Parents have heard the question before. A child glances up from a game or book, eyes fixed on a smartphone glowing in an adult’s hand. “Mommy, do you love your phone more than me?” The Frontiers in Psychology study published in June 2026 captured that raw sentiment. It revealed something deeper. Adolescents who saw their caregivers distracted by devices reported higher levels of insecure attachment. Both to mother-like and father-like figures.
The research, led by Don Grant and colleagues, developed the Device Attachment Interference Scale. In a sample of 600 U.S. adolescents aged 12 to 17, higher scores on this measure consistently predicted greater anxious and avoidant attachment. The numbers stood out. Regression coefficients showed strong links. For mothers, avoidant attachment rose with a b of 0.630. Anxious climbed to 0.819. Patterns held for fathers too. Yet this wasn’t about AI. It examined everyday tech interference. Phubbing. Technoference. Behaviors now amplified by generative tools that demand constant attention.
Fast forward months. AI systems handle more cognitive work. Summarizing reports. Generating code. Brainstorming strategies. And fresh evidence piles up. Reliance risks eroding core human capacities. But outcomes hinge on exactly how people engage the technology.
Zara Abrams laid this out clearly in the American Psychological Association’s Monitor on Psychology on July 1, 2026. “There’s clear evidence that people can perform many tasks better with AI,” said Brooke Macnamara of Purdue University. “The question is: What happens to our skills if we start relying on it regularly?” The piece cataloged studies showing both gains and losses. Physicians using AI for polyp detection saw accuracy drop from 28.4 percent to 22.4 percent in three months. One Lancet Gastroenterology & Hepatology paper from 2025 documented the slide.
Offloading explains much of it. Humans act as cognitive misers. They choose the path of least resistance. Evan Risko from the University of Waterloo pointed to that tendency. Memory research backs him. When people expect external storage, recall suffers if access vanishes. Lauren Richmond of Stony Brook University noted the pattern in experiments dating back years.
Yet passive use drives the worst effects. Transactional queries. Ask. Receive. Move on. Mutlu Cukurova of University College London warned these interactions foster atrophy over growth. “People learn best when they engage in active exploration, problem-solving, and sensemaking,” Shiri Melumad of the Wharton School added in the same APA report. AI summaries short-circuit that process. Research becomes passive consumption.
James C. Kaufman sees amplification instead of replacement. The UConn educational psychology professor tested storytelling tasks. Participants worked alone or with AI. Those strong in creativity and intelligence beforehand performed even better alongside the tools. “What we found is that creativity and intelligence still matter,” Kaufman told UConn Today on January 13, 2026. “Participants who were more creative without AI also tended to perform better when collaborating with AI.” AI generates ideas effectively. Evaluation demands human judgment. Selection. Refinement. Those steps remain ours.
But equity questions loom. Paid versions improve faster. Free tiers lag. Kaufman flagged the divide. “If you already have strengths in a domain, you should be able to use AI more effectively,” he said. “AI doesn’t suddenly make everyone equally creative.” Creativity has long been democratic. Differences across gender, culture, and income stayed modest. Technology could widen gaps.
Wharton researchers uncovered another twist. AI sharpens individual outputs. It flattens collective diversity. Christian Terwiesch and colleagues reexamined creative tasks. One involved inventing toys from a fan and building bricks. Human-only groups produced completely unique concepts. AI-assisted ones converged. Only 6 percent of ideas stood apart. “If you rely on ChatGPT as your only creative advisor, you’ll soon run out of ideas, because they’re too similar to each other,” Terwiesch explained in the Knowledge at Wharton article from July 2025. The model averages likely completions. Same prompt. Similar distribution. Diversity suffers.
Lennart Meincke, a Mack Institute fellow, elaborated. “When you give the model the same prompt, it tries to average the most likely completions based on that input.” The result? Predictable output. Teams risk echo chambers. Innovation stalls without deliberate safeguards. Vary prompts. Begin with human ideas. Test multiple models. Protect divergence explicitly. “Diversity is often overlooked, but it needs special protection,” Terwiesch stressed. “If you don’t solve for it explicitly, you won’t get it.”
