By Jagdish Prasad Yadav, Senior Occupational Therapist and Founder, Locus Thrive (Newbee Hub Limited)
Messaging fatigue. Fewer close friends. A shorter fuse for attention. Low-grade anxiety. The feeling something bad is about to happen. These don’t show up alone — they arrive together, in the same people. Here’s what the data actually says.
Attention is shrinking. The average time someone stays focused on one screen has dropped from 2.5 minutes in 2004 to 47 seconds today (UC Irvine). One 2026 study found it takes 26.8 minutes to fully recover focus after a single interruption.
Friendships are collapsing. Americans with no close friends: 3% in 1990 → 17% in 2024. Time spent with friends dropped 37% between 2014–2023. The US Surgeon General now calls this a loneliness epidemic, comparable in health impact to smoking 15 cigarettes a day.
Anxiety has a name now: doomscrolling. Compulsive negative-content consumption, reinforced by infinite scroll, is consistently linked to anxiety, depression, and stress across dozens of 2025 studies. The loop: discomfort drives checking, checking surfaces more alarm, the alarm drives more checking.
Adults can’t opt out either. Even motivated adults given self-set screen limits mostly fail to hold them — willpower alone consistently loses to impulse (2026 study). What actually worked: small built-in friction, which cut app opens by 57% in one trial — not discipline, design.
AI is adding a new layer. A 2026 Drexel study found teen AI-companion use mapping onto the textbook stages of behavioural addiction. Separate 2026 research found teens substituting AI chatbots for real conversations — even after arguments with real partners — which researchers worry is quietly resetting what young people expect a relationship to feel like.
The way forward isn’t willpower. The evidence points elsewhere: build friction into the design, not the person. Build the underlying skill — attention, delay, sitting with discomfort — early, while it’s still a skill in progress rather than a habit already set. And simply naming the mechanism — this is engineered, not a personal failing — is often what changes how people respond to it first.
Sources: UC Irvine (Gloria Mark) attention-span research · Survey Center on American Life, friendship data · 2025 doomscrolling scoping review, Mental Health and Digital Technologies · 2026 Ulysses Donation Pledge study, ScienceDirect · Drexel University & Arizona State University/Lancet Child & Adolescent Health, 2026 AI-companion studies