Decoding Artificial Intelligence with Monish Darda

Monish Darda, Co-founder and CTO-Icertis

Artificial Intelligence (AI) and Machine Learning (ML) can be used by businesses to completely transform their operations. This is an edited excerpt from an interview of Monish Darda, the co-founder and CTO of Icertis, conducted by Amit Paranjape, Chairman of the IT & ITES Committee at MCCIA, as a part of MCCIA’s YouTube series, ‘Decoding Artificial Intelligence with Industry Leaders’. Icertis is a cloud-based contract software management company which became a unicorn in 2019. They talk about how Icertis is using its cloud-based architecture from an AI standpoint and the future of AI in the enterprise software space.

Mr. Darda begins by talking about how Icertis is currently going through a positioning exercise. They are moving from contract management to a contract intelligence software company For them, this is a natural transition to ensure that they provide not just data but also knowledge to their customers through their systems and platform.

While talking about the role AI is playing in understanding contracts, he explains that today, the compliance requirements are becoming more stringent and so the value of the contract and the value of the intelligent management of the contract is increasing. There are terms and conditions in the contract which need to be inserted in the business process. AI helps in discovering and extracting information from complex and lengthy paperwork and simplifies the process. AI also connects the people and process involved in a contract to create a good decision support tool for overall business planning and business contingencies. AI and natural language processing (NLP) are also being used in the legal domain, specifically for the paralegal work, that involves looking through large volumes of past data. NLP is used to understand information contextually. For example, in a contract, if words are closer to each other, it can be inferred that it is a jurisdiction or payment term clause and so NLP digs deeper to find inferential meaning in the sentences.

He also speaks about the Generative Pre-Trained Transformer 3 (GPT-3). It is created by Open AI and is the largest language model ever made. It is an AI that can follow the language structure and generate content that appears as though it is written by a human being. It has 17 billion parameters and has exceeded the Moor’s Law in some way. This means that the higher the complexity, the more natural is the output. The GPT-3 can write a poem or a novel or have a conversation because it can figure out the connections between the trillions of megabytes of text it has ingested. Though there have been some experiments with GPT-3, in writing contracts, it is not yet writing full-fledged contracts as there are a lot of factors involved in drawing up a contract which it has not fully mastered. He further says that scaling up GPT-3 will not be a problem if a lot of money is available. According to some estimates, it takes close to 150–200 million dollars to retrain GPT-3 through transfer learning. This involves taking what GPT-3 has already learnt and giving it additional data to train it on a different domain.

Mr. Darda also elaborates on how Icertis is using cloud-based architecture from an AI point of view. The cloud has enabled them to scale up and because it is cheaper, it is possible to train larger models. As the model becomes richer, the learnings can be implemented across customer verticals as well.

When asked if their customers, many of whom are large multi-billion dollar companies, are worried about the learnings that could be used in different scenarios after including them in the general model, he says that there are two aspects to it. The first one is the ethical side, where it has to be ensured that they have rights to use the data, they are not misusing the permission, and are keeping the data unbiased. The second part is called ‘explainable AI’ where it is necessary to use technology such that, the decisions taken can be traced back to the data. Explainable AI is slightly tougher because it is heuristic in nature. In his opinion, explainable AI will be a compliance regime that will have to be dealt with by the entire industry.

He concludes by discussing the future of AI in various sectors. Consolidated data will be the trend in the enterprise software space. While regulations will be tighter, there will be slivers that go end-to-end and disrupt the business process. In his opinion, all negotiations will be assisted by AI..AI will have the strongest impact in areas that are high risk and high volume. While the manufacturing processes are already changing, the insurance sector will be the largest consumer of AI. It may also lead to insurance premiums changing every day. There will be areas like autonomous cars where AI will replace humans. He believes that in 5–10 years, most freight vehicles will be autonomous.AI will also transform the gaming industry. It will be a disruptor if one can have a VR experience with family and friends who are not present in the same location.

(You can watch the full video at https://www.youtube.com/watch?v=VFlVIMxcrKg )

Mahratta Chamber of Commerce, Industries and Agriculture | One of India’s oldest Chambers of Commerce — Established in 1934 | More than 3000 Corporate Members