1. Introduction: Aim of This Analysis

Public administrations are increasingly using AI to develop public policies in cities^[1], even going so far as to adopt formal legal decisions in a fully automated way, as happened in late 2023 in the world-famous case of the Brazilian city of Porto Alegre, which passed a water meters ordinance entirely written with generative AI^[2].

The purpose of this study is to analyse the use of AI in smart cities, considering its possibilities and its legal limits.

The OECD defines smart cities as «initiatives or approaches that effectively leverage digitalisation to boost citizen well-being and deliver more efficient, sustainable and inclusive urban services and environments as part of a collaborative, multi-stakeholder process»^[3]. According to an EU study, «Smart cities are therefore at the interface between social and technological dimensions»^[4]. The technological dimension of the smart city relates to the role of AI: «While urban and local authorities can resort to AI to improve the efficiency of a given decision-making process or enhance the delivery of a public service, AI makes the most of its potential in a smart-city environment. The combination of AI and other digital technologies (big data, Internet of Things, cloud and telecommunication infrastructure) enables the development of smart cities»^[5].

Machine learning AI can be of great use to process big data and automate city management and make it more efficient in various areas such as public service provision (e.g. emergency services), environmental control, traffic, etc. In this study, I aim to concentrate on what could be called smart urban planning and management, and smart housing and homes.

In relation to local formalised legal decisions, legal rules such as urban planning or ordinances, and administrative acts such as traffic sanctions or building permits, AI can also play a relevant role, although, as will be briefly highlighted later, the law imposes certain legal limits on its use due to its potential risks.

In this sense, to conclude this introduction, the idea that we would like to highlight is the following. While AI has a number of important advantages for the promotion of the right to good administration of people living in cities (art. 41 of the European Charter of Fundamental Rights, ECFR^[6], and national legislation of member states, e.g. art. 97 of the Italian Constitution or art. 30 of the Statute of Autonomy of Catalonia), it also presents a number of environmental costs and risks to people’s rights.

The advantages of AI have been highlighted by scholarship and jurisprudence^[7]. Among those who write on the subject, it has been pointed out how AI is capable of processing huge amounts of data, making predictions using correlations, avoiding cognitive biases and what is known as “noise”^[8], or generating new employment opportunities.

For example, in Italy, the decision of the Italian Council of State of December 13, 2019 (number 8472), underlines that (English translation is mine):

«7.1 In general terms, it should be stressed that the public administration must also be able to exploit the considerable potential of the digital revolution. In this context, the use of computer algorithms for decision-making in the public and private spheres is based on the claimed gains in terms of efficiency and neutrality. In many fields, algorithms promise to become the tool through which the distortions and imperfections that typically characterise the cognitive processes and choices made by human beings, highlighted, especially in recent years, by an impressive range of literature on behavioural economics and cognitive psychology, can be corrected. In this context, the decisions made by the algorithm thus assume an aura of neutrality, the result of aseptic rational calculations based on data».

Likewise, the decision of the same body of 4 February 2020 (number 881), stresses that (translation is mine):

«As already highlighted in the previous section, the usefulness of this operational method of public interest management is particularly evident about procedures, such as the one at issue in this litigation, which are serialised or standardised, involve the processing of large numbers of applications, and are characterised by the acquisition of certain and objectively verifiable data and the absence of any discretionary appreciation. The full admissibility of these instruments is in line with the canons of efficiency and cost-effectiveness of administrative action (Article 1 of Law No. 241 of 1990), which, in accordance with the constitutional principle of good administration (buon andamento) (Article 97 of the Constitution), require the administration to achieve its objectives with the least expenditure of means and resources and by streamlining and accelerating the procedure». However, along with the undoubted advantages of AI, scholarship and case law have also highlighted the risks that can arise with the use of this technology. From bugs that can give rise to serious errors on an enormous scale, to biases, with the consequent danger of discrimination, adding the opacity generated by black boxes or the hallucinations that these systems can suffer, for example, which, on the other hand, will generate a replacement of humans in the workplace in both the public and private sectors.

Several judgments in different countries have identified a variety of problems related to the use of AI that have led to serious damage and public administrations being held responsible.

In the Australian case called “Robodebt”, certain administrative tasks in the social field performed by humans were automated. The result of automation errors has been 400,000 welfare recipients in Australia who were wrongly accused of misreporting their income to the welfare agency and were consequently fined. From July 2015 to November 2019, the system used algorithms to calculate overpayments, collecting more than $1.2 billion, which the government will be forced to repay, because of a 2019 court ruling that the scheme was unlawful^[9].

