (Machine Ethics, Metaethics, Normative Ethics, Applied Ethics) : Invite
to a reflective practice, reasoned moral reflection, & therefore deliberative which is based on reasons:
consideration of the wider contexts & comprehensive relationship networks in which AI systems are embedded
●
Ethics washing, Principle Washing : Companies can enjoy the appearance of ethics without the substance.
* "The Ethics of AI Ethics: An Evaluation of Guidelines", Thilo Hagendorff 2020
"The uselessness of AI ethics", Luke Munn, 2022
** "L’éthique au coeur de l’IA", Lyse Langlois, Marc-Antoine Dilhac, Jim Dratwa, Thierry Ménissier, Jean-Gabriel Ganascia, Daniel Weinstock,
Luc Bégin et Allison Marchildon, 2023

Current Landscape of AI &
AI Ethics and Data Privacy
in Morocco
Prof. Karim Baïna
Full Professor

44 / 56
Morocco AI Projects

45 / 56
EU AI Act
risk-based approach
Connecting the dots in trustworthy Artificial Intelligence: From AI principles, ethics, and key requirements to responsible AI
systems and regulation, Natalia Díaz-Rodríguez et al. Information Fusion’2023

46 / 56

47 / 56

48 / 56

49 / 56
Loi n° 09-08 relative à la protection des personnes physiques à l'égard du
traitement des données à caractère personnel
(Moroccan 09-08 law)
Cette loi vise à assurer une protection efficace des particuliers contre les abus
d’utilisation des données de nature à porter atteinte à leur vie privée et d’harmoniser
le système marocain de protection des données personnelles avec celles de ses
partenaires notamment européens. En outre, la loi institue une Commission
Nationale de protection des Données Personnelles (CNDP).
Loi n° 05-20 relative à la cybersécurité
(Moroccan 09-08 law)
La loi 05-20 relative à la cybersécurité vise notamment à mettre en place un cadre
juridique préconisant un ensemble de règles et de mesures de sécurité afin d’assurer
et renforcer la sécurité et la résilience des systèmes d’information des
administrations de l’Etat, des collectivités territoriales, des établissements et
entreprises publics et de toute autre personne morale de droit public de l’Etat ainsi
que des infrastructures d’importance vitale disposant des systèmes d’information
sensibles.

50 / 56

51 / 56
• Compared to its 88th position in 2023, Morocco has dropped in ranking,
101st globally, with a score of 41.78 out of 100.
• While Morocco has made progress in AI infrastructure, it lags behind in
governance and regulatory frameworks, which impacts its overall
readiness.

52 / 56

Ethics and Trust in AI:
Which Opportunities for
Researchers in the Human
and Social Sciences ?
Prof. Karim Baïna
Full Professor

Ethical Use of AI in SHS Research
« The future of ethical AI in Social Sciences
depends on whether scholars choose to act as
(i) Gatekeepers, reinforcing existing ethical
constraints, or as (ii) Revolutionaries,
pioneering new paradigms that redefine how AI
interacts with Society, Knowledge production,
and Policymaking. »
Remus Runcan et al. Ethical AI in Social
Sciences Research: Are We Gatekeepers
or Revolutionaries ?
https://www.mdpi.com/2075-4698/15/3/62/pdf?version=1741250479
"Ethical AI in Social Sciences Research: Are We Gatekeepers or Revolutionaries ?" Remus
Runcan et. Al, 2025

1. The AI-Enhanced HSS Researcher :
– Uses AI tools (e.g., NLP, machine learning, LLMs) to support SHS research (e.g., text
mining, discourse analysis).
2. The HSS Analyst of AI (Researcher on AI applied to SHS) :
– Studies how AI transforms society, language, labor, education, inequality, etc.
3. The Ethically-Informed Co-Designer :
– Participates within interdisciplinar (or transdisciplinar) team in co-designing, auditing, or
red-teaming of AI systems to ensure value alignment.
4. The Normative Guardian (HSS Gatekeeper Ethicist) :
– Applies and enforces existing ethical, legal, and institutional norms (e.g., GDPR, CNIL,
research integrity charters, AI Ethics Frameworks).
5. The Critical-Epistemic Revolutionary (HSS Revolutionary Ethicist) :
– Challenges the epistemic, political, or ontological foundations of AI systems; proposes
alternatives, by pioneering new paradigms that redefine how AI interacts with society,
knowledge production, and policymaking
6. The HSS Integrator of Ethical Frameworks :
– Bridges SHS insights with operational AI development (standardization, policy, hybrid
evaluation).
Ethical Use of AI in SHS Research
– Researcher Personas

Ethical Frameworks for
Trustworthy AI –
Opportunities for Researchers
in Human and Social Sciences
Prof. Karim Baïna
Full Professor

Ethical Frameworks for Trustworthy AI – Opportunities for Researchers in Human and Social Sciences

More Related Content

Similar to Ethical Frameworks for Trustworthy AI – Opportunities for Researchers in Human and Social Sciences (20)

More from Karim Baïna (20)

Recently uploaded (20)

Ethical Frameworks for Trustworthy AI – Opportunities for Researchers in Human and Social Sciences