Thomas Germain explored these risks in a May 2026 BBC Future feature. Adam Green of Georgetown University put it bluntly. “If you are not doing as much of the actual thinking, then your capability to do that kind of thinking is going to atrophy.” Heavier AI users scored lower on critical thinking tests. A survey of 494 students linked more ChatGPT use to self-reported memory issues. AI ideas proved less original. More formulaic.
Yet brains adapt. Past technologies reshaped habits without collapse. GPS reduced spatial memory reliance. People navigated less independently. Jared Benge’s meta-analysis of 57 studies and 411,000 adults found no “digital dementia.” Tech sometimes lowered cognitive impairment risk. The difference lies in intent. Add friction. Draft ideas on paper first. Ask AI to quiz you afterward. Form your own rough view before consulting the machine. Tolerate discomfort. Slow down on purpose.
Nita Farahany of Duke University captured the societal stakes in the APA monitor. “One of the things that’s seductive but also troubling about generative AI is the ability to offload much more executive functioning than we have in the past. What does it mean for society if humans are passively receiving information but no longer able to critically interrogate it?” Passive mode erodes competence. Active engagement builds it.
Michael Gerlich’s 2025 study in Data reinforced the nuance. Deliberate prompting enhanced critical thinking. Passive reliance diminished it. Jackson G. Lu of MIT observed metacognition as the key moderator. Employees high in self-reflection used AI more productively. They reflected on outputs. Challenged assumptions. Turned tools into collaborators rather than crutches.
Mindy Shoss from the University of Central Florida focused on organizations. High workloads and job insecurity push workers toward quick AI fixes. Without norms or incentives for exploration, skills stagnate. “Organizations tend to be most successful when they set clear norms and guardrails for AI use,” she said. David Evans at Microsoft offered optimism. Extra time from automation could fuel curiosity. Learning. Upskilling. If leaders create space for it.
The Frontiers paper on parental devices hinted at relational costs. Caregivers distracted by screens modeled divided attention. Adolescents felt it. Insecure bonds followed. Scale that to AI. Colleagues glued to chat interfaces during meetings. Parents prompting models while kids wait for eye contact. The attachment study controlled for age and gender. Associations held. Device interference acts as a relational signal. Availability matters.
Current X discussions echo the tension. Designers lament clients generating logos in seconds that rival human concepts. One post from July 10, 2026, captured resignation. AI understood brand values and produced minimal symbols. Yet responses pushed back. Human creativity persists in evaluation and context. Engineers still hire specialists for nuanced UX. Machines accelerate. They don’t replace judgment.
Robert J. Sternberg argued in a 2026 Frontiers in Education article that AI has already compromised aspects of intelligence and creativity. The question isn’t future risk. It’s current reality. And mitigation.
So what now? Treat AI as a sparring partner. Not an oracle. Draft first. Query second. Critique always. Organizations must measure more than output speed. Track skill development. Idea diversity. Critical depth. Individuals guard their cognitive muscle. Write longhand occasionally. Solve problems offline. Question AI answers aggressively.
The adolescent voices in that 2026 psychology study delivered a simple warning. Attention signals care. When devices or algorithms claim focus, relationships shift. Cognitive skills follow similar logic. Use them. Or watch them fade. The evidence grows clearer each month. Human strengths in originality, evaluation, and connection don’t vanish. They require exercise. In an age of capable machines, that practice becomes deliberate choice.
Recent work from Stanford scholars in March 2026 showed promise in tuned AI for visual artists. Better collaboration. Precision prompts. Yet the core lesson repeats. Tools augment best when paired with strong human foundations. Creativity and cognition endure. Provided people refuse to outsource everything.
WebProNews is an iEntry Publication