In France, the judgment of December 7, 2022 of the Paris Administrative Court annulled a decision to assign a student to a secondary school which was only the sixth option entered by the student in the “Affelnet” application, on the grounds that the methods of calculation of the student’s points scale were incorrect and that these “material errors” influenced the overall line of the decision issued by the preparatory commission for the process and, consequently, the assignment decision taken by the academic director of the National Education Services^[10].

In a British case, between 1999 and 2015, the British Post Office accused nearly 4,000 workers of theft, fraud and false accounting, of whom nearly 900 were convicted of crimes they did not commit, and which have been attributed to an error in the company’s computer system. The case generated great suffering, including several suicides by those affected and even prison sentences, and has given rise to a television series, “Mr. Bates vs the Post Office” on the ITV channel.

In Spain, the ruling of December 30, 2021 of the Central Contentious-Administrative Court No. 8, which is the first judicial decision, except for errors or omissions, in relation to an algorithmic system in Spain, and the ruling of the National High Court of April 30, 2024, appeal no. 51/2022, which resolves the appeal against the former, have denied the access to the source code of the Bosco program used by the Spanish Administration to solve applications to obtain the energy bonus in case of energy poverty. The litigation has arrived in 2024 to the Spanish Supreme Court which will decide in the future about this conflict definitively.

Therefore, the use of AI in smart cities by the authorities will always be double-edged: the improvements in public management (effectiveness of the right to good administration of people living in cities) vs. the risks of violating rights such as equality, transparency or data protection and privacy, for example. Hence the need for regulation of the use of AI, such as that developed by the Regulation (EU) 2024/1689 of the European Parliament and of the Council of June 13, 2024 laying down harmonized rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828  (also known as Artificial Intelligence Act, RUEAI) and those that may be approved in the future by national public authorities to protect said rights.

2. Artificial Intelligence and Urban Planning

There is no doubt that one of the areas of competence of national local authorities in Europe is that of urban development and housing^[11]. Decisions on land use and the location of housing and public and private infrastructures are of crucial importance in cities.

According to the above-mentioned EU study of 2021, there is evidence that smart cities that use AI are growing fast, in particular, after the COVID-19 crisis, although aggregate quantitative evidence is difficult to collect, in part due to the multiplicity of the forms smart cities can take, alongside their very varied use of AI. However, qualitative evidence from specific examples shows that the use of AI by local urban authorities is still mostly restricted to simple tasks. In general, incremental approaches are adopted to gradually develop initiatives and capacities around AI. There are interesting examples of larger and/or high-income cities, which have been early adopters of the smart-city concept, and which exhibit significant progress in the diffusion and exploitation of AI.

A field where several cities are working intensively is that of digital twins: digital reproductions of the urban fabric of cities where cities can experiment with different public policies (e.g. traffic and transport policy)^[12].

In general, the use of AI in the field of urban planning can be developed in several areas, for example:

(a) Analysis of big data: which makes it possible to know the urban situation of the city and proceed to develop actions against problems such as urban segregation caused by gentrification and exclusionary zoning.

b) The promotion of citizen participation in urban planning, as highlighted by Professor Rangone in relation to regulation in general^[13].

This use of AI to support or assist urban planning decisions makes it possible to improve management and, as said before, to make the right to good administration effective in the urban planning sphere. The question to ask is whether AI could go beyond helping or supporting human decision-makers (e.g. mayors or town councils) to make the decisions for which they are responsible, but also directly take the decision in question, with a total automation of the urban planning procedure. In several European cities, for example, work is underway to use AI to fully automate the procedure for granting building permits^[14]. I will return to this issue shortly.

3. Artificial Intelligence and (the right to) Housing

Having analysed some of the possibilities of AI in the field of urban planning, in this section we turn specifically to the use of this technology in relation to housing, which is considered a fundamental right by the European Court of Justice^[15] and a right, even a constitutional right^[16], in several European countries.

Again, AI can have several interesting applications here^[17]. For example:

a) Identification of risks of disaster and disaster prevention.

b) Improving the management of the public housing service (or in EU terminology, service of general interest, used by national legislation, e.g. Spanish Housing Act 2023, art. 4): there are experiences in this sphere in the field of care for homeless people, for example^[18].

c) Automation of homes: domotic or smart housing can enable elderly people to stay in their homes (ageing in place^[19]) and save on energy consumption by improving the energy efficiency of buildings^[20].

d) Law enforcement: in relation to inspections and detection of infractions in fields such as, for example, the tourist use of dwellings, a field where several Spanish cities are using AI^[21].

