Written Evidence Submitted by the Center for AI and Digital Policy (CAIDP)

(GAI0098)

The Center for AI and Digital Policy (CAIDP) welcomes the opportunity to provide comments on the Governance of Artificial Intelligence for the UK Parliament – Science and Technology Committee inquiry1.

The CAIDP is an independent, non-profit organization established to advise national governments and international organizations on AI and digital policy. In 2022 we published the CAIDP Artificial Intelligence and Democratic Values Index - 20212, a comprehensive report of the AI policies and practices in 50 countries, including the United Kingdom. As outlined in this report, we recommend that all countries:

Establish national policies for AI that implement democratic values
Ensure public participation in AI policymaking and create robust mechanisms for independent oversight of AI systems
Guarantee fairness, accountability, and transparency in all AI systems
Commit to these principles in the development, procurement, and implementation of AI systems for public services
Halt the use of facial recognition for mass surveillance
Curtail the deployment of lethal autonomous weapons
Begin implementation of the UNESCO AI Recommendation
Establish a comprehensive, legally binding convention for AI

Based on your call for evidence, our team of experts would like to add the following comments to the questions raised in the inquiry.

1 UK Parliament. Inquiry. Governance of Artificial Intelligence. Science and Technology Committee. November 5, 2022. https://committees.parliament.uk/work/6986/governance-of- artificial-intelligence-ai/

2 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

1. How effective is the current governance of AI in the UK?

1.1 What are the current strengths and weaknesses of current arrangements, including for research?

The United Kingdom is one of the 50 countries evaluated in our AIDV 2021 report to assess the state of governance of AI, their progress in AI policies, and implemented actions across 12 metrics3. In our 2021 AIDV review, we commended the outstanding efforts and plans of the UK to maintain a solid leadership in AI, through the enactment of the UK’s National AI Strategy4. The UK placed in Tier II out of the five tiers of our evaluation. We highlighted that a distinctive strength of the UK’s AI strategy was its commitment to ensuring that AI governance and regulatory systems foster innovation, growth, and investment while protecting the safety and fundamental rights of the public.

Noted in our report was the UK regulatory arrangement of cross-sectoral legislation5. Based on this approach, the UK at the onset opts for decentralizing the responsibility for AI oversight within the purview of different agencies (i.e. The Office for Artificial Intelligence, the Information Commissioner’s Office (ICO), the Medicines and Healthcare products Regulatory Agency, the AI Council, the UK AI Standards Hub and the Centre for Data Ethics and Innovation). Such arrangement poses a major challenge of “a lack of clarity, overlaps, inconsistency, and gaps in approach”6 as recognized by the own UK authorities. The decision of the Office of AI to develop a national position on governing and regulating AI is in line with the recommendations of OECD and UNESCO, and the proposed harmonized rules on artificial

3 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

4 GOV.UK. National AI Strategy (Sept 2021).

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file

/1020402/National_AI_Strategy_-_PDF_version.pdf

5 GOV.UK. National AI Strategy (Sept 2021). (p.54) https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file

/1020402/National_AI_Strategy_-_PDF_version.pdf

6 GOV UK. AI Regulation. Policy Paper. July 2022) https://www.gov.uk/government/publications/establishing-a-pro-innovation-approach-to- regulating-ai/establishing-a-pro-innovation-approach-to-regulating-ai-policy-statement

intelligence discussed in 2021 at the European Commission7 . CAIDP encourages the UK to continue the pursuit of a unified, standardized, and national position for AI oversight in line with the call from the Council of Europe for a “cross-cutting regulatory framework of AI”8.

CAIDP strongly suggests that the UK regulatory system address harmonization of rules and standards and the national, regional, and global level, to address the cross-border nature of AI and data systems9, in line with the resolutions of the Council of Europe. Such harmonization will ensure achieving a ‘common ground around security, safety, and system resilience”10.

The UK proposal for a new pro-innovation and risk-based approach to AI Regulation submitted to the UK Parliament in July 202211 advocates for a regulatory framework that is “proportionate, light-touch and forward-looking”. The pillar of ‘Governing AI effectively”, the third in the AI action plan aims at promoting clarity, and confidence in businesses and increasing public trust. The establishment of the Office of Artificial Intelligence, as the oversight body for AI; the Centre for Data Ethics and Innovation (CDEI) tasks with regulating the trustworthy use of data, and AI12 and the participation in ICO’s formulation of an Auditing Framework for AI13 are positive strategic actions.

