The ETILC Technology subcommittee assembled on 28th September 2021 to look into the ongoing developments related to AI, ML & NLP in the Indian economic landscape. The members followed a sectorial segmentation while discussing the same, with a particular focus on data extraction using AI. Along with tracking emerging trends in sectors like eCommerce, pharmaceuticals, and , prospective frameworks to strengthen the process of data mining and data science were also touched upon. The session was directed to the cause of bridging the gap between tech-based companies and other vertical industry players.
Pervasion of automation
“Playing with data is the new board game for the Corporates. Data mining is less about the amount of extractable data, but more about figuring out where the unstructured data lies and then structuring it to get it to the desired outcome.”
–Nitin Sahni, Director, Kamadhenu Technology
“Riding on the back of cheap storage and computational cost, synthetic intelligence and cognition have come of age, bringing in unprecedented advancement in extraction of meaningful, actionable insights from large and dynamic data sets. In the end, one needs to keep the business objective in focus while building these data intensive Operative Insight platforms. After all, our clients are interested in furthering their business goals from the insights we provide, rather than the underlying technology itself!”
–Dr. Harsh Vinayak, Sr. Vice President, Intelligent Automation & Data Solutions, NTT DATA Services
The widespread presence of AI and ML in all areas of our life, as well as economic operations, is extremely apparent. In a global survey conducted by PwC India, around 56 % of participants stood by the potential of AI to solve complex modern problems, ranging from intelligent information extraction to cyber security and privacy. The transformation of OCR and template-based data extraction into systems powered by AI and NLP(Natural Language Processing,) have effectively automated and evolved the mining of intelligent data from complex unstructured documents. Its wide-ranging functions encompass document import and classification, image enhancement, character recognition, validation, routing, verification as well as export. This digital revolution is discernible in the workings of QR codes, digital wallets, satellite image analysis, as well as in data analysis of medical images like MRI / XRAY.
In the corporate landscape, automated data extraction is being leveraged to achieve operational efficiency, enhanced customer experience, and risk management. Organizations are developing and synthesizing technical skills (ML algorithms, statistics, programming languages) with experiential skills (data parsing, interpretational skills, NLP) to adapt to the latest technologies.
The Problem of Efficiency
Today, the global spend on technology ranges somewhere between $5.3 to $5.5 trillion. Out of this, New Age Technologies comprise almost 10%. By 2025, this number may increase to 30%. However, this can only be achieved by adopting a framework for efficient data extraction. Far from scarcity of data, organizations have ample data spread out across websites, emails and product reviews, etc. Out of all the unstructured data, we may be only extracting around 10% to 20%. Organizations need to train themselves to assess the nature and position of the data, while also balancing the cost versus the value that the extracted data would result in. Examining data while weighing it against its possible outcomes is absolutely crucial.
Moreover, in cases of dysfunction in extraction and analysis, it has been found that human intervention is crucial to the AI loop. The reason behind this analytics failure may range from a lack of clear vision, improper value assessment, and inadequate analytics translators. From the perspective of content moderation, despite placing algorithms, statistical and analytical processes, human interposition remains vital.
In spite of the encouraging enthusiasm that companies worldwide demonstrate in their want for the adoption of AI, there’s still a significant lack of trust. The efficiency of the incorporation of AI in processing like recruitment is second-guessed.
“Despite the high intent of adopting automated data extraction, clients are often unable to really appreciate the value they might be able to get. It stems from a culture of distrust.”
–Sachin Garg (need his designation from Sakshi/Chetan)
As per a report titled ‘State of Artificial Intelligence in India’ by AIMResearch, AI services positioned in the Retail Sector along with eCommerce can be foreseen to value $5.25 billion by the end of the next decade. The integration of voice-to-text conversion, image-to-text conversion, along with physical and virtual amalgamation into eCommerce websites and other online platforms is transformational.
“We have been able to create data lakes that absorb varied forms of data; this data is then operated on to create useful insights that help improve the shopping experience.”
–Piyush Jha, Chief of Technology & Strategy, APAC | SVP, Digital, GlobalLogic
Apart from this, we have also at hand the creation of AI-fueled stock market predictions along with bot-based auto-generated articles. Moreover, another technological endeavor focuses on transforming sparsely documented codes into easily understandable data via automation. Furthermore, in the Financial Services Industry, AI and ML are being employed to tackle the issue of diversion of funds. Instead of adopting a post mortem approach, the AI utilizes pattern analysis and statistical engines to assess data against arrays of blacklisted companies and management audits. Such elements can access data that is out of the reach of humans.
“In the case of banks, data obsolescence can be countered using semantic analyzers that help track meaningful information, tailor it and teach the machine better.”
–Jaya Vaidyanathan, Chief Executive Officer, BCT Digital
Additional intelligent data extraction trends emerging in the Banking Industry are focused on employing complex algorithms to assess the value space of customers, based on each transaction. This helps in efficient value chain management.
“We are an operational system consisting of the joint functioning of OLAP AND OLTP engines that process sub milliseconds and sub nanoseconds to high-performance computing environments. Our technology layer facilitates value chain management across multiple industries.”
–Nanda Kumar, Chief Executive Officer, Sun Tec Business Solutions
The assimilation of edge AI into infrastructure is also currently being worked on. Embedded engineering creates smart components and software that are capable of capturing real-time data on power frequency in buildings.
“Although we are currently working on integrating edge computing and IoT into biometrics, password authentication, and the infrastructural space, I believe that we are yet to harness the immense potential of intelligent extraction.”
–Mathew Chacko, Chief Executive Officer, Precision Infomatic
In addition to this, AI is also being employed to calculate solar potential before solar panel installation. Satellite imaging calculates the area of rooftops. On the agricultural front, language conversion and translation are useful in extracting information from the field workers’ or shop operators’ local languages. The employment of voice-to-text conversion is extraordinarily serviceable in establishing clear communication and thus, systematic operation. Funding this challenge only affirms the necessary completion of predictive maintenance concerning machine learning.
“We use our technologies to identify crop health. In case of calamity or loss, it helps us anticipate crop damage as well as calculate compensation payable.”
–Agendra Kumar, President, ESRI
According to AIMSResearch, the Healthcare sector contributes only 1.9% to the Indian AI market. However, successive developments in this industry attest to its growth prospects. In 2020, an FDA-approved digital pill that works based on sensors registering stomach fluids and then conveying the data to a patch. This data is perused by both the pharmaceutical companies as well as healthcare providers; it helps ensure that the patients consume the medicine and that the desired effect can be achieved.
“Apart from the problem of data extraction, devices like this pill also address the issue of data creation to solve particular challenges that include anticipation of end outcomes.”
–Ram Singampalli, Chief Operating Officer, Hexaware Technologies
While it is unmistakable that automated data extraction and data science at large are headed to an evolutionary path, concerns related to the same persists. It is crucial that we actively consider and examine the factor of ethicality underneath the employment of all these technological processes. Transparency must be practiced when it comes to privacy concerns. Moreover, data anonymization should comply with data privacy guidelines. One must be vary of misuse of elements like the biometric database for particular agendas, be it political or otherwise. Along with the progress in fintech, smart agriculture, as well as dark technology, it is imperative that we follow the right and ethical direction.
Apart from that, the integration of AI, ML, cloud computing, IoT, and NLP are bound to accelerate India’s economic growth, be it through technological advancement, global collaborations, or job creation.
“While people are worried about losing jobs to AI, I predict that AI will create more jobs than you lose. All we need to do is to keep track of the AI and keep them in line.”
–Prasanna Pendse, Head of Technology – India, ThoughtWorks