In this field, it is worth highlighting the approval of Regulation (EU) 2024/1028 of the European Parliament and of the Council of 11 April 2024 on data collection and sharing relating to short-term accommodation rental services and amending Regulation (EU) 2018/1724 published in the Official Journal of the EU on 29 April 2024. It entered into force 20 days later and will start to be applied from May 20, 2026.

This European regulation establishes several obligations on both hosts and short-term rental platforms. Hosts must apply to the authorities for a registration number. Once obtained, hosts must give this number to the platforms, which will provide the authorities with the data relating to the host’s activity. With the information they can gather, the authorities will be able to assess the situation and apply the necessary public policies.

e) The application of AI in relation to housing also has its dark side, as it can lead to discrimination against tenants in the rental market. In the USA, tenant screening companies which screen possible tenants on the behalf of landlords or landladies (credit history, previous evictions, criminal record…) are becoming more and more frequent. That use of AI can create problems in the case of incorrect information containing errors, if irrelevant data are used or different treatment depending on race is implemented. The US Administration has issued in 2024 guidelines on the application of equal protection rights existing in US law in relation to this activity^[22] and several cases have already reached the US courts^[23].

4. Legal Limits to artificial intelligence in Smart Cities: the “Humanity Reserve”

Throughout this study we have pointed out the possible legal limits to the use of AI in smart cities in relation to urban planning and housing decisions. It’s now time to address this issue directly.

Is it legally possible to use AI to fully automate administrative procedures related to urban planning and housing? In other words, is it possible to think of an AI system that directly adopts an urban development plan or denies planning permissions?

The answer to this question is, in our opinion, that it depends on the type of legal decision to be taken and the type of AI to be used.

In the case of the use of AI to develop binding powers, there does not seem to be a major legal problem. This could be the case, for example, for building permits if there is no administrative discretion.

In relation to fully automated decisions in the field of discretionary powers, the question changes. The exercise of planning discretion implies a margin of appreciation, a volitional sphere, in which it is necessary, in accordance with the right to good administration and existing case law, to exercise due diligence and due care to take the best possible decision, considering all relevant factors and disregarding irrelevant ones^[24]. Therefore:

1. When using rule-based systems of AI (simple/deterministic AI) that develop deductive inferences: it is not possible to eliminate discretionary powers if they exist, due to the prohibition of fettering by the right to good administration^[25].

2. In the case of AI using statistics (complex/predictive AI) that develop inductive inferences (e.g. machine learning, deep learning): AI has no empathy, whilst being human is having the capacity to feel (some) identification with other humans (except psychos). It is true that AI can imitate empathy (like psychos), but it does not have mirror neurons, and it is not capable of developing real empathy^[26].

It implies that AI cannot exercise the margin of appreciation existing in discretionary powers in a proper way, because it cannot take into consideration all the relevant factors, including the situation of the humans affected by its decisions.

3. Moreover, AI cannot make abductions, that is thinking like human beings with common sense and developing hypotheses. Abduction is a form of reasoning in which assumptions are made to explain observations. For example, if an agent were to observe that a light was not working, they can hypothesise what is happening in the world to explain why the light was not working. An intelligent tutoring system could try to explain why a student gives certain answers in terms of what the student understands and does not understand.

The term abduction was coined by Peirce (1839-1914) to differentiate this type of reasoning from deduction, which involves determining what logically follows from a set of axioms, and induction, which involves inferring general relationships from examples^[27].

4. Therefore, due to both factors, using AI for developing discretionary powers would be a violation of the right to good administration.

5. In the case of discretionary and binding powers, the principle of audi alteram parte included in art. 41 of the ECFR and in national legislations avoids a complete automation of administrative procedures since citizens must have the opportunity to be heard and to have their points of view considered by the public administration.

6. On the other hand, the duty to give reasons (also included in art. 41 of the ECFR and national legislations and now in art. 86 RUEIA regarding explainability, a component of the reasons)^[28] is violated if AI systems that are black boxes are used.

