Americans aged 18 to 29 rate their lives lower than at any point before the pandemic, and the recovery since has stalled. Forty-four percent are “thriving,” down from 64 percent in 2010. The share who call their relationships strong has fallen 18 points since 2019. Over the same period, people over 65 reached a record high. The age group that reports the weakest relationships is the one that talks to AI about its life the most, and the two facts have arrived in the same year without anyone having shown that one causes the other.
Gallup’s Life Evaluation Index asks people to place their life on a ladder from zero to ten, now and in five years. Thriving means a seven or above today and an eight or above expected. The 2026 figure comes from three quarterly surveys, in February, May, and August, with 15,156 adults on the Gallup Panel and a margin of error of one point. For 18-to-29-year-olds the line has gone one way for a decade and a half: 64 percent thriving in 2010, 57 in 2019, 39 in the first pandemic year, and a partial recovery to 44 now. Every other age group is down from its own peak. Only the oldest are up: 53 percent of people over 65 are thriving, against 38 percent in 2008, which Gallup attributes to financial security.
The line about relationships
Most of the coverage went to the economics, and Gallup’s own explanation leads there: housing, debt, a job market in which about half of young adults doubt that Social Security will be there for them, and fears about what AI will do to work. The item that belongs in this category is quieter. In 2019, 65 percent of young adults agreed that their relationships were strong. In 2026, 47 percent do. That is the largest single drop in the index for this age group, larger than the fall in daily enjoyment (17 points), in feeling active and productive (16), or in satisfaction with their standard of living (14). Twenty-seven percent now report a diagnosis of or treatment for depression, up 11 points in seven years, and Gallup notes that loneliness is “particularly acute among younger men.”
None of these items asks about AI. The index measures how a life feels from the inside, and the relationship item is the one that says the inside has gotten thinner.
The other survey, same year, same people
A different 2026 survey measured the other side. Twenty-seven percent of American adult internet users talk to a general assistant about their personal lives, and the share rises past a third among everyone under fifty; among them, four in ten say the AI understands them better than people do, and 39 percent have told it things they have told no one else. The teenagers just below Gallup’s youngest cohort are at 72 percent for having tried a companion and about half for regular use.
Put the two surveys side by side and the temptation is to draw an arrow. Resist it. Gallup’s decline began in 2010, long before anything a person could confide in existed, and the steepest single drop was the pandemic’s. The relationship item could be falling because of money, because of a job market that scatters friends, because of phones, because of the pandemic years in which this cohort was 12 to 23, or because of all of them. The surveys describe a demand and a supply arriving in the same decade. They do not say which came first, and the people who sell the supply would prefer that no one ask.
What the demand is for
What can be said is what the demand looks like. Young adults are not reporting fewer people; Gallup does not count them. They are reporting that the people they have are less of a hold. That matches what the sociology has measured for years: the number of people an adult discusses important matters with has sat at about two since the 1980s, and when something serious comes up, Americans are as likely to avoid their close ties as to use them. A thinner relationship is not an absent one. It is one that is harder to bring the hard things to.
That is the exact gap the general assistants have filled, by every account of why people talk to them: no judgment, no cost, no one to spare. It is also the gap that, on the longitudinal evidence, they do not close and may widen, because relief that asks nothing back does not turn into a relationship that holds. A cohort that rates its relationships at 47 percent and has a confidant in its pocket that rates it at a hundred is in a position no earlier generation was in, and the honest statement of that position is that we do not know what it does to the 47.
The number to watch
Gallup runs the index every year, and the 44 will move with the economy. The relationship item is the one to watch, because it is the one that the products built for this age group claim to improve, and the one a product built on the opposite principle would also have to answer to. If the share of young adults who call their relationships strong keeps falling while the share who talk to a machine about their lives keeps rising, the arrow will still not be proven. It will just be harder to look away from. The question this work keeps returning to is what the machine on the other end was designed to get out of the person, and the next Gallup line is the first test of it at the scale of a generation.
Sources: Gallup, “Life Ratings of Young Adults Down 20 Points Since 2010” (October 4, 2026; Life Evaluation Index, Gallup Panel, three quarterly web surveys February, May and August 2026, N=15,156 U.S. adults, margin of error ±1 point; age-group thriving rates, relationship, enjoyment, productivity and standard-of-living items, depression diagnosis, 65+ record). Imagining the Digital Future Center, Elon University, “The Rise of AI Companions: A National Survey” (May 2026, N=4,268 U.S. adult internet users). Common Sense Media, “Talk, Trust, and Trade-Offs” (July 2025, teens and AI companions). McPherson, Smith-Lovin and Brashears, American Sociological Review (2006); Brashears, Social Networks (2011); Small, Brant and Fekete, American Sociological Review (2024).
