This article is written by Shahin Parveen of 1st Semester of Dr. Ram Manohar Lohiya National Law University
ABSTRACT
Artificial Intelligence is a rising field of development. And a hot topic of discussion after the advent of ChatGPT. Artificial Intelligence is an attempt to mimic the cognitive skills of human intelligence in order to perform those tasks which only human can perform. Earlier the AI was used to perform lighter tasks which need complete human interferences, today such AI systems are possible who can work with minimal human interferences, and research is going on for developing an AI system which can perform actions of human cognitive skills without any interference of humans. Growing use of AI is preferred because of its ability to make human life more comfortable through its efficient, accurate and proficient skills. It has issues related to it too. Like the issue of copyright, privacy, and data security etc.
The paper starts with an introduction to Artificial Intelligence. It further goes on with the technical definition of AI along with types of AI to provide a comprehensive idea about AI. In types of AI, weak and strong AI has been mentioned with descriptions. A case of Microsoft Tay has been mentioned to emphasize on how unregulated AI can create problems. The paper further goes on mentioning the legal issues related to AI such as, data privacy and protection, transparency and explainability, intellectual property, AI biases and discrimination, tort liability, liability of AI, ethical issues, and global governance.
The last half of the paper is dealing with legal design for AI which will help in resolving the problems which arises from AI. It also provides the idea of laws that are made or drafted to regulate AI. In the end, it gives the conclusion of the article to give a clear picture to the readers on the topic.
Keywords –
Artificial Intelligence, Data protection and security, Privacy, Transparency, Biases and discrimination, Legal design
INTRODUCTION
Humans are considered as the wisest creature on earth because of their ability to use their brains, which creates an apparent control of humans on other creatures. Some people consider humans are superior to other creatures. Humans create inventions to simplify their lives. But what if the inventions which are created for easing human life start possessing threats and make it more complex. Generally, every human invention creates more or less harm to mankind. But what if the harm can be of as big as destruction of mankind itself. One such invention is Artificial Intelligence. The simulation of human intelligence processes by machines, particularly computer systems, is known as artificial intelligence[1]. Through artificial intelligence (AI), a computer system simulates human decision-making by using math and logic to learn from fresh knowledge. AI has been created for making the good and even better decisions in the place of humans. AI is currently in its initial phase of development, and nobody knows how far it will go. Earlier, the AI had doing small works like navigation e.g., google maps, facial recognition e.g., used by phone for unlocking your phone, and smart assistance like Alexa and Shiri, etc. But the serious issues start arising with the onset of strong AI which are much more capable of doing various work like ChatGPT.
The famous billionaire Elon Musk have said “there is a risk that advanced AI either eliminates or constrains humanity’s growth.”[2]. The complexity and vastness of this emerging field of AI does not allow us to have such prediction right now but the threats from AI are genuine in various context. Having features to analyse huge toll of data with efficiency, accuracy and consistent results increases the productivity of companies, which tempt them to substitute AI instead of human labour forces. How quickly the machine is being incorporated into industries today is a glaring example of this. Employers are continuously engaged in this process of substitution. But this substitution can be problematic as there are deficiency of laws dealing with AI specifically. Talking about India, it does not have any law on AI, it does have IT laws and some guidelines on data protection. Exploration of legal issues with AI, and laws on it is highly required.
ARTIFICIAL INTELLIGENCE
A broad area of computer science called artificial intelligence (AI) is focused on creating intelligent machines that can carry out tasks that traditionally require human intelligence. Despite the fact that AI is an interdisciplinary discipline with many different approaches, developments in machine learning and deep learning in particular are ushering in a new era in almost all areas of the IT industry.
Machines can mimic or even outperform human brain functions thanks to artificial intelligence. Additionally, AI is gradually becoming a part of daily life and is a field in which businesses from every industry are investing, as evidenced by the emergence of self-driving cars and the proliferation of generative AI tools like ChatGPT and Google’s Bard.
