| 46 | | 47 | SPECIAL FEATURE: FRACTAL.AI Navigating the Interplay of Sustainability and Artificial Intelligence the data was obtained through consent and that it is being used in a secure environment. Prior to testing with realworld data, using synthetic data that imitates real-world data produced using generative or advanced AI models offer a secure alternative for training and experimenting with the test model, thus, avoiding data being compromised. Data minimisation and restricting access to only necessary data points further enhance security. With proper address and lineage, it can be tracked where the data is being employed averting unnecessary data exposure. This also helps in reducing carbon emission and making it a more sustainable practice. Machines or technologies stealing jobs is an incomplete tale. The advent of ATM machines in the 1970s stirred bank tellers as they feared unemployment. Nevertheless, the requirement for teller jobs increased as ATM machines made operating a branch cheaper, giving rise to increased number of branches and raising the demand for bank tellers. Now, their marketing and interpersonal skills have become more important. Similarly, integrating AI into our systems would redefine how we work, increase our productivity, enhance customer service, and benefit the economy as it would increase the number of jobs. The onus lies on us to embrace the change, adapt and upskill, and rid ourselves of an archaic mind-set. While such cutting-edge technology makes our lives convenient, many question if we have become too dependent on it. Over-dependency on technology may lead to social chaos as people would intend to perform every task via technology reducing human interaction. However, as humans are hardwired to interact with humans, the over-dependency would not last long. Social movements and protests are enabled through technology and their participation in large numbers underscores its positive influence. The Net-neutrality movement in India gained support from the populace through wise usage of technology. AI breakthroughs are crucial for knowledge about climate change and predicting its potential consequences. Similarly, predictive AI has tremendous potential for regulating our energy usage. For instance, it may moderate the engine and software of a car for better fuel efficiency, regulate the temperature of the air conditioner for power efficiency, and schedule washing machine runs With the rising awareness of the imperative of sustainability, it has become a paramount consideration for investors, consumers and companies alike. As we drive intensive measures to assess and establish sustainable practices, the intersection of sustainability and Artificial Intelligence (AI) promises a new evolutionary paradigm. Sustainability in the realm of AI divulges through the dual perspective of leveraging AI for sustainable outcomes while ensuring the very sustainability of AI itself. At its core, AI has the potential to enhance productivity while curtailing carbon emissions across sectors. While the intent is harnessing AI responsibly to nurture societal and environmental well-being, concerns still remain. Developing AI, research, and commercial development of models demand massive computational resources which have a significant carbon footprint. Globally, Google’s data centres use about twice as much electricity as the city of San Francisco. A critical first step in addressing this challenge is adopting sustainable practices at the organizational level. Pre-emptive thinking is necessary to strategize and optimize processes before creating multiple models. Additionally, the intent needs to shift to considering existing alternatives, such as training a smaller model to imitate the behaviour of a larger, already trained model. Rationalised AI development would go a long way in not just reducing emissions, but also minimising data usage. It is a common misconception that since organisations don’t employ data specific to race, region or gender the data would be devoid of biases. However, inadvertent biases often emerge through proxy indicators such as zip codes indicating colour or nationality and purchasing pattern reflecting marital status and gender. This necessitates awareness of possible biases in the data, quantitatively assessing the discrimination, and thereafter, having the right mechanism to mitigate them. However, integrating sustainable practices in the development of AI models should be inherent. As a car is accompanied by a seatbelt by default, so should sustainable practices be incorporated in machine learning or predictive modelling or AI workflows. As digital breakthroughs benefit us invaluably, they also raise concerns about privacy and security of information. These concerns should be addressed by ascertaining that Sray Agarwal is seasoned AI and analytics professional with a rich background spanning Financial Services to Hospitality. With a track record of pioneering Responsible AI frameworks for prominent banks in the UK and the US, Sray’s expertise is rooted in datadriven innovation. Additionally, he recently authored a book on Responsible AI published by Springer and has lent his expertise as a visiting faculty member at renowned institutions like NUS, ISB, and Jio Institute, where he imparts cutting-edge AI education. As a Principal Consultant at Fractal, Sray leads the charge in AI and Responsible AI endeavors. Sray Agarwal Head of Responsible AI, Fractal.ai to optimise its capacity percentage. The data collected by AI can be used by AI to reduce carbon emissions. Globally, cumulative data centralized for informed decision-making at the national level can contribute to the formulation of effective policies and practices, with a holistic overview of the process. In conclusion, responsible AI should be the norm as it can prevent revenue, reputation, and regulatory risks. But regulatory frameworks implemented by the government adjust slowly to the ever-evolving technology and often have gaps in addressing various aspects of technology. AI itself is not one technology or a singular development. It is a bundle of technologies whose decision-making abilities are a dynamic work in progress. Hence, AI governance cannot only be a top-down approach wherein legislative adoption and compliance is the only box to be checked. This responsibility trickles down to the lowest common denominator, the average consumer. However, it needs to be regulated at the organisational level through comprehensive guidelines and risk frameworks. Crucially, it must be executed in the form of technical safeguards that actively function and interact with the AI projects within the organization, for a safe and sustainable future, where AI is an enabler of unprecedented progress. •
RkJQdWJsaXNoZXIy NTcxNzM1