In 1976 Weizenbaum (a MIT scientist) called for a societal consensus that machines should not replace humans in work that benefits from wisdom and empathy, a state that a computer would be unable to have (but might mimic emotions like psychos). In that sense, drawing on the work of other contemporaries looking at specific instances in which AI would be inappropriate, Weizenbaum named the following as roles that ought to be fulfilled only by humans: customer service representatives, therapists, eldercare workers, soldiers, judges and police officers. I think that Weizenbaum felt short with this list: actually, beyond certain jobs, any kind of exercise of administrative discretion should be in the hands of human beings due to the reasons exposed.

We can find several examples of a “reserve of humanity” established by law:

a) National level

German Administrative Procedures Act (VwVfG), Section 35a, fully automated issuing of an administrative act: «An administrative act may be adopted entirely by automatic bodies, provided that this is permitted by law and that there is neither a discretion nor a margin of appreciation».

Administrative procedure Catalan Act 26/2020, Article 44 about Automated administrative action: «2. Only those acts that may be adopted with programming based on objective criteria and parameters shall be subject to automated administrative action».

Spanish Charter of Digital Rights of 2021: «XVIII.6 Efforts shall be made to promote citizens’ rights as regards artificial intelligence recognised in this Charter in the framework of administrative action, recognising in all cases the rights to: decision-making being reserved to persons, in the absence of legislation providing for the adoption of automated decisions with the necessary guarantees».

b) EU level

Art. 22 GDPR “Automated individual decision-making, including profiling”: «1. The data subject shall have the right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning him or her or similarly significantly affects him or her.

2.Paragraph 1 shall not apply if the decision:

A) is necessary for entering, or performance of, a contract between the data subject and a data controller.

B) is authorised by Union or Member State law to which the controller is subject, and which also lays down suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests; or

C) is based on the data subject’s explicit consent.

3.In the cases referred to in points (a) and (c) of paragraph 2, the data controller shall implement suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests, at least the right to obtain human intervention on the part of the controller, to express his or her point of view and to contest the decision.

4.Decisions referred to in paragraph 2 shall not be based on special categories of personal data referred to in Article 9(1), unless point (a) or (g) of Article 9(2) applies and suitable measures to safeguard the data subject’s rights and freedoms and legitimate interests are in place».

The EU regulation on AI of 2024 (REUAI) prohibits its use in some cases included in art. 5 (unacceptable risks), but no mention is made to administrative discretionary powers. However, in relation to judges, Annex III of the REUAI states that high-risk AI uses include:

«8. Administration of justice and democratic processes: (a) AI systems intended to assist a judicial authority in the investigation and interpretation of facts and law, and in the application of the law to a particular set of facts».

Recital 61 REUAI indicates that «The use of AI tools can support the decision-making power of judges or judicial independence but should not replace it: the final decision-making must remain a human-driven activity».

In relation to executive power, Annex III of the REUAI establishes that high-risk AI uses include «AI systems intended to assist competent public authorities for the examination of applications for asylum, visa and residence permits and associated complaints with regard to the eligibility of the natural persons applying for a status». In that case it could imply that AI systems can only assist decision-makers, not replace them. But in other references in Annex III this distinction is not so clear (e.g. «Migration, asylum and border control management»).

It is possible to consider the possibility of extracting a principle from Annex III interpreting words such as “assist”, “evaluate”, “establish priorities” in the sense that machines should be a supporting tool provided there is discretion but not replace human decision-making. In any case, to avoid confusion, such a prohibition of replacing the final discretionary human decision should be explicitly included in the future, in my opinion, in art. 5 of the REUAI to make it clear and to cover other cases in which there is no specific high risk but rather administrative discretion, the use of which can, in itself, constitute a high risk of maladministration and violation of citizens’ rights.

Therefore, in the current state of the art of AI, the precautionary principle and the legal limits seen above impose the avoidance of such uses of this technology in smart cities. Exclusion of fully automated discretionary decisions, taken only by AI, is not a consequence of technical reasons but of legal considerations.

5. Conclusions

As we have seen, AI could be a tool for good administration in the field of urban planning and housing. European cities are using AI increasingly in the context of smart cities. However, the fact that AI can be technically available to be applied in any area of administrative activity creating full automation does not mean it is always legally possible to do so. In this analysis we have discussed the need for a human reserve, that is, that some decisions (discretionary) in relation to some kinds of AI (rule based and non-symbolic) should be in the hands of human beings, due to legal considerations based on avoiding fettering and respecting the right to good administration.

I would like to finish by underlining that one of our big challenges ahead will be to be able to achieve the best from humans and the best from AI in the regulation of land use and housing and in the deployment of smart cities^[29].