7 European Commission Proposal for a regulation laying down harmonised rules on artificial intelligence. April 2021). https://digital-strategy.ec.europa.eu/en/library/proposal-regulation- laying-down-harmonised-rules-artificial-intelligence

8 CAIDP. Council of Europe AI Treaty. https://www.caidp.org/resources/coe-ai- treaty/#:~:text=The%20Council%20of%20Europe%20and%20AI&text=All%20Council%20of% 20Europe%20member,to%20shape%20modern%20privacy%20law.

9 CAIDP. CAHAI Multi-stakeholder consultation. (May, 2021) https://www.caidp.org/app/download/8317777163/CAIDP-CAHAI-06052021.pdf 10 CAIDP. Statement to UK Regarding AI Accountability. (Nov. 2021).

https://www.caidp.org/app/download/8356505263/CAIDP-Statement-UK-17112021.pdf

11 GOV.UK. AI Regulation. Policy Paper. Presented to Parliament by the Secretary of State for Digital Culture, Media and Sport. (July 2022). https://www.gov.uk/government/publications/establishing-a-pro-innovation-approach-to- regulating-ai/establishing-a-pro-innovation-approach-to-regulating-ai-policy-statement

12 GOV.UK. National AI Strategy (Sept 2021). (p.31) https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file

/1020402/National_AI_Strategy_-_PDF_version.pdf

13 European Commission. ICO releases AI and data protection framework. (May, 2022). https://eudatasharing.eu/news/ico-releases-ai-and-data-protection-framework

Notwithstanding this progress, we note that the strategic intent falls short of a more explicit plan to lead the promotion of a human-centric development of AI, where the protection of fundamental rights, democratic values, and rule of law is a priority. CAIDP urges the UK Parliament to consider that in devising robust AI governance systems, the UK adopts the UNESCO Recommendation for the Ethics of AI. UNESCO calls for the implementation of AI governance mechanisms for “effective protection, monitoring of impact, enforcement and redress” 14, that extend from soft governance (i.e. certifications of AI systems) to enforcement of “auditability and traceability of AI systems”.

We acknowledged that the UK has a strong record in the protection of Human Rights15 and that the formulation of the 2022-2025 strategic plan of the Equality and Human Rights Commission aims to address the impact of artificial intelligence on human rights.16 In the formulation of concrete plans to promote innovation, we advise the UK to integrate explicitly the UNESCO Recommendation for Ethical AI, to ensure that AI innovation is “anchor [ed] in human rights and fundamental freedoms, values and principles, and moral and ethical reflection”17.

In the R&D field, the agenda of the UK’s pro-innovation strategy places AI technology at the center of its short, medium, and long-term plans18. The UK has set a goal to launch a National AI Research and Innovation Program (R&I) with five aims to unlock the potential of AI technologies through discovery, maximization of creativity, development of capacity, international partnerships, and support for adoption. We note that these plans do not include an explicit recognition of the centrality of human rights and trustworthy AI in innovation. CAIDP

14 UNESCO. Recommendation on the Ethics of Artificial Intelligence. (Nov 2021) (p.11) https://unesdoc.unesco.org/ark:/48223/pf0000380455

15 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021. (Feb 2021) (p.465)

https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988

16 Equality and Human Rights Commiission. Strategic Plan 2022-2025. (Mar 2022) https://www.equalityhumanrights.com/sites/default/files/about-us-strategic-plan-2022-2025.pdf 17 UNESCO. Recommendation on the Ethics of Artificial Intelligence. (Nov 2021) (p.1) https://unesdoc.unesco.org/ark:/48223/pf0000380455

18 GOV.UK. National AI Strategy (Sept 2021). (p.8)

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file

/1020402/National_AI_Strategy_-_PDF_version.pdf

advocates for the criticality of integrating fundamental rights compliance, as “a necessary precondition for market participation”19 We believe that innovation without a human-centric approach is not true innovation. We, therefore urge the UK to adopt appropriate regulatory and policy measures that while fostering “innovation and competition for trustworthy AI20,” supports the transition from R&D to deployment, prioritizing the protection of human rights and fairness. We encourage the UK to lead the way in ensuring research on the ethical application of AI systems21, standardization of “internationally comparable metrics to measure AI research, development, and deployment” and reporting on progress in the implementation of these global principles.

