The topic for today's discussion at first look can appear extremely confusing and hazy. However, a deeper analysis would show that the topic is extremely well thought out and relevant.
The term Artificial Intelligence ("AI") has been used in recent years in almost every field of human life. Every social and economic aspect seems to have some benefit from the use of AI.
Intelligence is a gift which human beings have been vested with naturally. The level of intelligence varies from species to species and generations to generations. However, when intelligence is embedded in a machine, not naturally but by enabling it in the form of input, analysis and output, it is machine intelligence or AI. The advantage of artificial intelligence or AI as is commonly known is the capability to be able to process and analyse very high volumes of data for arriving at conclusions based on some pre-determined parameters.
For example, if an AI software is run on this particular event, by giving it the relevant data i.e., the footage, at the end of the event, the software can throw up results as to how many attendees were attentively listening, how many attendees had simply logged in and how many attendees were actually yawning and were bored 😊.
Thus, AI can be utilised for the most mundane jobs such as starting the microwave or the washing machine at a pre-determined time upon a simple command, or predict the outbreak of a pandemic.
As per the WHO's website, it received information from Wuhan about the virus on the eve of the new year i.e. 30th December 2019. A week thereafter, the first alert from the WHO about the coronavirus outbreak was issued. Flights were however continuing across the world and movement of people was not stopped. The pandemic was finally declared as a PANDEMIC on 11th March 2020.
However, a Canadian AI company called BlueDot had used AI- powered algorithms to analyze information from a multitude of sources to identify disease outbreaks and forecast how they may spread. By sifting through 100,000 news reports in 65 languages a day, BlueDot recognized patterns between health outbreaks and travel, and made predictions about the COVID outbreak on 31st December, 2019. If this prediction had been immediately transmitted across the world and international travel had been curbed, maybe the pandemic could have been contained.
The speed of AI though, sometimes threatening and overwhelming, can be used in certain areas for the betterment of society and humanity as a whole. So how does AI work and what is its interface with Intellectual Property and with Covid-19?
As the world continues to wrestle to find a prevention or cure for the virus, AI can be and, is in fact being utilised, for analysis, research of large quantum of data to identify various aspects of the virus. For example:
The challenge with AI is three-fold –
1) The availability of data, which could be privacy protected data or copyrighted datasets
2) The parameters or inputs required to analyse the data.
3) Finally, the treatment to be meted to the conclusions of the said analysis.
During the outbreak of Covid-19, reports have been received in respect of various strains of the virus, which have affected people. The strains are not identical and neither are the symptoms nor the intensity of the disease. While research is taking place in separate pockets across the world, there is no humanly possible way to analyse the global data emanating from the virus without the use of AI.
The sheer magnitude of analysis of this data can be imagined by considering the large number of players/ entities, who are involved in the diagnosis, management, treatment, prevention and cure of this disease. Diagnostic laboratories, government departments doing contact-tracing, the various applications and the data collected by the said applications, the medical records of all those persons who have undergone testing, the nature of treatment given in different parts of the world, the mortality data, the recovery data, quantities of masks, PPE kits etc. being used to predict the required demand and to correlate this with production – all this data cannot be analysed without the power of AI. Thus, AI can be an extremely efficient tool in finding a fast and effective solution to this unprecedented pandemic outbreak.
The global community is conscious of the power of AI. Hence, there are various steps that have been taken to facilitate the use of AI for the protection of public health. Various AI tools such as CORD-191, COVID-19 Research Explorer,2 COVID Scholar3 are all AI-based tools, which are made available to scientists and medical researchers to give the input data and to obtain answers to their various queries.
So long as the data is not copyright protected, it can be used for analysis and research and for providing positive outcome.
The interface between IP rights and AI can be of two kinds:
1) IP can be a barrier, which shields data from being used by AI softwares.
2) IPR can be used to protect the outcomes of AI
Thus, IP rights can be both a sword and a shield while dealing with AI. To ensure that IP is not a barrier, it is important to make access to copyrighted databases easy and possible. Organizations across the world are conscious of the need for making copyrighted databases easily available. Various endeavours have been made by right-holders to not insist on IP rights during the Covid-19 outbreak. Examples of these endeavours are as under:
In addition, various countries have enacted special provisions in their copyright laws to make Text and Data Mining (TDM) easier. Countries such as UK, Japan, and the EU have incorporated specific fair dealing and fair use provisions to enable use of copyrighted material and also reproduction of lawfully accessed copyright materials for use in AI systems. Examples of such statutes which have been amended are
"Copies for text and data analysis for non-commercial research
(1) The making of a copy of a work by a person who has lawful access to the work does not infringe copyright in the work provided that—
(b) the copy is accompanied by a sufficient acknowledgement (unless this would be impossible for reasons of practicality or otherwise).
"(Exploitation without the Purpose of Enjoying the Thoughts or Sentiments Expressed in a Work):
It is permissible to exploit a work, in any way and to the extent considered necessary…
(iii) if it is done for use in data analysis (meaning the extraction, comparison, classification, or other statistical analysis of the constituent language, sounds, images, or other elemental data from a large number of works or a large volume of other such data"
"Article 3: Text and data mining for the purposes of scientific research:
1. Member States shall provide for an exception to the rights provided for in Article 5(a) and Article 7(1) of Directive 96/9/EC, Article 2 of Directive 2001/29/EC, and Article 15(1) of this Directive for reproductions and extractions made by research organisations and cultural heritage institutions in order to carry out, for the purposes of scientific research, text and data mining of works or other subject matter to which they have lawful access.….."
