Redesigning work to work with Artificial Intelligence

Artificial Intelligence (AI) is increasingly permeating all aspects of our daily lives and transforming how we interact with the world. AI powers the voice assistants that read out our daily schedules, the recommendation algorithms that suggest our morning news, and the traffic prediction systems that guide our commute. At work, AI aids in data analysis, automates repetitive tasks, and even assists in complex decision-making processes. In healthcare, AI algorithms are revolutionising diagnostics and treatment plans, while in education, personalised and adaptive learning is being facilitated through AI.

However, amidst the rise of AI, one question has become increasingly pertinent: Will AI serve as a complement or a substitute for human effort? Throughout history, the emergence of new technologies has often been met with scepticism and resistance. The 19th century saw the advent of the steam engine, which sparked fears and protests among the rioting Luddites. Fast forward to the 20th century, the introduction of computers in the Indian banking sector too triggered similar fears of job losses and was met with significant opposition from workers and their unions. However, the adoption of computers in the mid-80s laid the foundation for core banking and technological advancements such as electronic fund transfer, sparking a wave of innovation in financial services and ultimately, greater employment opportunities. In the 21st century, we find ourselves grappling with similar dilemmas concerning AI and its implications for human capital development.

The COVID-19 pandemic presented a unique and valuable opportunity to assess the effect of AI adoption or exposure in firms on employee outcomes. Diverse firms leveraged their AI investments during this time to maintain business continuity and build economic resilience to the pandemic. Adoption was driven by the need to substitute unavailable labour, reorganise work, re-engineer processes, and introduce new products and services suited to a rapidly changing business environment due to the lockdown.  

In our paper, ‘Amenability to AI and Employee Outcomes: Evidence from the Pandemic,’ we delved into the impact of exposure to AI on a range of employee outcomes during the pandemic. We examined hiring trends, salary shifts, non-pecuniary benefits, and relational contracts across firms. In the process, we aimed to answer an important question: Does AI impact the demand for and value of labour, and if so, how?

AI Complements Human Capital

The COVID lockdown period provided a unique opportunity to examine the causal relationship between AI exposure and employee outcomes. Drawing from anonymised reviews on the India-based employee review platform AmbitionBox, we analysed data from over 600,000 employees across 2,971 firms during the lockdown (January 1, 2020 – February 28, 2021). This data was further refined down to 119,060 employee reviews spanning 73 occupations, as per the National Classification of Occupations (NCO, 2015).

To measure the exposure of AI in various professions, we surveyed 3,100 Indian workers using the Suitability for Machine Learning index, as defined by Eric Brynjolfsson and Tom Mitchell in their 2017 paper. This survey was part of an earlier study (The impact of AI on Indian markets, 2020) conducted by the Indian School of Business (ISB) in collaboration with Intel Corp, which aimed to assess the impact of AI on employees, occupations, firms, and the overall economy.

Our analysis revealed that AI adoption during the lockdown had a significant positive impact on firms' recruitment practices, salaries, and non-pecuniary benefits offered. We observed improved hiring outcomes (greater hiring continuity, lower hiring freezes, and job offer rescinds), less adverse impact on wages (more timely salary incidences, lower unpaid salary, or salary cuts), and a rise in non-monetary benefits (such as work-from-home arrangements and employee care programs) These findings demonstrate that firms that leveraged AI during the pandemic did so in a manner that complemented existing human labour as opposed to substituting it. Indeed, these firms maintained continual hiring and remuneration timeliness while notably reducing salary pauses and cuts.

Anecdotal evidence from diverse industries supports our findings. Retailers used AI-based conversational and messaging bots to front-end and contain customer conversations, and in turn, augment the capacity of existing sales representatives to significantly scale demand and reduce costs.  Similarly, innovative technologies like drones equipped with IoT and GPS/GIS-enabled satellite imagery were employed for policing and delivering critical medical supplies to people confined to their homes. The increased use of geospatial technology also enabled employees to reconfigure workflows, track consignments in real-time and set geo-boundaries for 'No-go areas' during the lockdown.

