Yesterday, the U.S. Census Bureau released its 2025 poverty, income, and health insurance data. I am becoming increasingly frustrated that in the years since the Biden Racial Equity Executive Order recommended the disaggregation of data, data is still reported at the banal race or sex disaggregation. The poverty data provided poverty rates by sex: the 2025 poverty rates for men and women both showed statistically significant decreases of 0.4 and 0.6 percentage points, respectively. Hispanics were the only race/ethnic group with a statistically significant decrease in poverty rates, at 1.2 percentage points. As the diversity of the U.S. continues to grow, the Census must report poverty rates at the intersection of race and sex.
In the Supplemental Poverty Measure (SPM), poverty among people without a high school diploma rose 2.0 percentage points, to 32.4 percent, a statistically significant increase. The official poverty measure increased by 0.6 percentage points, but that change was not statistically significant. So the SPM, which accounts for the cost of living, safety-net benefits, and tax payments and credits, increased for those without a high school diploma. Translation: even after counting food assistance, tax credits, and other safety-net benefits, more Americans without a high school diploma fell below the SPM poverty threshold this year than last. Their SPM poverty rate is 8.6 points above the official rate, a gap wider than for any other group in the data (see Figure 1). Whatever the safety net is doing for those with a high school diploma or greater, it is doing less for those without one.
Figure 1. Poverty Rate, No High School Diploma, Age 25+
Source: U.S. Census Bureau, Income, Poverty and Health Insurance Coverage in the United States: 2025.
Tamara Jeffries shared this Instagram post with me, and I was dismissive. She and I have been talking about AI since I started using Claude.ai and Perplexity.ai. But the SPM poverty rate increase for those without a high school diploma has me thinking: are the conversations about rogue cluster agents hijacking the internet or infrastructure, or AI killing all of mankind, a deflection from the harm that is already being caused? We can see reports about potential job losses, but the increase in the number of Americans in poverty is real, and so is job automation. The National Equity Atlas found that jobs requiring no more than a high school diploma carry an average automation risk of 80 percent; accommodation/food services, administrative and waste management, real estate, retail trade, and transportation/warehousing are industries with over 60 percent of jobs automatable, affecting roughly 31 million jobs in this cluster of industries.
Next year’s SPM numbers will tell us whether 2025 was a blip or the start of a trend. Given where automation risk is concentrated, I’m not betting on a blip
Stay sharp and mind the deflections.
Rhonda V. Sharpe is the president and founder of the Women’s Institute for Science, Equity and Race. Her research focuses on gender and racial inequality, STEM diversity, and the demographics of higher education.