In our forthcoming AIDV 2022 report, we incorporated a new metric22 to evaluate how countries are implementing the UNESCO Recommendations for on AI Ethics. We recommend adopting and referencing this framework in the current governance arrangements and reporting on the commitments and progress, to lead by example to other nations in the world.

2. What measures could make the use of AI more transparent and explainable to the public?

CAIDP commends the Science and Technology Committee for looking for ways to make the use of AI more transparent and explainable to the public.

Algorithmic transparency is one of the key metrics of the CAIDP’s evaluation of AI policies and practices23. CAIDP emphasizes the importance of algorithmic transparency as the

19 CAIDP. Statement to the European Commission, the European Parliament, and the European Council, on Draft EU AI regulation. https://www.caidp.org/app/download/8312964663/CAIDP- EU-AI-20042021.pdf?t=1619014103

20 OECD. OECD AI Principles (May 2019) (para 20.)

https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449

21 UNESCO. Recommendation on the Ethics of Artificial Intelligence. (Nov 2021) https://unesdoc.unesco.org/ark:/48223/pf0000380455

22 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988 23 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, iii, p.4. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988

cornerstone of trustworthy and human-centric AI24 and recommends that all countries “must guarantee fairness, accountability, and transparency in all AI systems”25.

Algorithmic Transparency is well recognized in the AI policy of the United Kingdom, as can be seen in the Information Commissioner’s Office’s (ICO) guidance26 which includes the concept of algorithmic transparency as derived from the UK’s Data Protection framework27.

We particularly applaud the recent efforts of the Central Digital and Data Office (CDDO) and the Centre for Data Ethics (CDEI) in piloting the UK Algorithmic Transparency Standard to improve transparency regarding algorithmic tools in the public sector28.

We highly recommend pursuing the plans of the CDDO and the CDEI to extend the standard to more public offices, including seeking the endorsement of the Data Authority of the standard as an official cross-government standard, as announced in their joint announcement29. Such an extension would be beneficial for increasing the public’s understanding of algorithmic systems used by the government.

In addition, we recommend that the government explicitly implements the UNESCO Recommendations on the Ethics of Artificial Intelligence30 concerning algorithmic transparency

24 CAIDP Statement to UK Regarding AI Accountability. Nov 17, 2021, p.2 https://www.caidp.org/app/download/8356505263/CAIDP-Statement-UK- 17112021.pdf?t=1660245988

25 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.4. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988 26UK ICO, Guidance on AI and Data Protection (July 30, 2020), https://ico.org.uk/for-

organisations/guide-to-data-protection/key-dp-themes/guidance-on-ai-and-data-protection/what- do-we-need-to-do-to-ensure-lawfulness-fairness-and-transparency-in-ai-systems/

27 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.464. (Feb 2021), https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988

28 UK CEIDP, (July 1, 2022), Piloting the UK Algorithmic Transparency Standard, Centre for Data Ethics and Innovation Blog, https://cdei.blog.gov.uk/2022/06/01/piloting-the-national- algorithmic-transparency-standard/

29 UK CEIDP, (July 1, 2022), Piloting the UK Algorithmic Transparency Standard, Centre for Data Ethics and Innovation Blog, https://cdei.blog.gov.uk/2022/06/01/piloting-the-national- algorithmic-transparency-standard/

30 UNESCO Recommendations on the Ethics of AI (2021), https://unesdoc.unesco.org/ark:/48223/pf0000380455

for decision-making algorithms in its guidance on AI, such as the ICO guidance on AI and Data Protection31.

The UNESCO supports algorithmic transparency regarding decision-making AI systems and recommends:

People should be fully informed when a decision is informed by or is made based on AI algorithms, including when it affects their safety or human rights, and in those circumstances should have the opportunity to request explanatory information from the relevant AI actor or public sector institutions. In addition, individuals should be able to access the reasons for a decision affecting their rights and freedoms and have the option of making submissions to a designated staff member of the private sector company or public sector institution able to review and correct the decision32.