"Article 4: Exception or limitation for text and data mining
1. Member States shall provide for an exception or limitation to the rights provided for in Article 5(a) and Article 7(1) of Directive 96/9/EC, Article 2 of Directive 2001/29/EC, Article 4(1)(a) and (b) of Directive 2009/24/EC and Article 15(1) of this Directive for reproductions and extractions of lawfully accessible works and other subject matter for the purposes of text and data mining.…."
Insofar as India is concerned, a database is protected as a literary work under Section 2(c) of the Copyright Act. In view of the fact that databases, which are currently being assimilated in various organizations, would not be individual-centric but machine-centric, these databases would be computer-generated databases. Thus, the author of these databases under Section 2(d)(vi) would be the person or entity who is causing the work to be created. Thus, in the case of laboratories, hospitals, government entities, research organizations, university etc. who are collecting and assimilating data, the respective entities would be the authors of these collective databases and would be the owners of the copyright in respect of these databases. The use of these databases for the purpose of research would be permissible so long as it constitutes fair dealing under Section 52(1)(a)(i). Insofar as the law of fair dealing is concerned, the legal principles governing fair dealing are quite well-settled4. These judgements clearly lay down that fair dealing is permissible, especially if the use is of a transformative nature. Broadly, the purpose of the use of the work, the commercial nature of the exploitation, the competition that it may provide to the original owner, the character of use etc. would determine as to whether the use is fair or not.
Thus, insofar as the use of copyrighted datasets is concerned, the respective entities, who own the copyright therein, can do their own research using AI softwares. If any independent researcher wishes to use the same, it may be permissible if it is fair dealing. One word of caution would be the use of medical records of patients available in laboratories and in hospitals. Recently, Justice Srikrishna who headed the Committee for drafting India's data protection law, while speaking at a Webinar on the challenges in personal data protection, said that in the situation of Covid, where data is necessary for statistical probability, the law should ensure data anonymisation, where only numbers and no personal information can be utilised.
Coming to the second aspect i.e., the outcome of use of AI, what are the rights and who owns the rights in the same?
Various rights could be generated from the outcomes of AI. Several copyrightable data sets may be created. Diagnostic tools, better pandemic management products, quicker manufacturing using a combination of AI and 3D printing, potential vaccines and drug molecules etc., may be generated using AI analysis. The rights in these outcomes would belong to the organisations which have undertaken the research using AI. All outcomes of AI are initiated by humans who are in turn employed by organisations. The same rules as are applicable to other forms of IP generated by these organisations would be applicable even here. The fact that AI may be used, does not vest the IP in the computer or the AI tool.
Coming now to AI and law - how AI can be used to speeden up dispensation of justice, there are various solutions that are available to use AI as an effective tool for justice dispensation. AI softwares would primarily require input data and analytical parameters to arrive at a conclusion. In law, the input data would be the facts and documents relating to a particular case, the parameters for analysis would be the settled judicial precedents and the applicable statutes. The outcome would be the recommendation of the AI system as to what the order or judgment should be.
To put it simply, if the facts and the documents of a case are fed into the AI system and the AI system is pre-programmed with the relevant statutes and the case laws, it would apply the facts to the law and give the most suitable conclusion based on the said parameters.
An algorithm would be, therefore, required to be developed at 2 levels. Level 1 would be the structure of such an AI system, which could be a constant for all legal disciplines. Level 2 would be the parameters for a specific area of law or discipline. Each AI system developed in this manner could be used for expedited adjudication of the cases, especially in areas which are not so complex. For example, in case of traffic challans, complaints under the Negotiable Instruments Act, bail applications, and other similar areas, AI tools can be easily used to predict the outcome.
There could be concerns such as bias in the parameters which are programmed and the lack of a humane touch in adjudication. It is not in every case that a human touch would be required to adjudicate a dispute. There are various examples wherein AI has been used for granting a bail, or for resolving a traffic challan and has been proved to be extremely effective. For instance, the validity of outcome by an AI based bail recidivism tool (COMPAS) was judicially upheld by the Supreme Court of Wisconsin in US in the case of Loomis v Wisconsin.5 The Court said that if used properly with an awareness of the limitations and cautions, a circuits court's consideration of a COMPAS risk assessment at sentencing does not violate a defendant's right to due process.
The interface between AI and law is one that could provide a very effective solution for expedited adjudication of certain categories of cases where human intervention can be minimal. The use of AI in law could be easily described as PAC- Law: Processing, Analysis and Conclusion.
The conclusions, which are arrived at using AI systems, can also be supervised or monitored by a fully qualified Judicial Officer. The time required to simply monitor or supervise, even on a random basis, would be much less than what is spent today. AI systems in law can provide enormous assistance to Judges, who require research on a daily basis in particular fact situations. The data that is collected on the National Judicial Data Grid can, if subjected to an AI analysing software, produce miraculous results in terms of reducing the inconsistency in decision making, contradiction in decision making, and inefficiency in decision making.
Thus, PAC can be an effective AI tool used in the field of law for the purpose of processing analysis of facts and law to arrive at a just conclusion.
Every technological tool comes with its own risk. However, before employing such tools, the risk-benefit analysis would have to be done in a gradual and a phased manner so as to ensure that the AI tool or system does not bypass the Judge. For those who have read Dan Brown's Origin, where an AI software bypassed its creator to create havoc, the picture can be bleak. However, the sheer large quantum of AI tools being currently used in the world in almost all fields, clearly, shows that such outcomes as in this book are at best, fictional. Adequate controls can be placed to ensure that AI is used just as digital technologies for greater efficiency, better solutions with a human face.