The global generative AI market is projected to experience explosive growth in the coming years, with a 45 per cent compound annual growth rate expected from 2021 through 2028. As the commoditisation of AI services becomes more widespread, business models of industries, from software development to entertainment, will change drastically.
LLMs (Large Language Models) and Generative AI are set to automate various tasks that require natural language understanding – for instance, summarisation, translation, answering questions, coding, and even conversation. AI-powered coding assistants like GitHub Copilot, Amazon CodeWhisperer, ChatGPT and Tabnine are rapidly gaining popularity and helping developers undertake routine tasks, freeing them to focus on more complex issues.
According to a report by GitHub, a vast majority of developers in the U.S. have embraced AI coding tools, integrating them into their workflows both professionally and personally. 92 per cent of programmers based in the U.S. are now leveraging AI to supplement their coding abilities.
As these technologies upset the mode of producing and consuming media products and information, there will be significant economic challenges such as the disruption of markets, creation of inequalities, reduction of incentives for human creativity and innovation, and the displacement of workers. A large majority of developers believe that AI coding tools will give them an edge in the workplace, according to a GitHub report. They expect several key benefits from using AI coding assistants, including more accurate, efficient and faster coding.
Tasks that involve routine information processing, data entry and filling out forms in sectors such as customer service, research, even blue-collar jobs and legal segments, may be affected. Even with partial automation, almost 5 to 10 per cent of roles in the sectors may cease to exist in the near future. This will create hundreds of millions of unemployed skilled and semi-skilled workers. It will also impact developed and developing nations differently.
Nations and societies that do not rapidly reskill their workforce will be disproportionately affected. There is also no guarantee that generative AI and related technologies will create new jobs to make up for the lost ones.
The impact of LLMs and Generative AI will be felt across all industries. In India, sectors that involve routine information processing will be affected, including customer service, research, blue-collar jobs, and legal vocations.
India has been a leader in IT services primarily due to the availability of cheap coders domestically. The advantage of having an English-speaking population will soon wither away. Advance planning through workforce training programmes, new policies, and social support measures that help people through this transition are the need of the hour.
Funding and incentivisation of the transition of workers by helping them gain new technical skills should be a high priority. Policy and legal measures that help workers with the transition, severance payments, advance notice of automation and restrictions on discriminatory AI systems are necessary. The state can also think of providing tax breaks and other incentives to help businesses retrain workers. Enhancement in social safety nets may require changes in pensions, insurance, and employment rules. These new social safety nets should recalibrate unemployment benefits, perhaps even think of unemployment insurance, create income supplement opportunities (maybe as a temporary measure), and create job placement services to help displaced workers apply for new roles.
India is not as well-prepared as China and the U.S. to face the onslaught of Generative AI and related technologies. The country doesn’t have any major investments in AI chip hardware design. The absence of audited data sets for training and fine-tuning models is a major shortcoming. India also doesn’t have its own foundational or generative model like GPT 3 or Wu Dao. Compared to China and the U.S., India has significantly fewer experts with PhDs in fields related to AI.
There are limitations in access to cloud computing in India for training large language models, and it is expensive. India does not have large corporations that invest heavily in in-house AI research. The number of AI policy think tanks and research institutes is also much higher in the U.S. and China. Any good quality talent in these fields in India will quickly migrate to these destinations. There is a serious lack of coordination between academia and industry in India. As the race for large language models heats up, data security and privacy concerns in India will reduce our chances of getting valuable data that can train robust models. In fact, India seems to lack a comprehensive and composite AI strategy that connects government, industry, academia, and society.
If India has to proactively utilise the gains from these disruptive technologies and adapt to the changes in the economy, society, and everyday life, it should do the following: one, develop a comprehensive national AI strategy that connects stakeholders to provide a roadmap for responsible AI deployment. Increase funding for AI research. Two, establish AI policy think tanks and research institutes to foster AI innovation and nurture talent. Three, incentivise businesses to invest in AI R&D and support workers’ training for the changing technological landscape. Four, implement policies to protect workers from job displacement and enhance social safety nets. Five, foster collaborations between academia, industry, and internationally to develop responsible real-life AI applications and stay updated on best practices.
Brijesh Singh is Adjunct Distinguished Fellow, Cybersecurity Studies, and a senior IPS officer.
This article was first published in the Indian Express.