1. This study is a result of the research project PID2023-151396OB-I00 (La mejora de las decisiones administrativas mediante el uso de IA, la experimentación y los sandboxes: especial atención al ámbito urbano y de vivienda), funded by the Spanish Government. ↑
2. See for example The Washington Post, 4 December 2023: «A Brazilian city passed a law about water meters. ChatGPT wrote it», https://www.washingtonpost.com/nation/2023/12/04/ai-written-law-porto-alegre-brazil/. ↑
3. OECD, Enhancing the Contribution of Digitalisation to the Smart Cities of the Future, 2019: https://www.oecd-ilibrary.org/docserver/f6970913-en.pdf?expires=1733825724&id=id&accname=guest&checksum=90234C9A8259F0814115ABE4B0D6721B ↑
4. European Parliament, Artificial Intelligence and Urban Development, 2021: https://www.europarl.europa.eu/RegData/etudes/STUD/2021/690882/IPOL_STU(2021)690882_EN.pdf. ↑
5. European Parliament, Artificial Intelligence and Urban Development, 2021, p. 21. ↑
6. Article 41,“Right to good administration”: «1.Every person has the right to have his or her affairs handled impartially, fairly and within a reasonable time by the institutions, bodies, offices and agencies of the Union. 2.This right includes: (a)the right of every person to be heard, before any individual measure which would affect him or her adversely is taken; (b)the right of every person to have access to his or her file, while respecting the legitimate interests of confidentiality and of professional and business secrecy; (c)the obligation of the administration to give reasons for its decision. 3.Every person has the right to have the Union make good any damage caused by its institutions or by its servants in the performance of their duties, in accordance with the general principles common to the laws of the Member States. 4.Every person may write to the institutions of the Union in one of the languages of the Treaties and must have an answer in the same language» ↑
7. In general, see J. Ponce, El Reglamento de Inteligencia Artificial de la Unión Europea de 2024, el derecho a una buena administración digital y su control judicial en España, Marcial Pons, Madrid, 2024. ↑
8. D. Kahneman, O. Sibony, C. Sunstein, Noise. A flaw in human judgement, William Collins, London, 2021. ↑
9. Details available at: https://www.theguardian.com/australia-news/2020/may/29/robodebt-government-to-repay-470000-unlawful-centrelink-debts-worth-721m. ↑
10. Details accessible at: https://www.actu-juridique.fr/administratif/des-erreurs-de-donnees-traitees-par-un-algorithme-peuvent-etre-cause-dillegalite/. ↑
11. However, the EU is increasingly intervening in this field, thanks to the impact of other EU powers on the cities. See J. Ponce, The European union role in ensuring access to affordable housing and in fighting against urban segregation and gentrification: moving forward, in Journal européen des droits de l’homme = European journal of human rights, no. 5, 2019, p. 311-334. ↑
12. See in the case of Barcelona: https://eurocities.eu/stories/barcelona-shapes-the-future-of-city-planning/. ↑
13. N. Rangone, Artificial intelligence challenging core state functions. A focus on law-making and rule-making, in Revista de Derecho Público: Teoría y Método, Vol. 8, 2023, pp. 95-126. ↑
14. For the Vienna case https://digitales.wien.gv.at/en/projekt/brise-vienna/. In Spain, see the cases of Madrid or Vitoria: https://blogs.vitoria-gasteiz.org/medios/2024/07/04/el-ayuntamiento-recurrira-a-la-inteligencia-artificial-para-agilizar-la-concesion-de-licencias/; https://www.madrid.es/portales/munimadrid/es/Inicio/Actualidad/Noticias/Madrid-avanza-hacia-la-automatizacion-para-agilizar-la-tramitacion-de-licencias-urbanisticas/?vgnextfmt=default&vgnextoid=5dbcf5d9881bd810VgnVCM1000001d4a900aRCRD&vgnextchannel=a12149fa40ec9410VgnVCM100000171f5a0aRCRD. ↑
15. Court of Justice, judgment 10 September 2014, Case C‑34/13, KUŠIONOVÁ, ECLI:EU:C:2014:2189, paragraph 65: «Under EU law, the right to accommodation is a fundamental right guaranteed under Article 7 of the Charter that the referring court must take into consideration when implementing Directive 93/13». ↑
16. E.g. Belgian Constitution, art. 23.3, Spanish Constitution, art. 47 or Finnish Constitution, chapter 2 Section 19. ↑
