The conversation around artificial intelligence (AI) has rapidly expanded from research labs and tech companies to everyday discussions in mainstream media, coffee shops, and street corners. AI is a hot topic, with concerns ranging from its potential to revolutionize industries to fears of its impact on jobs and society. Recently, a friend shared a story about overhearing a group of men discussing the threats posed by AI, with one of them expressing his frustration by saying, “I want to punch AI in the face!”
While AI doesn’t have a face to punch, this reaction highlights the anxiety many people feel about the rise of machine intelligence. The potential dangers of AI are indeed concerning. It can be misused to reinforce biases, destabilize political systems, and deepen social inequities. Some experts even warn that AI could one day challenge our dominance as a species. But beneath these fears lies a more immediate concern: the potential economic impact of AI, particularly on jobs.
One key aspect of this fear is the worry that AI might take over our jobs. However, as futurist Roy Amara observed, society often overestimates the short-term impact of technology while underestimating its long-term effects—a concept known as Amara’s law. This perspective is particularly relevant when considering AI’s influence on the labor market. While it is challenging to predict the long-term impact of AI on job loss, we do know that AI does not eliminate jobs in a straightforward way. Instead, it automates tasks—specific tasks within a job, rather than entire roles. As a result, AI will likely transform jobs rather than eliminate them entirely.
Take the software development industry, for example. Automated coding tools are becoming increasingly common, and Matt Welsh, a startup founder and computer scientist, likened their arrival to an alien spaceship suddenly landing in our backyard. Yet, rather than replacing programmers with AI, Welsh encouraged his team to use these tools because they increased productivity by 30–40%. My MIT colleague Armando Solar-Lezama suggested that AI coding tools will shift the focus toward high-level awareness and structure of the codebase. While AI may plant certain trees, we still need skilled people to tend to the forest.
The question then arises: who benefits from these productivity gains? Will workers who become more productive earn higher wages, or will companies reduce their workforce and pocket the savings? Ideally, increased productivity would lead to higher wages for workers, recognizing their enhanced skills and output. However, there’s also the risk that efficiency gains will lead to workforce reductions, with the benefits primarily accruing to corporations.
While much of the focus on AI’s impact has been on large companies, small- and medium-sized enterprises (SMEs) will also be affected. Larger businesses may have the resources to explore and invest in AI, potentially widening the competitive gap. However, there are efforts to help smaller companies adapt. For example, Global Partnerships in AI has introduced a portal to educate business owners about AI tools and services available in their industries.
To illustrate the division of tasks between humans and AI, consider a small interior design firm. Design work, rooted in creativity, empathy, and subjective judgment, remains a distinctly human endeavor. However, AI can assist by automating routine tasks, such as creating mood boards, optimizing layouts, or sourcing materials, allowing designers to focus on higher-level creative and strategic work. While AI might generate design renditions, the results will be the average of existing knowledge, not something new and fresh. Thus, tasks that require creativity, client interaction, and ethical decision-making are likely to remain in the human domain.
This example highlights that jobs are composed of a variety of tasks, some of which are suitable for AI automation and others that are not. Erik Brynjolfsson and Tom Mitchell suggest that jobs should be analyzed based on the tasks they involve, distinguishing those that can be automated from those that require human intelligence. Tasks that involve large datasets and well-defined goals are ideal for machine learning, while those requiring common sense, physical dexterity, or subjective judgment are less so. For instance, skilled trades like plumbing and electrical work are unlikely to be automated soon.
In summary, while AI will undoubtedly change the nature of work, it is not a simple matter of replacing jobs. Instead, it will reshape how tasks are performed, requiring us to adapt and find new ways to leverage human creativity and ingenuity alongside machine intelligence.
By Impact Lab
