FSPL Faculty Scholarship at Penn Law GDPR General Data Protection Regulation HRQ Human Rights Quarterly. UPJIL University of Pennsylvania Journal of International Law WYBAJ Windsor Yearbook of Access to Law.
The adjudicative landscape in South Africa
The next wave of technology
23 Duin en Bakhshi 2017 https://www2.deloitte.com/nl/nl/pages/deloitte-analytics/articles/part-1-kunstmatige-intelligence-gedefinieerd.html?id=nl:2sm:3tw:eng_da_bus:ai1 . 28 Duin en Bakhshi 2017 https://www2.deloitte.com/nl/nl/pages/data-analytics/articles/part-2-kunstmatige-intelligence-technieken-explained.html.
Problem statement and background to study
The most recognized benefit of AI tools in legal practice appears to be the improvement of efficiency. The fact is that South Africa is not even in the top ten on the list of countries that are advanced in terms of technology.46 Therefore, the application of AI systems in the South African legal system and legal fraternity may take some time.
Research question
Since South Africa is a constitutional democracy49, the application of AI will have to be in such a way that it does not harm the fundamental rights of people. This can pose a challenge because AI can only make decisions based on data fed into the system (such as a person's age, race, socio-economic background) and this can turn out to be biased, leading to unfair discrimination may result.50 This may be the major challenge in applying AI in a country like South Africa, more likely than the cost implications involved in its application.
Premises, assumptions and hypothesis
Assumptions
56 McGregor et al. The Universal Declaration of Human Rights at 70 – Putting Human Rights at the Heart of the Design, Development and Deployment of Artificial Intelligence 12. 57 McGregor et al. The Universal Declaration of Human Rights at 70 – Putting Human Rights at the Heart of the Design, Development and Deployment of Artificial Intelligence 26.
Research aim
Research objectives
Research methodology
The US and the European Union were chosen because of their progressive use of AI in pilot projects.
Limitations of study
Framework of dissertation
In this chapter, the current state of the relationship between artificial intelligence and human rights will be discussed. Finally, it discusses how AI and human rights can be reconciled and how the use of AI systems should always aim to promote and elevate human rights.
A brief history of Artificial Intelligence
How AI actually works
90 Prediction Machines from Agrawal, Gans, and Goldfarb: The Simple Economics of Artificial Intelligence 43-51; see Lander and Beigelmacher 2020 https://www.cloudfactory.com/training-data-guide who point out that AI systems can be declared useless without quality data. Therefore, one can conclude that AI systems are only as good as the quality of the data entered.
Types of AI
- Machine learning
107 Perset, Nishigata and Aranda Artificial Intelligence in Society 20 and 21; see also O'Carroll 2017 https://codebots.com/artificial-intelligence/the-3-types-of-ai-is-the-third-even-possible. General AI is thus the first among the types of AI to pass the Turing test.
Big data
- Training data
They did, and they collected a total of 1,460,462 examples from a subset of machine learning system training data.146. Above is an illustration of AI software tasked with processing and dealing with the data fed to it.153 AI systems improve their predictions based on the quality and quantity of data fed to them.154 This happens because the training data is not unchanged and improving through testing, iterations and updates.155.
A brief history of the law and AI
Today, however, AI is already being used in the legal field in the United States, where AI systems are being used to make decisions about whether an applicant should be granted bail or not.172 This is evidence not only of the rapid development of artificial intelligence as a technology, but also its ability to be successfully implemented in law. What is already quite evident, however, is the fact that AI systems can greatly assist humans in the adjudication process by making the process more efficient and possibly much faster.
A brief history of the application of AI within the law
Natural Language Processing
NLP and machine learning technology will greatly enable AI systems to automatically analyze and also render large volumes of legal text material. Since law, as a discipline, consists mostly of written material, if not exhaustively, it stands to reason that NLP systems will become effective tools to help AI systems make better predictions when it comes to legal matters.223.
Working together
This is where AI systems can prove very useful in the judging process; it is currently possible to use AI systems as a detection tool that can check why bailiffs arrive at incorrect outcomes.229. A similar outcome, as illustrated above, is likely to occur in the legal field, more specifically when it comes to bail hearings and trials; due to the simple fact that AI systems and humans are superior in various facets of predictions.233 Therefore, their combination will both compensate for their individual weakness and enable extremely high prediction accuracy, as well as an extremely low misprediction rate.
Conclusion
A brief history of human rights
The Republic of South Africa is one, sovereign, democratic state based on the following values: a) Human dignity, the achievement of equality and the promotion of human rights and freedoms. 246. The fact that the first section in the Constitution mentions the importance of human rights makes their importance in South African society very clear.
Brief overview of AI’s effect on human rights
The current impact of AI on human rights
Another example of AI systems being found guilty of human rights violations occurred in a study conducted by Sweeney.274 Sweeny's investigation into the discriminatory effects of AI systems surprisingly began with a Google search275 of her name: Latanya Sweeny . This therefore means that organizations using sophisticated AI systems must have assessment and monitoring systems in place that are solely designed to protect against human rights abuse by AI.291.
Big data and human rights
A similar legislation to GDPR exists in South Africa; it is called the Privacy Act.301 This Act, like the GDPR, exists to protect people's right to privacy as enshrined in the Constitution.302 The Act aims to protect people's privacy by ensuring that their personal data is processed and collected in a manner that is lawful and in accordance with internationally accepted standards for the processing and collection of people's personal information.303. Although the rules of the GDPR only apply within the EU, international companies doing business in the EU must also comply with them.304 Therefore, the scope as well as the influence of the GDPR has extended beyond the borders of the EU. 305 It has also set a high standard for other non-EU nations to follow in terms of the way states should protect and regulate the use and processing of their citizens' personal digital data to ensure the protection and promotion of human rights.306.
