Since the advent of advanced generative AI tools, both employees and employers have been contemplating whether AI will take, alter, or eliminate jobs. While there are instances where organisations have downsized their teams to incorporate AI assistance, a recent study suggests that AI is generally too costly to fully replace humans in most jobs.
The Massachusetts Institute of Technology conducted a study to address concerns about artificial intelligence replacing humans across various industries. The findings reveal that, at present, artificial intelligence cannot efficiently replace the majority of jobs in a cost-effective manner.
In one of the initial comprehensive examinations of the potential for AI to replace human labour, researchers focused on modelling the cost-effectiveness of automating different tasks in the US particularly jobs utilising computer vision, such as teachers and property appraisers.
The study revealed that only 23% of workers, measured in terms of dollar wages, could be efficiently replaced. In certain instances, the cost of installing and operating AI-assisted visual recognition made human involvement more economically viable.
The widespread adoption of AI across various industries gained momentum last year, particularly fuelled by the success of OpenAI's ChatGPT and other generative tools that showcased the technology's potential.
Tech giants including Microsoft Corp. and Alphabet Inc. in the U.S., as well as Baidu Inc. and Alibaba Group Holding Ltd. in China, introduced new AI services and intensified their development initiatives.
However, some industry leaders expressed concerns about the rapid pace of these advancements, cautioning that it may be too hasty. Apprehensions about the impact of AI on employment have consistently been a central concern in discussions surrounding the technology.
"‘Machines will steal our jobs’ is a sentiment commonly voiced during periods of swift technological advancement. This apprehension has resurfaced with the development of large language models," as stated by researchers from MIT’s Computer Science and Artificial Intelligence Laboratory in their 45-page paper titled "Beyond AI Exposure."
The paper notes, "We find that only 23% of worker compensation ‘exposed’ to AI computer vision would be cost-effective for firms to automate because of the large upfront costs of AI systems."
Computer vision, a subset of AI, empowers machines to extract meaningful information from digital images and other visual inputs. Its widespread applications are evident in object detection systems for autonomous driving and in assisting with the categorisation of photos on smartphones.
The cost-benefit ratio of computer vision is particularly favourable in sectors such as retail, transportation, and warehousing, where major players like Walmart Inc. and Amazon.com Inc. operate prominently. The feasibility of its application extends to the healthcare context, as noted in MIT’s paper.
The authors suggest that a more aggressive AI rollout, especially through AI-as-a-service subscription offerings, could enhance the scalability and viability of other applications. The study, funded by the MIT-IBM Watson AI Lab, utilised online surveys to gather data on approximately 1,000 visually-assisted tasks across 800 occupations. Presently, only 3% of such tasks can be cost-effectively automated, but this figure could potentially rise to 40% by 2030 if data costs decrease and accuracy improves, according to the researchers.
The sophistication exhibited by ChatGPT and its counterparts, such as Google’s Bard, has reignited concerns about AI displacing jobs. These advanced chatbots demonstrate proficiency in tasks that were traditionally exclusive to human capabilities.
The International Monetary Fund (IMF) recently stated that nearly 40% of jobs worldwide would be affected, emphasising the need for policymakers to carefully balance AI’s potential with its potential negative consequences. Discussions at the World Economic Forum in Davos last week prominently cantered around the impact of AI on the workforce.
Mustafa Suleyman, the co-founder of Inflection AI and Google’s DeepMind, expressed the view that AI systems are "fundamentally labour-replacing tools." The paper included a case study involving a hypothetical bakery.
Bakers, who visually inspect ingredients for quality control daily, spend only 6% of their time on this task, according to the researchers. Despite potential time and wage savings from implementing cameras and an AI system, the cost of such a technological upgrade still outweighs the benefits, they concluded.
Neil Thompson, the director of the FutureTech Research Project at the MIT Computer Science and Artificial Intelligence Lab, commented, “Our study examines the usage of computer vision across the economy, evaluating its applicability to each occupation across nearly every industry and sector. We show that there will be more automation in retail and health care, and less in areas like construction, mining, or real estate,” in an email statement.