Artificial Intelligence: The Origin

Artificial intelligence is particularly machine learning that is the machine’s ability to keep improving its performance without human having to explain exactly how to accomplish the assigned task. There is no doubt that within just few years’ machine learning has become far more efficient,insomuch so that we can now build systems that can learn how to perform tasks on their own.

Such peculiar impressive systems have its share of story of origin. Just like everything owes its origin to something or someone, so does the impeccable Machine learning alias Artificial Intelligence.

It all began in a conference, convened by John McCarthy and Marvin Minsky in 1956, where the term ‘artificial intelligence’ was coined by the former and large groups of top researchers of various fields participated to exchange their ideas with regard to the AI’s achievable goals. Subsequently, the interest in AI saw surge during the year 1957 to 1974 with steady advancements at par with the resources and knowledge of the leading brains of that time. The reigning years of AI began in 1990s and 2000s, when in the year 1997, the champion and Grandmaster Gary Kasparov was defeated by IBM’s DeepBlue, a chess playing program and paved way to the artificially intelligent decision making program. The way, we now see AI in our daily routines was just the fragment of one amongst the possibilities, back then, which has evolved over the periods of time and gradually, has taken the form that we see it in now.

Artificial Intelligence (AI): In Present

With the passage of time, machine learning has been adopted as well as adapted to the persistent demands of the time.It operates on a fundamentally different approach rather it can be said that for once it worked on algorithms[i], but now the Machine learning systems are replacing older algorithms in many applications and have become superior at handling many tasks that were once done best by humans. It would not be wrong to say that AI-based systems when surpass human performance at a given task, they are much more likelier to be in demand to use. For instance, Aptonomyand Sanbot, makers respectively of drones and robots, are using improved vision systems to automate much of the work of security guards. The software company Affectiva, among others, is using them to recognize emotions such as joy, surprise, and anger in focus groups. And Enlitic is one of several deep-learning startups that use them to scan medical images to help diagnose cancer.[ii]

Through time testing researches, Artificial intelligence and machine learning has successfully taken a formwith its learning abilities categorized into supervised learning, unsupervised learning, semi-Supervised and reinforcement learning. Amongst these four, the most commonly seen and talked machine learning ability is Supervised learning system(whereby the machine is given lots of examples of the correct answer to a particular problem,for instance, the inputs might be pictures of various animals, and the correct outputs might be labels for those animals: dog, cat, mouse, horse). This system is very well-known and acknowledged in the business world,  even the news that made rounds in the year 2017 is worth mentioning here- In the year 2017, JPMorgan Chase introduced a system for reviewing commercial loan contracts; work that used to take loan officers 360,000 hours, and now which can be done in few seconds.[iii]

The increased utility along with time saving has now become the positive drive to include AI more and more into daily lives, whether be it in the form of voice recognition(millions of people are now using it such as Siri[iv], Alexa[v], and Google Assistant[vi]), Vision systems, such as those used in self-driving cars or ML systems to improve the cooling efficiency at data centers by more than 15% as being used by the Google’s DeepMind[vii] team. Just recently, while keeping up with the times, Honorable Supreme Court of India has adopted the AI system “Supreme Court Portal for assistance in court efficiency” to deal with the huge amount of data.

Challenges and Risk Associated With AI

AI has significantly emerged as a harbinger of technology revolution but its integration in various sectors has raised challenges. Recently, many reports and incidents have come forward which displayed the exponential potential of AI to disrupt the financial stability, social harmony and even, machine bias as well as its ability to control the decision making power of people at large. For instance, in 2018 fortune reported that “Amazon reportedly killed an AI recruitment system because it couldn’t stop the tool from discriminating against women”,[viii]and within the same year, there was another report by the Guardian, which read “A beauty contest Judged by AI and robots didn’t liked dark skin”.[ix] These are just few accounts but are powerful enough to reveal the human biases being projected back at us through the lenses of AI.

With the ability of AI to recognize, interpret, process, and simulate human emotions and the same being used for psychological profiling either to get access to personal data or to spread fake news/ hate speeches, the concerns of safety, privacy of the users or people, at large is in jeopardy. The challenges with AI assimilation in our daily routines is not just limited to Loss of employments, deep fakes, social harmony, unfair discrimination, but lack of accountability has set foot on the ground. The vulnerability of AI systems cannot be overlooked as under malicious attack, perturbation[x] included by attacker has the potential to alter the output of system. Such as, an attack on AI system of stock exchangemay result in huge financial disturbance across the impacted nation or any perturbation in the AI system of traffic management may result in serious consequences, disrupting the traffic and endangering the lives of commuters. Therefore, paying regard to such situations, there has arisen a demand to bind AI with human centric notions and to make it accountable within the legal framework.  