How big data can help protect human rights
However, it should be possible to use big data systems to help promote human rights without compromise. 326 Sarfaty UPJIL 76; See Ramcharan Preventive Human Rights Strategies; Sicilianos Preventing Human Rights Violations (International Human Rights Studies).
The inherent conflict between AI and human rights
344 McGregor et al. The Universal Declaration of Human Rights at 70 – Putting Human Rights at the Heart of the Design, Development and Deployment of Artificial Intelligence 12. The threat to people's privacy is probably the most controversial part of the discussion of AI and human rights.
Reconciling AI and human rights
AI systems on people's human rights.379 This will ensure that the use and application of AI systems in society is done in a way that is aware of the importance of protecting and enforcing human rights. Another way in which AI systems and human rights can be reconciled is by promoting AI literacy among the general public.380 The first step in promoting AI literacy involves first addressing the issue of transparency and opacity .
AI advancing Human Rights
AI in the European Court of Human Rights
AI systems have the potential to be used successfully for the promotion and protection of human rights. 388 Aletras et al PEERJ 3; see also European Court of Human Rights The ECHR in 50 Questions 7.
Conclusion
Using the above fundamentals, the AI system that was used is able to predict ECtHR decisions with an accuracy rate of 79%.393 The results of their experiment have proven that the facts of the case are the single most important aspect that allows for accuracy. of the AI system.394. Medvedeva, Vols, and Wieling conducted a similar study, where they took data from the ECtHR and created an AI algorithm, and just as in the experiment discussed above, they were able to predict the court's decision with remarkable accuracy.395 In their study, they were able to predict ECtHR decisions with an accuracy rate of 75%.396 From these two experiments it is clear that AI systems can be used in such a way that they do not violate human rights; in fact, these experiments have proven that AI systems can be used in such a way as to help promote human rights.
AI systems in adjudication
AI in the Supreme Court of the United States
Katz, Bommarito, and Blackman created an AI algorithm for the specific purpose of trying to detect whether it can accurately predict the decisions of Supreme Court justices.421 As discussed above, the effectiveness (i.e., the accuracy of the predictions) of AI largely depends on depends in part on the quantity and quality of data entered into the system. In the case of the AI algorithm dealing with the Supreme Court, large amounts of data were fed into it, which were of very high quality.422 The AI system was fed data in the form of court cases dating back to the 1800,423.
AI in the European Court of Human Rights
This means that the algorithm created by Aletras et al did not require as much training data as the one created by Katz, Bommarito, and Blackman—although it was slightly more accurate.458. The AI system created by Aletras and others is slightly different, although it is intended to perform the same function as the one created by Katz, Bommarito, and Blackman.
Ensuring transparency
Keeping AI systems accountable
As noted above, AI systems must be transparent in order to be a fully effective tool in trials and all other legal proceedings.489 Transparency in AI systems will go a long way in ensuring that AI systems are held accountable. Because access to the code and inner workings of AI systems will only reveal the actual source code and therefore will not reveal why it made a particular prediction or even what data it used in its decision making process; see also Doshi-Velez et al AI liability under the law: The role of clarification 4.
Legal framework for implementing AI systems in adjudication
As briefly mentioned above, the government indeed recognizes the revolutionary potential of technology such as machine learning, NLP and AI systems.511 However, the South African government has recognized its responsibility to develop policy that regulates these technologies in a way that is holistic and comprehensive .512. It has developed a strategy called the Government Transformation Strategy; and the goal is to help the government take full advantage of transformative technologies, such as machine learning, internet of things and AI systems.513.
Keeping it simple
527 Kroll et al 2017 UPLA 696, this process will involve many iterations made to the AI systems. The process of ensuring that all relevant parties are involved in the development process of AI systems will go a long way in ensuring that those systems are easy to use in adjudication, and other similar and related legal processes.
Conclusion
What this means is that AI systems must be used in conjunction with human judges in the judging process. Therefore, through legislative and other regulatory frameworks, AI systems can be used in adjudication in South Africa (and in other jurisdictions around the world).
Summary
AI systems in bail applications
The experiment showed that judges need some kind of decision-making support,568 especially when it comes to deciding on bail applications and perhaps when. Similar to how a hearing aid would be used if you have hearing problems572, judicial officers should also use decision-making aids when making decisions in court proceedings.
Using AI systems to predict judicial outcomes
Again, what should be made clear is that the use of AI systems must first and foremost be used as aids to referees. Therefore, the combination of AI systems with experienced and skilled judges will result in improved judicial outcomes and therefore improve the judicial process as a whole.
Research outcomes
Without a clear strategy, the use of artificial intelligence systems in the legal field, especially in litigation, is most likely to be successful. Therefore, now is the time to start developing regulatory frameworks for the development and use of artificial intelligence systems.
Recommendations
When evaluating the above extract, it becomes clear that the analogy used by Alertas et al to illustrate how the development and application of AI systems should be treated is completely correct. Therefore, it is of utmost importance that the government, along with having legislation and regulatory policies in place that govern the use of AI systems for adjudication.
Conclusion
OECD The Impact of Big Data and Artificial Intelligence (AI) in the Insurance Sector OECD The Impact of Big Data and Artificial Intelligence (AI) in the Insurance Sector (2020) available at https://www.oecd.org/finance/The -Impact-Big- Data-AI-Insurance-Sector.pdf accessed 24 November 2020. Lewis T 2014 A Brief History of Artificial Intelligence https://www.livescience.com/49007-history-of-artificial-intelligence.html.
