Rethinking AI’s role in income inequality

The rapid rise of generative AI has prompted a familiar set of fears—particularly rising inequality as some professions become more valuable and others get automated away. But those concerns could be overstated. A year-long research and modelling effort by PwC economists, experts and outside academics found that under certain conditions, AI could actually reduce income inequality.

The analysis looked at the potential impact of AI across three potential scenarios between now and 2035. In the most optimistic scenario, widespread AI adoption boosts productivity and revenue growth at companies, along with employee wages (even for the industries and jobs most automatable through AI. Our 2025 Global AI Jobs Barometer indicated these types of wage increases are already occurring). Some jobs will go away—as with virtually all technological advances—but the overall effect is a net positive. As a result, productivity growth leads to rising GDP growth, real wage increases for most workers and expanded societal prosperity overall.

To quantify this impact, we looked at a metric called the Gini index, which gauges wealth and income inequality among a country’s population on a scale of 0 to 100. Higher Gini index scores indicate greater inequality. In the US, our analysis (again, assuming the most optimistic scenario) shows the Gini index dropping from 37.5% in 2023 to 36.7% by 2035. 

An improvement of 0.8% may seem small, but it’s significant given the generally slow rate of change for overall income and wealth levels. Between 1980 and today—a period of steeply rising inequality in the US—the Gini index only increased by about 0.8%. Other countries, including Germany, India, Japan and the UK, also showed a potential reduction in inequality by 2035, but to a lesser degree. 

China showed a more sizeable impact, potentially due to its high volume of medium-skilled workers whose jobs would be complemented by AI—and the ability of AI to modestly boost opportunity for the country’s large base of low-income workers (e.g., agricultural technology for farmers or automated factories that create jobs for rural transplants).     

These aren’t guaranteed outcomes. The analysis also looks at less rosy scenarios in which AI productivity gains accrue unevenly across nations or AI adoption is slowed globally by widespread mistrust and lack of governance. In those cases, income and wealth inequality could worsen. 

In that way, the world is at a crossroads. The future impact of AI depends heavily on the business strategies, policy choices and workforce investments we make today. To that end, business leaders and policymakers should start by taking several steps to build trust. 

• Technology companies need to prove that AI is secure, be transparent about its abilities and limitations, make the underlying reasoning of AI models as understandable as possible and show that their offerings have measurable value.
• Company leaders need to put sufficient guardrails around the use of AI, including the data it draws on and the use of AI-generated insights and recommendations. 
• Both the public and private sector need to invest in initiatives for digital and technical upskilling to ensure that the productivity benefits of AI are spread as widely as possible across the workforce. 
• Governments need to align national and global standards as much as possible, support access to AI infrastructure and data across regions, and incentivise companies to adopt responsible AI practices.