Regulatory and Legal Framework for Artificial Intelligence

The recurrent challenges posed by AI is quite overwhelming sometimes, and might not look a good idea to continue to work with, but still it is contributing in resolving many issues quickly and that too, in cost effective manner including, but not limited to data analysis. However, considering the serious challenges associated with it and which might increase in coming years, the need for legislation and regulatory framework for responsible and accountable artificial intelligence system appears inevitable. Some states have even started working towards the alignment of AI governance practices vis-a vis interests of stakeholders, like Singapore has released “Model AIGovernance Framework”[xi] to serve as guide to develop human centric AI systems. Similarly, United States has ‘Principles for the Stewardship of AI Applications’ and European Union has non-binding guidelines for trustworthy AI that put forth the seven key requirement[xii] that AI system should meet in order to be trustworthy. Moreover, European Union also have GDPR Act, 2016 in place to protect the data and a sort of sectorial regulation for Artificial intelligence.

As regard India, presently, we don’t have any guidelines or regulatory framework exclusively for AI, except for few sector specific regulations such as Medical Device Rules, 2017, SEBI circular on AI, Personal Data Protection Bill, but for the first time, in the year 2018, NITI AYOG released National Strategy on Artificial Intelligence, an approach paper which focused on ‘Towards Responsible AI for All’ wherein, said document discussed in detail the various parameters of artificial intelligence and was launched as a blueprint for the AI ecosystem. Further, to protect the rights and interests of the stakeholders which includes not only public or private sectors but individuals too, who might have to undergo far-reaching effects, while relying upon the AI systems, makes it all the more important to be regulated. Hence, India needs regulatory framework as per the circumstances and impact that AI will have on society, business, financial structure, economic circumstances.No doubt, such structures have to be developed within the confines of the Constitution, while upholding the basic tenets of human lives vis-à-vis AI systems. The need of the hour is to introduce a regulatory framework which can at least lay down the guidelines to promote responsible AI and the first step would be to issue regulations  through amendment in the existing “The Information Technology Act, 2000”.

Conclusion

Machine learning is fundamentally different from the software that preceded it: The machine learns from examples, rather than being explicitly programmed for particular outcome.  Although, AI is contributing in resolving many critical issues and provide quick and cost effective solutions for many commercial and health challenges, but, there are looming risks and challenges associated with it as ML systems have low “interpretability”, meaning that humans have difficulty figuring out how the systems reach their decisions. This creates three risks: the machines may have hidden biases; it is often impossible to prove that an ML system will work in all mission-critical situations; and when the ML system does make errors, diagnosing the problem and correcting it can be difficult.[xiii]Even, NITI AYOG in its draft document has stated that AI may impact three million IT Jobs and ten million manufacturing Jobs, apart from that for a country like India having huge population excess dependency and promotion of AI may impact the employment and livelihood of people, though it’s equally right that alternative employment solutions associated with AI are required to be developed, however, promotion of AI without any limitation may adversely impact the country like India, where, a high percentage of population is illiterate and is dependent on laborious task which majorly relates to manufacturing and other related areas, therefore, India needs a policy document as well as regulatory guidelines for a responsible AI.

Author: Dixit Mehta, Founding Partner of Ductus Legal Law Firm

Co-Author: Neha Sharma, IP Attorney Practicing in The Acme Co. Law Firm

---

[i]A set of unambiguous instructions that a mechanical computer can execute.

[ii] Thomas H. Davenport et al., Artificial Intelligence: The Insights You Need from Harvard Business Review (Boston, Massachusetts: Harvard Business Review Press, 2019).

[iii]Debra Cassens Weiss, ‘JPMorgan Chase uses tech to save 360,000 hours of annual work by lawyers and loan officer’, March 2, 2017, https://www.abajournal.com/news/article/jpmorgan_chase_uses_tech_to_save_360000_hours_of_annual_work_by_lawyers_and.

[iv]Siri is a virtual assistant that is part of Apple Inc.’s iOS, iPadOS, watchOS, macOS, and tvOS operating systems.

[v]Alexa, is a virtual assistant AI technology developed by Amazon, first used in the Amazon Echo smart speaker and the Amazon Dot, Amazon Studio and Amazon Tap speakers developed by Amazon Lab126.

[vi]Google Assistant is an artificial intelligence–powered virtual assistant developed by Google that is primarily available on mobile and smart home devices.

[vii]‘Machine Learning finds new ways for our data centres to save energy’, Environment Projects, September 2019, https://sustainability.google/progress/projects/machine-learning/.

[viii]David Meyer, ‘Amazon Reportedly Killed an AI Recruitment System Because It Couldn’t Stop the Tool from Discriminating Against Women’, October, 10, 2018, https://fortune.com/2018/10/10/amazon-ai-recruitment-bias-women-sexist/

[ix]‘A beauty contest was judged by AI and the robots didn’t like dark skin’, The Guradian, https://www.theguardian.com/technology/2016/sep/08/artificial-intelligence-beauty-contest-doesnt-like-black-people.

[x]Perturbation is the algorithm through which outcome can be disrupted on account of third object interacting with the system.

[xi]‘AI Governance Initiative’, Infocomm Media Development Authority, https://www.imda.gov.sg/AI.

[xii]The seven key requirements includes i) human agency and oversight, ii) technical robustness and safety, iii)  privacy and data governance, iv) transparency, v) diversity, non-discrimination and fairness, vi) environmental and societal well-being and vii) accountability.

[xiii]Thomas H. Davenport et al., Artificial Intelligence: The Insights You Need from Harvard Business Review (Boston, Massachusetts: Harvard Business Review Press, 2019).