CAIDP believes that disclosing the decision-making process of AI systems to individuals is crucial to discover bias or unfair decisions, and allows affected individuals to take appropriate actions33.

Regarding the explainability of AI, CAIDP equally recommends explicitly implementing the UNESCO recommendations. Explainability and transparency are closely related. UNESCO describes explainability in this sense:

Explainability refers to making intelligible and providing insight into the outcome of AI systems. The explainability of AI systems also refers to the understandability of the input, output, and functioning of each algorithmic building block and how it contributes to the outcome of the systems. Thus, explainability is closely related to transparency, as

31 UK ICO, Guidance on AI and Data Protection (July 30, 2020), https://ico.org.uk/for- organisations/guide-to-data-protection/key-dp-themes/guidance-on-ai-and-data-protection/what- do-we-need-to-do-to-ensure-lawfulness-fairness-and-transparency-in-ai-systems/

32 UNESCO Recommendations on the Ethics of AI (2021)(Recommendation #38), https://unesdoc.unesco.org/ark:/48223/pf0000380455

33 CAIDP Statement to EDPB on EDPB Guidelines 01/2022 on Data Subject Rights - Right of Access, 11 March 2022, p.6, https://edpb.europa.eu/sites/default/files/webform/public_consultation_reply/CAIDP-Statement- EDPB-11032022.pdf

outcomes and sub-processes leading to outcomes should aim to be understandable and traceable, appropriate to the context.34

UNESCO therefore recommends:

AI actors [controllers and processors] should commit to ensuring that the algorithms developed are explainable. In the case of AI applications that impact the end user in a way that is not temporary, easily reversible, or otherwise low risk, it should be ensured that a meaningful explanation is provided with any decision that resulted in the action taken for the outcome to be considered transparent.35

We strongly support the UNESCO recommendations on AI concerning transparency and explainability and believe that the Science and Technology Committee should take the opportunity of the inquiry regarding the Governance of Artificial Intelligence to incorporate the UNESCO text, which has already been endorsed by the UK. We note also that Gabriala Ramos, the guiding force behind the UNESCO Recommendation, gave a keynote speech on UK AI at the Alan Turing Institute earlier this year.36 In her remarks, Ramos said:

We cannot ignore the real negative impact that unregulated technologies are having on our democracies, on our children and with the reproduction of biases and discrimination, and hate speech.

We also consider the fact that more and more decisions are taken with the support of Artificial intelligence, in the public and private sectors. To have a loan, to have a job interview, etc, and we should ensure they do not open the door to more inequalities, that start with the fact that half of the world is not connected to the internet, and more than that may not be represented in the data.

34 UNESCO Recommendations on the Ethics of AI (2021)(Recommendation #40), https://unesdoc.unesco.org/ark:/48223/pf0000380455

35 UNESCO Recommendations on the Ethics of AI (2021)(Recommendation #40),

https://unesdoc.unesco.org/ark:/48223/pf0000380455

36 The Alan Turing Institute, UNESCO to give key note speech at AI UK 2022: The Alan Turing Institute welcomes UNESCO’s Gabriela Ramos to AI UK 2022, https://www.turing.ac.uk/news/unesco-give-key-note-speech-ai-uk-2022

At UNESCO, we make the call to foster innovation, but foster it within an ethical framework. The 193 countries that signed up to our framework, the Recommendation on the Ethics of Artificial Intelligence, which includes the UK, ascribe to this vision. 37

3. How should decisions involving AI be reviewed and scrutinized in both the public and private sectors? Are current options for challenging the use of AI adequate and, if not, how can they be improved?

The UK Data Protection Act currently includes an individual’s right to the intervention regarding automated decisions in Art. 22 UK GDPR38. Recently, however, the Taskforce on Innovation, Growth, and Regulatory Reform proposed removing this provision39. After the majority of the respondents to a public consultation noted that the human review of automated decisions is an important safeguard, the government abandoned its plans to remove Article 22 UK GDPR in June 2022 and instead now seeks to amend Article 22 ”to cast it “as a right to specific safeguards, rather than as a general prohibition on solely automated decision-making” to allow for more innovation40.