So basically, AI has function to perform those tasks which are based on human cognitive skills. Sometimes it learns to perform the task without the help of human intervention and sometimes it learns with the supervision of humans. [3]
TYPES OF ARTIFICIAL INTELLIGENCE
On the basis of their abilities to perform works, AI has been divided into two categories:
- Weak AI
Weak AI has limited functions. Weak AI concentrates on a single task and performs it much better than humans while working under significantly more restrictions than even the most primitive human intellect. Due to its restricted capability, it can easily automate that particular operation, and its narrow concentration has enabled it to drive numerous technical advances in the previous few years alone. It is also termed as “specialised AI” as it performs specialized tasks. It is the maximum that have been done in the AI field till now. It has various examples such as chatbots, Spotify shuffle that decides next song in playlist according to the mood, Email spam filters which keeps the spam messages separated from important one, smart assistances like Siri, Cortana, and Alexa, self-driving cars or automatic cars, google maps etc.[4]
2. Strong AI
Strong AI is nothing greater than an idea of making such kind of AI who matches cognitive levels of human and have all the special intelligence that humans only humans have. This idea of AI is not just to match the human intelligence but to create an artificial intelligence which can even exceeds the human intelligence. This type of artificial intelligence has the capacity to act and behave in the same manner as people. AGI mimics human general intelligence and can learn new skills and solve problems in a manner akin to our own. Strong, or general, artificial intelligence can generalise information and apply that information to other tasks, plan ahead using the most recent information, and adapt to an environment as changes take place. There are no real-life examples of strong AI as it really does not exist, but we can see examples of it in fictional movies such as Star Trek: The Next Generation.[5]
Till now everything about AI seems nice. But every invention is prone to be misused. Since humans are not perfect in the same way their inventions too. Even though the development on AI is still going on and is far to go, but the massive use of AI in every field and industries introduces new ways of crime and appeal for regulations.
Microsoft Tay case
Microsoft Tay was a chatbot developed by Microsoft company in the year 2016. It was a twitter account through which it can tweet, comments and reply to the comments. Basically, it can perform all the feature a normal twitter handle can perform. It was an experiment by the company in “conversational understanding”. It was design in a way that it would learn from the experiences it will gain after talking to humans. Soon after the launch within 15 hour the chatbot had to made dysfunctional as it started tweeting racist, misogynist, and other kind of discriminatory tweets. [6]
These kinds of instances have been occurred which raises questions like, what if smart assistance learn people’s biases prevailing against different communities and start targeting people of that communities. Various time it happens that weak AI like google map shows wrong route which even does not exist, but it has learnt that from some sources. Which shows the weaknesses of AI through which it can be manipulated and hence misused.
LEGAL ISSUES
Humans have a variety of cognitive biases, including confirmation bias, which is manifested in our behaviours and, consequently, in our data. Biasness is one of the various legal issues arises from AI. There are many other issues too.
- Data Privacy and Protection
AI being a non-human relies on data from humans to work. It uses vast amount of data that it has collected in doing its job. As it works rely highly on data and it collects data from people. People are concerned about security and privacy of their data. As some data are common some are confidential, data privacy is one of the main legal challenges surrounding associations’ use of AI.
The associations using AI must ensure that the collected data must be used in adherence to the privacy laws of that nation and should comply with the international standard, and not use otherwise. Confidential data should be avoided to put on AI, it does not have that intelligence to differentiate between the owner of the data and third party. If confidential information has been put on AI, there must be use of specific algorithms to protect it in order to maintain its confidentiality from third party. [7]
2. Transparency and Explainability
AI also include transparency and explainability. People may find it challenging to figure out how decisions are being made since AI systems can be opaque and challenging to comprehend. To ensure that AI systems are open and understandable, and that people can question decisions made by AI, laws and regulations will be required. Businesses must make sure they can demonstrate the fairness, precision, and transparency of the decisions made by their AI systems and provide justification for them.[8]
3. Intellectual Property
As the work AI do are based on prior information feed in it. The data it generates on the demand of the user is not the original work of AI, in reality it is work of someone else. It might be possible that based on the prior information AI would create some new information on such new work of authorship like articles, artistic work etc. In such scenario of new work, the concerned association can ensure their rights on it. But they have to be mindful while working with AI that they are open about who or what developed such works, and they have the required rights and licences to use and distribute them and ensure that they do not infringe any third-party copyright, patent, or trademark rights.[9]
4. AI Bias and Discrimination
Another legal issue with AI is biasness and discrimination. In particular, if AI systems are trained on data that reflects historical prejudices or inequities, they may unintentionally perpetuate prejudice and discrimination. Associations must ensure that their AI is free from biases or discrimination on any ground. Associations must have methods to detect biasedness and correct it. If someone has been affected by those biases, there should be a mechanism to redress it[10]. For example, the discrimination by Amazon’s algorithm against women. Amazon’s automated recruitment system was used for evaluating applicants on the basis of their resume. By examining the resumes of prior applicants, the system learned the ability to determine whether a candidate was qualified for a position. The AI system assumed that men were deliberately chosen since women were previously underrepresented in technical professions. As a result, it gave female applicants’ resumes a lower ranking[11].
