MIT professor pours cold water on the prevailing hype about AI and the economy

A study by MIT economist Daron Acemoglu appears to challenge the prevailing optimism surrounding artificial intelligence (AI) and its economic impactA study by MIT economist Daron Acemoglu appears to challenge the prevailing optimism surrounding artificial intelligence (AI) and its economic impact. While many experts predict a future fuelled by AI-driven productivity booms and reduced inequality, Acemoglu’s research paints a more cautious picture. His findings suggest that AI’s impact on productivity and inequality may be far less dramatic than anticipated, and could even exacerbate the gap between the rich and the poor.

This research stands in stark contrast to the wave of bullish forecasts about AI’s potential. Just last year, financial giants like Goldman Sachs projected a 7 percent increase in global GDP over the next decade due to AI. Acemoglu, however, views these predictions with scepticism. He argues that past waves of automation have primarily benefited those who own capital, such as business owners and managers, while leaving workers facing job displacement and stagnant wages.

Drawing parallels with past automation efforts, Acemoglu suggests that AI’s ability to significantly boost productivity may be limited. While AI excels at automating routine tasks, many future jobs will likely involve complex problem-solving and human interaction – areas where AI remains in its infancy. This suggests a more modest impact on productivity growth, translating to a potential GDP increase of only 0.93 percent to 1.16 percent over the next ten years.

Acemoglu further tempers expectations by highlighting the potential drawbacks of some common AI applications, such as deepfakes. Combating such negative uses of AI will require resources that could be better directed towards more productive endeavours, potentially offsetting some of the purported economic gains, according to Acemoglu.

The study also raises concerns about AI’s impact on inequality. While some argue that AI will free up workers for higher-skilled jobs, Acemoglu suggests this may not be the case. In fact, he posits that AI could disproportionately harm certain already disadvantaged groups. Furthermore, as AI creates new opportunities for those who own and develop the technology, the gap between the rich and the poor is likely to widen, he suggests.

Despite this scepticism, Acemoglu acknowledges the potential of AI to revolutionise various sectors. However, he emphasises that this potential hinges on the responsible development and deployment of the technology. AI’s true value lies not in creating sophisticated chatbots or novelty applications, he suggests, but in providing people with access to reliable and accurate information that can enhance productivity across various professions.

The current focus on developing general-purpose AI models capable of human-like conversation, argues Acemoglu, may not be the most effective approach. Instead, prioritizing the development of AI tools specifically designed to provide reliable information to educators, healthcare professionals, and skilled workers like electricians and plumbers, could yield far greater economic benefits.