CAIDP applauds the decision to retain Article 22, but strongly cautions against limiting the scope of Article 22. Article 22 is the cornerstone for oversight of AI techniques41, and follows the Data Protection Working Party’s assessment of Article 22 as a fundamental

37 Gabriela Ramos, The Alan Turing Institute – AI UK 2022, March 29, 2022, https://gabrielailianramos.wordpress.com/2022/03/29/the-alan-turing-institute-ai-uk-2022/

38 Art. 22 UK GDPR, https://www.legislation.gov.uk/eur/2016/679/article/22; CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.464. (Feb 2021)

39 Government of the UK, Open consultation by the Government of the UK, Data: A new direction, 10 September 2021, https://www.gov.uk/government/consultations/data-a-new- direction.

40 Government of the UK, Data: A new direction: government response to consultation, 23 June 2022,

41 CAIDP Statement Importance of AI Accountability in response to the DCMS consultation

Data: a new Direction, https://www.gov.uk/government/consultations/data-a-new- direction/outcome/data-a-new-direction-government-response-to-consultation

safeguard towards protecting fundamental rights and the rule of law42. Limiting its application to promote innovation could pose unwanted risks to individuals subjected to automated decisions.

4. How should the use of AI be regulated, and which body or bodies should provide regulatory oversight?

In September 2021, the UK government launched its first National Artificial Intelligence Strategy43, in which one of the three key pillars is ensuring that the UK gets the national and international governance of AI technologies right to encourage innovation, and investment and protect the public and the country’s fundamental values. At CAIDP, we believe that true innovation promotes human-centric approaches with an emphasis on fairness, transparency, accountability, adherence to the rule of law, and protecting the fundamental rights of the public.

We applaud and support the plans by the UK government to develop a national position through the Office for Artificial Intelligence on governing and regulating AI. We call upon the UK government to ensure the Office for Artificial Intelligence has independent oversight on governing and regulating AI, working closely with the existing sector-specific bodies with statutory regulatory powers to enhance coordination and harmony.

5. To what extent is the legal framework for the use of AI, especially in making decisions, fit for purpose?

The fast-paced advancements of AI technologies make it the most critical to develop legislation that is fit for purpose of protecting the rights and freedoms of people. The only way to reach the objective of a fit-for-purpose legal framework for the use of AI systems is through a human-centric approach, that prioritizes transparency, and fairness, and promotes adherence to

42 European Commission. Article 29 Data Protection Working Party, Guidelines on Automated individual decision-making and Profiling for the purposes of Regulation 2016/679, 3 October 2017. https://ec.europa.eu/newsroom/article29/items/612053/en .

43 Government of the UK, National AI Strategy (Sept. 2021), https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file

/1020402/National_AI_Strategy_-_PDF_version.pdf

the rule of law, and the protection of democratic values44. CAIDP strongly advocates for these to be non-negotiable tenets of the development of the UK’s AI legal framework.

The Council of Europe called attention to the need to revisit national legislation to address the “opacity, complexity, bias, (..) unpredictability and partially autonomous behavior of certain AI systems45”. CAIDP concurs with the view of the Council about automated decision- making AI systems that may result violate human rights in access to employment, credit scoring, social scoring, or judicial decisions46. We highly recommend that in devising or revising legal frameworks for AI systems the UK integrates algorithm transparency, explainability, human oversight, accountability, safety, and a red line to mass surveillance systems of facial recognition and social scoring47. CAIDP encourages the UK to maintain its multi-stakeholder consultation approach to integrate the views of all actors in the formulation or revision of an AI legal framework that is fit for purpose of protection of human rights.

6. Is more legislation or better guidance required?

CAIDP commends the UK for putting in place various legislations for regulating the design, development, procurement, and use of AI48. We believe that uses of AI that have the risk and potential to affect the fundamental rights of the public should be regulated through legislation to ensure more accountability, and transparency, and uphold the rule of law.