5. Tort Liability
There can be a legal issue of tort liability in working of an AI. The association could be held accountable for any damages if an AI system generates erroneous, careless, or biased results that hurt members or other end users. In case of health sector, it could have disastrous consequences. As a result, associations must make sure that their AI systems are trustworthy and accurate, and that all work products that arise from them are thoroughly examined for accuracy, veracity, completeness, and effectiveness. Associations must have insurance policies in order to guard against potential liability claims in each of these areas of legal risk.[12]
6. Liability of AI
There is huge debate going on from many years for considering AI as a person. The debate has not been concluded and in several instances different court has taken stand that only humans constitute person. This arises the question that who should be blame for the fault of AI. Laws must need to take a clear stand on the accountability of actions of AI use by individuals and organisations. People must be aware of the legal liability arise from the usage of AI for any work. [13]
7. Ethical issues
Ethical issues are also important issues connected to legal issues. AI raise a lot of ethical issues such as its implication of AI for employment, impact on society, and the potential use of AI for harmful purposes. AI accelerates progress and, in many circumstances, outperforms our capacity as humans to keep up with it not only in positive manner but in negative manner too. Such as, use of AI has accelerated hacking multiple times.
8. Governance at International Level
As AI is not confined to one country or state, it raises issue of international governance. AI is borderless, a person sitting in different country can use the AI software of different country. There is need of international laws and standards to maintain uniformity across the borders. So, people can comply with them, and conflicts can be avoided.
LEGAL DESIGN
Legal design is an idea to develop a design of AI with features addressing the ethical and legal issues related to AI. Legal design aims to bridge the gap between legal professionals, AI developers, and end-users by employing visual elements, intuitive interfaces, and user-centred approaches[14]. Legal designs have different attributes which will address the issues related to functioning of AI.
- Transparency
One of major issue related to AI is lack of transparency and explainability. Legal design can help in resolving the issue, with the help of interactive diagram, intuitive dashboards, and flowcharts, it enables the stakeholders to understand how AI operates.[15]
2. Compliance and Accountability
Legal design will be created to facilitates to create AI system in the compliance with legal requirements. It will help in drafting user-friendly terms of services, privacy policy and consent form etc[16].
3. Justice and fairness
Bias and discrimination are one of the important concerning issues with AI. Legal design will help in mitigating these biases by identifying and rectifying it by involving equality, multidisciplinary teams and promoting fairness within AI system. [17]
4. Privacy
Privacy is one of the major concerns for the users on the advent of AI. As the work of AI is based on the data it collects from users. Users constantly worry about how their data is being used. Legal design helps businesses develop visually appealing user interfaces that let customers understand and have some influence over the data that AI systems collect and process. Legal design enables people to make informed decisions and protects their privacy rights by offering transparent consent procedures, user-friendly privacy settings, and transparent data protection rules[18].
The solutions offered by legal designs should be included to solve the legal issues concerning AI. Inculcating legal design in an AI system will make it user-friendly and workable overall.
LAWS ON ARTIFICIAL INTELLIGENCE
There are various countries and organization attempting to create laws on Artificial Intelligence. Like China, Brazil and European Union (EU). EU has even passed a law recently for regulating the AI.
- European Union’s Artificial Intelligence Act
The European union have tried to make AI fit for future use by making a legislation to regulate AI. As it is quite clear that AI is future of world, these regulations have attempt make them fit for continuing use while respecting the privacy and security of data along with other ethical and legal concerns. It has provided a definition of AI along with categorization in AI. It has divided AI in four categories on the basis of the risk the poses: unacceptable, high, limited and minimal risk. It has introduced different guidelines for different level of AI to which the businesses have to comply. In case of non-compliance heavy penalty will be imposed. Also, in case of false compliance, punishment will be imposed. It has also introduced regulation for general purpose AI, which is used for different purposes like ChatGPT.[19]
2. Brazil
Brazil is drafting its first artificial intelligence law too. The three main pillars of AI regulation in the draft of Brazilian regulation, are protecting the rights of those impacted by the technology, identifying the level of risk, and forecasting governance procedures for businesses that produce or run AI systems. Also, it does categorization and poses obligation with respect to the risk category. It grants rights to the data subjects against users and providers of the AI system. It dictates the terms for governance of AI and also introduces the liabilities and sanctions related to it. [20]
3. India
Talking about India, India does not have any codified law specifically dealing with AI yet. But there are several laws that is used for interim regulation of the functioning of AI. Such as the IT Act of 2000, which regulates e-commerce and cybercrime. It penalises for the offences like hacking, publishing private image of someone, obscene contents, cyber-terrorism, breach of privacy and confidentiality etc.