Not surprisingly, we find that firms with greater AI exposure had significantly greater economic immunity during the pandemic. Using official data on real economic activity, that is, daily volume and value of payments at the state-sector level from the payments gateway, Unified Payments Interface (UPI), managed by the National Payments Corporation of India (NPCI), we find that while on average, the lockdown led to an average decline of 25% in the number of payments received, a 5% increase from the mean AI exposure of sectors stemmed this decline by 9%.

AI Use Improves Employee Satisfaction

While AI investments complemented human capital to improve the economic performance of using sectors during the pandemic, they also correlate with significant restructuring and reengineering of work, requiring employees to learn new skills, adapt to new work routines and processes, and respond to new operating models and cultures that are centred on data, technology, modular work, and employee autonomy. Therefore, they may be potentially disruptive to the socio-emotional outcomes of employees such as job security and employee satisfaction with work and the organisation. Therefore, we additionally assessed employee review data to discern perceptions of job security, organisational culture, and overall satisfaction with work and the organisation. Our results indicate a significant positive impact across all these outcomes, indicating that AI-enabled changes at work affected during the lockdown sustained beyond the pandemic and strengthened relationships of employees with their firms. Specifically, employees in occupations more amenable to AI expressed a more favourable overall evaluation of their employers, perceived a higher level of job security, experienced greater job satisfaction, and held a more positive view of their company culture.

Using novel, high-frequency data, our study presents essential evidence contributing to the existing literature on AI's impact on human labour. Adopting Artificial Intelligence does not necessarily imply workforce displacement or reduction in a given occupation. Instead, our findings suggest that AI can play a pivotal role in enhancing hiring, job satisfaction, and overall resilience in the face of economic challenges.

Accommodate to thrive!  

Like other ground-breaking General-Purpose Technologies, such as steam-powered turbines, electricity, computers, or information and communications technology, AI holds the potential to unlock new employment opportunities. If thousands of years of human history serve as an indicator, we have consistently adapted to change and remained at the forefront of innovation.

Emergent management research emphasises that Machine Learning will lead to the re-engineering of business processes and the reorganisation of tasks. Unlike previous waves of automation, whose impact was limited to simple and routine work, Machine Learning will impact diverse occupations and workforce segments, including many professional jobs.  Therefore, there is a need for a shift in the debate surrounding AI's impact on work, from full automation of jobs to a more nuanced approach that focuses on redesigning jobs and workflows.

Our study's results reinforce this perspective, suggesting that AI's most significant influence will be in redesigning work and business models that strengthen employees' relational contracts with firms. This insight paves the way for firms and governments to rethink their long-term strategies and prioritise human centricity in policy initiatives centred around the proper adoption of AI. By embracing AI's potential, we can unlock new opportunities and ensure a more adaptive and resilient future for the workforce.

In conclusion 

The universalisation of AI signifies a new era of technological advancement. As AI becomes pervasive in its reach and impact, it holds the potential to fundamentally redefine work and occupations. The future will require a workforce that can collaborate with AI, leveraging its capabilities to drive innovation and efficiency. As AI continues to evolve, individuals and organisations need to adapt and upskill.

Embracing this change while navigating its challenges will be vital to thriving in this AI-driven world. Our findings prescribe essential implications for business managers and policymakers. To the extent that AI-enabled reorganisation of work and new business models are sustained in the future, our results emphasise the need to ease resource constraints that facilitate AI adoption, investments in worker training, and management practices that complement the technology to improve firm performance.

This article is based on the research by Madan Pillutla, Professor, Organisational Behaviour, Deepa Mani, Professor, Information Systems, Nikhil Madan, Assistant Professor, Organisational Behaviour, Shekhar Tomar, Assistant Professor, Economics from the Indian School of Business and Abhishek Bhatia, doctoral student at London Business School.