44 CAIDP Statement Importance of AI Accountability in response to the DCMS consultation Data: a new Direction, https://www.gov.uk/government/consultations/data-a-new- direction/outcome/data-a-new-direction-government-response-to-consultation

45 EC. Proposal for a Regulation laying down harmonized rules on artificial intelligence. (April, 2021). https://ec.europa.eu/newsroom/dae/redirection/document/75788

46 CAIDP. Statement to the European Commission, the European Parliament, and the European Council, on Draft EU AI regulation. https://www.caidp.org/app/download/8312964663/CAIDP- EU-AI-20042021.pdf?t=1619014103

47 CAIDP. Statement to CAHAI Multi-stakeholder consultation. (Nov 2021). https://www.caidp.org/app/download/8357849663/CAIDP-Statement-CAHAI- 23112021.pdf?t=1660245988

48 GOV.UK. AI Regulation. Policy Paper. Presented to Parliament by the Secretary of State for Digital Culture, Media and Sport. (July 2022). https://www.gov.uk/government/publications/establishing-a-pro-innovation-approach-to- regulating-ai/establishing-a-pro-innovation-approach-to-regulating-ai-policy-statement

We call your attention that The Court of Appeal established that there are “fundamental deficiencies” in the existing legal framework for regulating AI in areas such as facial recognition technology49.

CAIDP recommends that The Office for Artificial Intelligence working with other sector- specific agencies that have regulatory authority must determine any gaps where current legislation may not be sufficient, and any subsequent regulation which may be needed in addressing AI issues like fairness, transparency, and explainability.

7. What lessons, if any, can the UK learn from other countries on AI governance?

We encourage the UK to examine the leading initiatives of other countries regarding AI governance, encompassing ethical AI, independent AI oversight, algorithmic transparency, and trustworthy AI innovation. Countries that are exemplars of AI governance include Korea, Germany, France, and Italy, positioned in Tier I of the CAIDP AIDV2021 report.

Korea stands out for its National Strategy for Artificial Intelligence which includes an explicit declaration to “prevent AI dysfunction and establish AI ethics” gathered from public consultation with academia, industry, and civil society.50 Korea amended the Framework Act on National Informatization51 to regulate the development and use of AI.

Italy’s National Strategy for Artificial Intelligence is an example of the harmonization of national policies and practices, with multilateral entities, such as the OECD, the Global Partnership on AI (GPAI), and the European Commission52. The Agency for Digital Italy (AgID) issued the RenAIssance strategy, to promote a human-centric approach to AI within the national

49 R (Bridges) v Chief Constable of South Wales Police & Ors [2020] EWCA Civ 1058, 11 August 2020, para 91

50 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p 282. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988 51 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p 284. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

52 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p 242. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

strategy of the country, promoting: “AI for human beings, AI for a reliable, productive and sustainable digital ecosystem and AI for sustainable development” 53

Leaders in independent AI policy oversight are Germany and France54. Germany has implemented the Plattform Lernende Systeme which presents a regional mapping of the AI developments in the country. France fosters a multi-agency landscape for AI oversight, including the Commission Nationale Informatique & Libertes (CNIL), which supervises data protection and data privacy; and the Defender of Rights (Défenseur des droits) which focuses on preventing risks of bias and harm due to automated discrimination by AI systems.

Algorithmic transparency is one of the most important policy issues today and takes the position that individuals should have the right to access the logic, the factors, and the data that contributed to a decision concerning them55. Current examples we commend are the new European Digital Services Act56, providing transparency obligations for large online platforms and search engines, and the proposed US Algorithmic Justice and Online Platform Transparency Act57, which among others requires transparency in the use of algorithmic processes and content moderation. Such legislative movements are steps in the right direction and can serve as raw models. Legislation like the Digital Services Act will e.g. make ad targeting more transparent and ensure that large companies will be held accountable, according to European Commissioner

53 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p 243. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

54 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p 540. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

55 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.503. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

56 EU Digital Services Act, Proposal for a REGULATION OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on a Single Market For Digital Services (Digital Services Act) and amending Directive 2000/31/EC, (2020). https://eur-lex.europa.eu/legal- content/en/ALL/?uri=COM:2020:825:FIN;

https://ec.europa.eu/info/strategy/priorities-2019-2024/europe-fit-digital-age/digital-services-act- ensuring-safe-and-accountable-online-environment_en#new-obligations

57U.S. Proposed Algorithmic Justice and Online Platform Transparency Act, Text - H.R.3611 - 117th Congress (2021-2022): Algorithmic Justice and Online Platform Transparency Act, H.R.3611, 117th Cong. (2021), https://www.congress.gov/bill/117th-congress/house- bill/3611/text.

for Competition Margrethe Vestager58. The proposed US Algorithmic Justice and Online Platform Transparency Act equally aim to provide transparency to users of online platforms.