CONCLUSION
Artificial Intelligence is a rising tool which will change the shape of mankind in few years. AI is primarily a machine with ability of human to perform certain works. It is in its developing phase right now and is growing exponentially. Every now and then new developments occurs in AI field. In today’s scenario humans are steadily getting dependent on AI even in their domestic arrangements. AI’s quality to make human life more comfortable leads to its rapidly increasing use. But everything comes at a cost, is same with AI. There are a lot of legal and ethical issues related to the use of AI, which need instant addressal as the use of AI is almost universal now, everyone is using it either on a large scale such as for the purpose of refining CVs of candidates or for playing music suggested by music app. The major issues related to AI are security, privacy and protection of data, discrimination, biases, intellectual property rights, transparency, tort liabilities and global governance etc. These issues can be addressed by using legal design for AI which will help in maintaining transparency, free from prejudices, data privacy and security etc. Some countries have already started to draft legislation to regulate this upcoming future of AI, a few have even passed the law. In order to safeguard the rights of its citizens who utilise AI in their daily lives, India must take serious action in response to this important development.
[1] Artificial Intelligence, tech target, available at https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence, last seen on 12/07/2023.
[2] Elon Musk warns AI could constrain humanity’s growth, Yahoo Finance, available at https://finance.yahoo.com/video/elon-musk-warns-ai-could-134212033.html?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAAMqFMSHGLQfhBrk6uFGZcucMKEoaalkK6UFie4abjH1nAwOo8ZnYu-N-vQMLflVrfF33vVJJrMS7g08b9xYT3HSuy5pj3GAiYtdYc2-7i2z-0FF5Wx8fJvDTxZD5QNWoOsK1W0juG28fm6RX0-AgSqsEkyzVEQGKyk9XnV16zRzv#:~:text=%2D%20Elon%20Musk%20is%20issuing%20a,us%20now%20with%20the%20details. ,Last seen on 12/07/2023.
[3] Alyssa Schroer, Artificial Intelligence. , Builtin, Available at https://builtin.com/artificial-intelligence/strong-ai-weak-ai last seen on 12/07/2023.
[4] Ibid.
[5] Ibid.
[6] James Vincet, Twitter taught Microsoft’s AI chatbot to be a racist asshole in less than a day,The Verge, available at https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist, last seen on 12/07/2023.
[7] Jeffrey S. Tenenbaum, Five key Legal Issues to consider when it comes to Generative AI, Asae, available on https://www.asaecenter.org/resources/articles/an_plus/2023/4-april/five-key-legal-issues-to-consider-when-it-comes-to-ai,Last seen on 12/07/2023.
[8] John Giles, Legal issues related to AI according to ChatGPT, Michalsons,available on https://www.michalsons.com/blog/ai-law-according-to-chatgpt/63492#:~:text=AI%20bias%20and%20discrimination&text=AI%20systems%20have%20the%20potential,discrimination%20have%20avenues%20for%20redress., last seen on 12/07/2023.
[9] Supra 7.
[10] Supra 8.
[11] Supra 7.
[12] Ibid.
[13] Supra 8.
[14] Nathan-Ross Adams, How legal design helps you leverage AI lawfully, Michalsons, available at https://www.michalsons.com/blog/how-legal-design-helps-you-leverage-ai-lawfully/66722, Last seen on 12/07/2023.
[15] Ibid.
[16] Ibid.
[17] Ibid.
[18] Ibid.
[19] The European Union’s Artificial Intelligence Act-explained, World Economic Forum, available at https://www.weforum.org/agenda/2023/06/european-union-ai-act-explained/#:~:text=Jun%2030%2C%202023,various%20development%20and%20use%20requirements. , Last seen on 12/07/2023.
[20] Dr. Benedikt Kohn, AI regulation around the world, Taylor Wessing, available at https://www.taylorwessing.com/en/interface/2023/ai—are-we-getting-the-balance-between-regulation-and-innovation-right/ai-regulation-around-the-world,Last seen on 12/07/2023.