17. See C. Compton and Jessie Hohmann, AI and the Right to Housing, in A. Quintavalla, J.Temperman (eds), Human Rights and Artificial Intelligence: A Deskbook, Oxford University Press, 2023. ↑
18. H. Chan, E. Rice, P. Vayanos, M.Tambe, M. Morton, Evidence From the Past: AI Decision Aids to Improve Housing Systems for Homeless Youth , in Proc. of AAAI Fall Symposium Series on Cognitive Assistance in Government and Public Sector Applications, 2017, https://teamcore.seas.harvard.edu/publications/evidence-past-ai-decision-aids-improve-housing-systems-homeless-youth-0. ↑
19. RL, Fritz, G. Dermody, A nurse-driven method for developing artificial intelligence in “smart” homes for aging-in-place., in Nurs Outlook. 2019 Mar-Apr, 67(2), p. 140-153. ↑
20. R. O. Yussuf,  O. S. Asfour, Applications of artificial intelligence for energy efficiency throughout the building lifecycle: An overview, in Energy and Buildings, Vol. 305, 15 February 2024. ↑
21. For example, Valencia: https://inspain.news/valencia-on-the-hunt-for-illegal-tourist-apartments-with-ai-and-chatbots/. ↑
22. US. Department of Housing and Urban Development, Guidance on Application of the Fair Housing Act to the Screening of Applicants for Rental Housing, April 2024, https://www.hud.gov/sites/dfiles/FHEO/documents/FHEO_Guidance_on_Screening_of_Applicants_for_Rental_Housing.pdf. ↑
23. See the American judicial decision Conn. Fair Hous. Ctr. v. CoreLogic Rental Prop. Sols, 478 F. Supp. 3d 259 (D. Conn. 2020), https://studicata.com/case-briefs/case/conn-fair-hous-ctr-v-corelogic-rental-prop-sols/ . SafeRent, an AI screening tool used by landlords, will no longer use AI-powered “scores” to evaluate whether someone using housing vouchers would make a good tenant. US District Judge Angel Kelley issued final approval for a roughly $2.3 million settlement to prevent SafeRent from discriminating against tenants based on income and race in November 2024, according to information included at https://www.theverge.com/2024/11/20/24297692/ai-landlord-tool-saferent-low-income-tenants-discrimination-settlement. ↑
24. J. Ponce, The Right to Good Administration and the role of. Administrative Law in promoting good government, in A. Cerrillo, J. Ponce, Preventing Corruption and Promoting good Government and Public Integrity, Bruylant, p. 25 ff. ↑
25. P. Craig, Administrative Law, Sweet and Maxwell, Thompson Reuters, 9th edition, 2021, paragraph 10-025. ↑
26. C. Misselhorn, ‘Empathetic’ AI has more to do with psychopathy than emotional intelligence – but that doesn’t mean we can treat machines cruelly, in The Conversation, 21 March 2024, https://theconversation.com/empathetic-ai-has-more-to-do-with-psychopathy-than-emotional-intelligence-but-that-doesnt-mean-we-can-treat-machines-cruelly-225216. ↑
27. See term abduction in the Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/abduction/peirce.html. https://artint.info/html/ArtInt_133.html. Also D. L. Poole & A. K. Mackworth, Artificial Intelligence: Foundations of Computational Agents, 3rd Edition, Cambridge University Press, 2023, https://artint.info/3e/html/ArtInt3e.Ch5.S8.html ↑
28. Article 86, “Right to explanation of individual decision-making”: «1. Any affected person subject to a decision which is taken by the deployer on the basis of the output from a high-risk AI system listed in Annex III, with the exception of systems listed under point 2 thereof, and which produces legal effects or similarly significantly affects that person in a way that they consider to have an adverse impact on their health, safety or fundamental rights shall have the right to obtain from the deployer clear and meaningful explanations of the role of the AI system in the decision-making procedure and the main elements of the decision taken. 2. Paragraph 1 shall not apply to the use of AI systems for which exceptions from, or restrictions to, the obligation under that paragraph follow from Union or national law in compliance with Union law. 3. This Article shall apply only to the extent that the right referred to in paragraph 1 is not otherwise provided for under Union law». ↑
29. As the great American computer scientist J.C. Licklider pointed out many years ago in his famous article Man-computer symbiosis, in IRE transactions on human factors in electronics, 1, 1960, p. 4-11. ↑