In addition to legislation addressing AI issues like transparency, and explainability, CAIDP highly recommends the UK prioritize investment in R&D for trustworthy AI. Countries in the Top Tier of CAIDP’s AIDV Index, such as Germany, made explainability and transparency one of its funding priorities for AI R&D projects59. The German Ministry of Education and Research stated that improving explainability and transparency are two of the Federal government’s central research goals60. Korea, another Tier I Country in CAIDP’s AIDV Index also emphasizes the importance of explainability in its AI R&D activities61. A final Top Tier country is Italy, with a thriving AI research ecosystem, coordinated strategies and actions, and government-funded schemes for new businesses and SMEs in the digital sector. Italy’s outstanding AI Strategic Programme identified 11 priority areas and six objectives for AI, with specific mention of the aim to “develop and adopt human-centered and trustworthy AI” and “develop AI-driven policies and services in the public sector”.

We commend such strategic actions, which signal that is possible to achieve progress and a leading position in AI innovation with consistent and determined progress in the promotion of human-centric innovation, fundamental human rights, rule of law, and democratic values.

Thank you for the consideration of our views. We would welcome the opportunity to discuss these recommendations with you.

Sincerely yours,

58 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.18. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

59 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.190. (Feb 2021)

https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

60 Bundesministerium für Bildung und Forschung, KI-Erklärbarkeit und Transparenz, https://www.softwaresysteme.pt-dlr.de/de/ki-erkl-rbarkeit-und-transparenz.php

61 CAIDP. Artificial Intelligence and Democratic Values AIDV 2021, p.278. (Feb 2021)

https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf

Marc Rotenberg	Merve Hickok	Karine Caunes, PhD
CAIDP President	CAIDP Research Director	CAIDP Program Coordinator

Tina Lassiter	Grace S. Thomson	Eric Wamugu
CAIDP Team Leader	CAIDP Team Leader	CAIDP Team Leader

Attachment

Artificial Intelligence and Democratic Values 492-500 (CAIDP 2022) (UK Country Report) Pg 458-465.

(25 November 2022)

References

Art. 22 UK GDPR, https://www.legislation.gov.uk/eur/2016/679/article/22; Bundesministerium für Bildung und Forschung, KI-Erklärbarkeit und Transparenz,

https://www.softwaresysteme.pt-dlr.de/de/ki-erkl-rbarkeit-und-transparenz.php

CAIDP Statement about Importance of AI Accountability in response to the DCMS consultation Data: a new Direction, https://www.gov.uk/government/consultations/data-a-new- direction/outcome/data-a-new-direction-government-response-to-consultation

CAIDP Statement to EDPB on EDPB Guidelines 01/2022 on Data Subject Rights - Right of Access, 11 March 2022, p.6, https://edpb.europa.eu/sites/default/files/webform/public_consultation_reply/CAIDP- Statement-EDPB-11032022.pdf

CAIDP. Artificial Intelligence and Democratic Values AIDV 2021. (Feb 2021) https://www.caidp.org/app/download/8376927963/AIDV-Index-2021.pdf?t=1660245988

CAIDP. Council of Europe AI Treaty. https://www.caidp.org/resources/coe-ai- treaty/#:~:text=The%20Council%20of%20Europe%20and%20AI&text=All%20Council%20 of%20Europe%20member,to%20shape%20modern%20privacy%20law.

CAIDP. Statement to CAHAI Multi-stakeholder consultation. (Nov 2021). https://www.caidp.org/app/download/8357849663/CAIDP-Statement-CAHAI- 23112021.pdf?t=1660245988

CAIDP. Statement to the European Commission, the European Parliament, and the European Council, on Draft EU AI regulation. https://www.caidp.org/app/download/8312964663/CAIDP-EU-AI- 20042021.pdf?t=1619014103

CAIDP. Statement to UK Regarding AI Accountability. (Nov. 2021).

