Artificial Intelligence (AI) is the new kid on the block, trying to settle in and make friends, but it’s not without its challenges. As Jeff Goldblum’s character in the 1993 movie “Jurassic Park” said, “life finds a way,” and humans are remarkably proficient at adapting to their ever-changing surroundings.
However, when we look at the evolution of the human race, it’s easy to appreciate the sheer scale of the change and marvel at the resilience with which humans have adapted. What’s perhaps less visible is the struggle of the everyday – the fact that we are constantly barraged with something new, unfamiliar, and that requires us to adapt repeatedly.
Today, it’s AI that’s demanding our attention, raising new questions, and making us uncomfortable. But by the time we finally get comfortable with AI, there will likely be something else entirely demanding our attention, making us uncomfortable yet again.
Challenges with AI
One of the main problems with AI is its tendency to hallucinate, or provide false information. In a recent decision of the Supreme Court of India, the Court set aside orders of the National Company Law Tribunal (NCLT) and National Company Law Appellate Tribunal (NCLAT) after noting that NCLT had relied upon non-existent and AI-hallucinated case-law precedents.
This incident is just one of several instances demonstrating the unreliability of AI in its current form. This is compounded by the lack of due application of mind, as AI is essentially governed by its original programming and algorithm.
Carefully crafted prompts and engagement through feedback can minimize hallucinations, but without a reasonable level of oversight, constant filtration, and verification, instances of hallucination will continue to plague the usage of AI, negatively impacting its credibility.
Indiscriminate Usage of AI
Another issue with AI is its indiscriminate usage, particularly in the legal profession. As a lawyer with over 16 years of experience, it’s common to see AI being used to churn out first drafts that lack originality and often have misplaced reasoning.
AI-generated drafts often include facts beyond the prompts and documents supplied as base, which are presented with an air of casualness that’s hardly warranted. This can lead to a complete abandonment of review in drafts so produced, which can have serious consequences.
When the author joined the profession, case-law research software was still in its nascent stages, and incidents of typographical errors, omission of sentences and paragraphs, and jumbled text were common. However, instead of rejecting research software, a different kind of hard work was employed, leading to their improvement and transformation into acceptable authorities.
Adopting AI
To facilitate the integration of AI, it’s essential to frame a methodology on its use, coupled with role-based access. Organisations can collaborate with AI developers to cater to their specific needs, and train people on how to use AI and review its output.
Inculcating a culture of cross-checking and verification is also critical, as well as encouraging originality and rewarding it to minimize AI dependency. Investing in institutional accountability to ensure hallucinated outcomes are identified and rejected in a timely manner is also essential.
Improving data security while using AI, supplemented with a robust legal framework against any unauthorized access to data, is also vital. Updating the Intellectual Property law framework to protect original works against usage by AI is also necessary, as courts have already been confronted with the question of copyright vesting in AI.
It is important to let AI facilitate our lives, rather than replace our cognitive abilities. There is no equivalent to the art of originality. The reason why we become lawyers in the first place, on some level, is to develop the ability to “sell ice in the middle of winter, and that too at a profit.”
We are also inching closer to original works produced by AI, for which we will have to equip ourselves, in terms of law, to consider questions of infringement from a whole new perspective. Courts have already been confronted with the question of copyright vesting in AI – See ‘Stephen
