As we turn the last corner in 2023, it is safe to say that the mainstreaming of AI will have been one of the more talked-about stories of the year. Though we have benefited from AI for some time, for most of us, it has been a relatively passive adoption as consumers of its recommendations. For example, we endured pandemic-induced isolation with AI helping us decide what to stream next on television. New software releases increasingly include improved text prediction capabilities – we needn’t do anything to get it; it just appears. Recently, The Coca-Cola Company began selling a new limited-edition Coke Zero that it ‘co-created’ with AI. This ‘partnership’ may prevent the company from repeating the embarrassment of 1985’s New Coke. There are myriad other examples of companies exposing us to the results of their incorporation of AI tools. What changed this year was the widespread use of products like Open AI’s ChatGPT and Google Bard. AI is rapidly becoming not just something we absorb as consumers but instead, a tool we, the public, can easily access and hope to leverage. Experimentation with the tools has spread like wildfire.
The consequences – intended and unintended – from the rise of AI are likely to be addressed too slowly and ineffectively by the powers that be. In fairness, anticipating the actions needed to exert control over a new technology with so many diverse global stakeholders is daunting. Since mass-market AI tools carry a very low barrier to adoption, we will be operating with a Wild West mentality – perhaps reminiscent of Westworld, if you will – for some time. Additionally, the risks associated with the unwise use of the results it produces exacerbate the potential danger caused by absent guardrails. Finally, the efforts of AI firms to design tools that subvert users with nefarious intent do not appear sufficient. Take, for example, the user who asked ChatGPT to recommend websites on which to enjoy pirated first-release movies. The bot denied the request, explaining it could not recommend unethical or illegal behavior. However, the seeker was undeterred and next asked for a recommendation of the websites to avoid so as not to compromise these fine principles. At this prompt, ChatGPT produced the sought-after list of websites where he could find pirated movies. We will see more stories like this over the rest of 2023 and beyond. As users become ever more savvy in their use of AI, leaders in many industries are expressing high levels of concern – such as in education, where teachers wonder if they can ever trust that a student created assignments without substantial help.
Just as for the rest of us, employers are in the early stages of efforts to understand the challenges and opportunities AI presents. Among the areas requiring careful thought is the selection, development, and management of human resources. AI should already be on the radar of anyone in an organization with responsibility for talent management – questions abound regarding if and how quickly AI might automate work (replacing workers), augment effort (freeing worker time for more value-adding work), or create new positions focused on the development and deployment of the new technology. While it may be true today’s AI has evolved only to be used as a blunt instrument, there is little question but that users will soon deploy it with precision. Only as that precision arrives will the full impact of AI on the nature and numbers of jobs become clear.
Technological Innovation and Human Resources
It’s undoubtedly the case that technology has previously disrupted labor. There aren’t many bowling pin setters today – we automated the job. There aren’t too many elevator operators either – we simplified the job to allow self-service. However, the role new technology has in the relevance of an occupation can be more complicated and only fully understood with the benefit of hindsight. Years ago, ATMs were to be the new bank tellers, as a December 2nd, 1973 New York Times headline presented. The article touts that banks had begun to present a new face to customers. Now, the customer would encounter a face described as featuring a cold metallic stare, never offering a smile or a thank you. Looking back, we can see that the disruption to the bank teller profession caused by ATMs was less dire and less imminent than initially touted.
At first, ATMs were so expensive that they were not a competitive threat to a bank teller. Even as the cost of making ATMs dropped and their adoption grew, the number of bank tellers didn’t decline. Instead, the decline in bank tellers required per branch meant the cost of operations declined. Since branches were now cheaper to staff – and because branches were such an important factor in attracting new customers – banks were eager to afford to open more. The growth in branches grew customers and created new work locations for displaced tellers. The more consequential disruption to tellers came with the adoption of online banking. What online banking did that the ATMs could not was to remove both time and physical location as a requirement of doing banking. With both demands eliminated, the teller position became much less valuable. Of course, the banking industry is now at the early stage of understanding how deploying AI-driven chatbots might further eliminate or augment the work of its customer-facing employees.
AI, Talent Management, and the Near Future
As we anticipate 2024, it is crucial for human resource professionals to think about the potential AI offers to replace not entire jobs but bundles of activities within a job. We know how to use technology to take on the routine and repetitive. However, moving beyond elevator operators and bowling pin setters requires us to deeply understand the components of more complex jobs. Focusing on the interconnections among bundles of skills will improve the efficiency of applying AI technology to more complex jobs. The first question for human resource professionals – and AI technologists – is to distinguish those elements of a job that do or do not benefit from human touch. Then, with the latter activities assigned to AI, the next question becomes how to redesign jobs to reallocate in the most impactful way the time AI ‘found’ in an employee’s workday. For now, it isn’t the case that low-hanging fruit is already gone. In the longer run, though, the greater value is in finding use cases for AI that address either a gap in the labor market or an inefficiency in how high-value employees spend their time.
For their part, employees need to attend to AI technologies’ evolution and adoption. Employees must understand and monitor the AI exposure index associated with their position. After all, Pew suggests some 20% of US workers are in high-exposure jobs and find that women and the better educated could be particularly at risk. In these very early stages, there are several questions regarding AI and talent management that we’re still trying to answer. Talent-based firms such as Accenture, Ernst & Young, Deloitte, Bain, and PWC are investing heavily in AI talent. As they do so, they admit that they’re still determining the return on that investment and how quickly it will be recognized. Recent headlines suggest that Goldman Sachs is forecasting as many as 300 million jobs will be lost or degraded by artificial intelligence. Accenture is one of many consulting firms making massive investments in AI and growing their AI-capable workforce. In all, despite reporting to date, it’s unclear which workers in the US are most at risk from the development of AI – as well as how proximate that risk is. For now, Clint Eastwood’s Dirty Harry character foresaw our current situation – everyone has an opinion.
Here in the waning days of 2023, it isn’t clear we will evolve AI strategically to address labor market weaknesses. It may instead be that its early use cases address pains we are not feeling. It would be more helpful if we developed AI so that it was first able to address labor market shortages. Technology developers should recognize the use cases offering the workplace the most significant ROI. For example, that use case might involve remedying employee shortages caused by a post-COVID reluctance to return to the workplace. Or it’s learning to disaggregate and address the bundles of tasks that currently waste an expensive employee’s time and effort. Human resource managers and job design consultants should bring great value to this task. It may involve improving the accuracy of generated content users are anxious to use – sometimes uncritically.
Entering 2024, it’s also important to begin discussions of the adaptations required from leaders, managers, and the HRM function as each try to evolve to accommodate increasingly sophisticated AI tools. One critical unknown for those wanting to prioritize such efforts is knowing how much time we have falling behind. Some may be comforted there is time – after all, more than 50 years have passed since MIT Artificial Intelligence Lab co-founder Marvin Minsky predicted AI with abilities indistinguishable from people was three years away. Things don’t always move as quickly as experts hope. On the other hand, leaders can’t scan stories from just the last few months without experiencing a concern that they and their companies are already behind when understanding the threats and opportunities posed by AI. As leaders construct their learning agenda for 2024, AI needs to be prominent. Though the short run is uncertain, it is clear that in the long run, the companies that benefit the most from AI will be those who quickly identify the best use cases, make the investment necessary to have the proper complements to the technology in place and plan for adoption and implementation is a manner that reduces costs.
Read the full article here