Future of job prediction and underlying contradictions

Elon Musk's claim that artificial intelligence (AI) will render all jobs obsolete got high global media attention. The underlying reason has been that we all are apprehensive about the future of jobs and the unfolding potential of AI. At the dawn of the twenty-first century, we got a scary feeling about the future of jobs due to rising waves of technologies like AI, robots, and automation. A series of reports kept announcing massive job loss predictions. One of the notable such publications has been the World Economic Forum's (WEF's) Future of Jobs Report. However, many such predictions have suffered from gross contradictions-worsening the concern. The added problem of future jobs has been what to believe in. Hence, we need to determine the underlying cause of future job loss predictions and why they contradict the unfolding reality. To begin with, let's look into recent updates on the future of jobs prediction.
As reported in the media, a study claimed the likely loss of 53.8 lakh jobs in five key sectors of Bangladesh by 2041. According to the study findings, the ready-made garments sector would be the worst sufferer, with likely sixty per cent job losses. The furniture-making industry is in the second worst position, with the threat of losing 55 per cent job losses. Artificial intelligence, Robotics, and Automation unleashing the fourth industrial evolution was attributed to the likely massive job loss.
The highly cited Future of Jobs Report 2020 of the World Economic Forum made several claims about job loss prediction, primarily due to AI. The report's findings predicted the displacement of 85 million jobs due to a shift in the division of labour between humans and machines by 2025. The report also indicated that by 2025, humans and machines will equally share time, as opposed to 33 per cent by machine in 2020, on current tasks. However, the Future of Jobs 2023 report finds only a one per cent increase in the level of automation over three years. Hence, it does not sound rational to predict that by 2025, the role of machines will increase from 34 per cent in 2023 to 50 per cent. Therefore, the current report finds that the actual pace of automation contradicts expectations from the 2020 survey. To remain in sync with the unfolding reality, respondents in the 2023 survey have lowered their expectations. They now predict that by 2027, only 42 per cent of business tasks will be automated. Such findings raise a few vital questions. Does it mean that the survey opinion-based approach fails to predict reality? Does it mean that intuitive extrapolation of experience is failing to keep pace?
The next issue has been job loss prediction due to AI. WEF's Future of Jobs 2020 report predicted that by 2025, AI would kill 85 million jobs. However, the reality is far from the prediction. Hence, the 2023 report changed the perspective of AI implications from job loss to creation and augmentation. It has been stated that the reality is not that machines will simply replace humans. Instead of fully taking over jobs from humans, AI will augment humans to be more productive. The expectation of taking over physical and manual work has decreased. Instead, AI is predicted to take over primarily algorithmic roles, requiring codified knowledge and skill. Contrary to the common belief of taking over mundane manual jobs needed in apparel or furniture making, a new report predicts that AI will find it easier to automate white-collar roles like reasoning, communicating, and coordinating.
The question is why there is a shift in making predictions. The answer could be simple: to reflect the reality more accurately. The change of focus from killing to augmenting also raises a question: will AI not kill jobs at all? Or, will we change the expectation once AI kills jobs? Perhaps such prediction exercises create far more questions than they answer. Consequently, this creates a decision-making dilemma. Hence, we need to know beyond what opinion survey finds. Due to the high importance of predicting the future of jobs, we need to help people get insights at a deeper level than what appears on the surface to make better choices.
To dig down further substance, we need to go beyond the opinion of experts surveyed, summarised, and reported by WEF. We cannot just keep adjusting expectations to close down the gap. We need to figure out why experts are facing difficulty in predicting the future based on intuitively extrapolating their knowledge and experience about the past and present.
Of course, AI will create jobs. Some of the jobs that AI will create are R&D jobs. High-end professionals will be analysing data, developing training data sets and training machine learning models, and developing algorithms for machines to take over or augment human roles. Besides, information security jobs will be created due to the growth of AI machine capability and connectivity.
Technology affecting economic and environmental trends will be the most significant cause of creating, refining and destroying jobs. Technologies like AI, Robotics, Software, and Automation will be driving creative waves of destruction. However, will technologies keep growing linearly? Besides, will technologies face varying difficulties in taking over human roles in work? Furthermore, what capabilities do humans rely on to play meaningful roles in work?
First, technology possibilities show up with a faint signal of potential for driving creative waves of destruction. In the beginning, they are caught in high-level noise. Besides, upon showing initial performance, sometimes, they stop growing. Often, they suffer from premature saturation before crossing the thresholds. It happens that AI notoriously suffers from these symptoms. Hence, many early demonstrations of AI performance, like autonomous vehicles, have not been scaled up to reach critical adoption levels. Their growth paths have been suffering from chasm. Hence, extrapolation of expert opinions developed through observation of early demonstration have failed to parallel the unfolding reality. Therefore, the survey-based opinion-gathering and compilation approach has suffered gross contradictions.
The next issue is about human roles in work and the capabilities they apply. Humans are eligible for work for their education and training earned codified and experience produced tacit abilities. In any occupation, they use all of them. However, the relative percentage varies. Besides, complexity of automating them also varies. For example, due to the software-centric AI technology core, AI finds it relatively easy to automate codified knowledge and skills. Codifying experience-earned tacit capability has also become easy for AI to take over. However, it has been challenging to take over in-born innate sensory, physical, cognitive, and psychomotor abilities. Such reality creates an illusion of AI's progress in fully taking over jobs.
For this reason, despite the $80 billion R&D investment, robot cars are caught in the chasm. Due to similar reasons, the reality has surfaced that it is far easier to take over knowledge-centric work, leaving physical and manual ones to humans. Therefore, it's time to question the reality of Elon Musk's observation about the future of jobs.
As explained, expert opinion-based survey results are highly deviating from reality. The underlying reason has been the technology life cycle, pervasive uncertainties, and varying complexities of automating different human abilities. Hence, to overcome contradictions about the future of job predictions, instead of surveying expert opinions, we need to focus on underlying technology life cycles, pervasive uncertainties, risks of getting caught in a chasm, and complexities of automating human abilities in target tasks.

M. Rokonuzzaman, Ph.D is academic and researcher on technology, innovation, and policy. [email protected]