EC. Proposal for a Regulation laying down harmonized rules on artificial intelligence. (April, 2021). https://ec.europa.eu/newsroom/dae/redirection/document/75788

Equality and Human Rights Commission. Strategic Plan 2022-2025. (Mar 2022) https://www.equalityhumanrights.com/sites/default/files/about-us-strategic-plan-2022- 2025.pdf

EU Digital Services Act, Proposal for a REGULATION OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL on a Single Market For Digital Services (Digital Services Act) and amending Directive 2000/31/EC, (2020). https://eur-lex.europa.eu/legal- content/en/ALL/?uri=COM:2020:825:FIN;

European Commission Proposal for a regulation laying down harmonised rules on artificial intelligence. April 2021). https://digital-strategy.ec.europa.eu/en/library/proposal-regulation- laying-down-harmonised-rules-artificial-intelligence

European Commission. Art. 29 Data Protection Working Party, Guidelines on Automated individual decision-making and Profiling for the purposes of Regulation 2016/679, 3 October 2017. https://ec.europa.eu/newsroom/article29/items/612053/en

European Commission. ICO releases AI and data protection framework. (May, 2022). https://eudatasharing.eu/news/ico-releases-ai-and-data-protection-framework

European Commission. The Digital Services Act: ensuring a safe and accountable online enviornment. https://ec.europa.eu/info/strategy/priorities-2019-2024/europe-fit-digital- age/digital-services-act-ensuring-safe-and-accountable-online-environment_en#new- obligations

GOV. UK, Open consultation by the Government of the UK, 10 September 2021, https://www.gov.uk/government/consultations/data-a-new-direction.

GOV.UK. AI Regulation. Policy Paper. Presented to Parliament by the Secretary of State for Digital Culture, Media and Sport. (July 2022). https://www.gov.uk/government/publications/establishing-a-pro-innovation-approach-to- regulating-ai/establishing-a-pro-innovation-approach-to-regulating-ai-policy-statement

GOV.UK. National AI Strategy (Sept 2021). https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data

/file/1020402/National_AI_Strategy_-_PDF_

Government of the UK, Data: A new direction: government response to consultation, 23 June 2022,

https://www.caidp.org/app/download/8356505263/CAIDP-Statement-UK- 17112021.pdf?t=1660245988

OECD. OECD AI Principles (May 2019)

https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0449

R (Bridges) v Chief Constable of South Wales Police & Ors [2020] EWCA Civ 1058, (Aug 2020), (para 91) https://www.judiciary.uk/wp-content/uploads/2020/08/R-Bridges-v-CC- South-Wales-ors-Judgment.pdf

U.S. Proposed Algorithmic Justice and Online Platform Transparency Act, Text - H.R.3611 - 117th Congress (2021-2022): Algorithmic Justice and Online Platform Transparency Act, H.R.3611, 117th Cong. (2021), https://www.congress.gov/bill/117th-congress/house- bill/3611/text.

UK CEIDP, (July 1, 2022), Piloting the UK Algorithmic Transparency StandardCentre for Data Ethics and Innovation Blog, https://cdei.blog.gov.uk/2022/06/01/piloting-the-national- algorithmic-transparency-standard/

UK ICO, Guidance on AI and Data Protection (July 30, 2020), https://ico.org.uk/for- organisations/guide-to-data-protection/key-dp-themes/guidance-on-ai-and-data- protection/what-do-we-need-to-do-to-ensure-lawfulness-fairness-and-transparency-in-ai- systems/

UK Parliament. Inquiry. Governance of Artificial Intelligence. Science and Technology Committee. November 5, 2022. https://committees.parliament.uk/work/6986/governance-of- artificial-intelligence-ai/

UK Parliament. Call for evidence. Governance of Artificial Intelligence. Science and Technology Committee. November 5, 2022. https://committees.parliament.uk/call-for- evidence/2750/

UNESCO. Recommendation on the Ethics of Artificial Intelligence. (Nov 2021) https://unesdoc.unesco.org/ark:/48223/pf0000380455