A growing number of elite AI researchers are grappling with a deeply personal question: does it make sense to plan a future when they believe their profession may cease to exist within 18 months? This isn't hyperbole or burnout chatter. It's a documented psychological phenomenon now recognized by psychiatrists as a clinical condition, and it's spreading from AI labs to mathematics departments and beyond.

What's Happening Inside AI Labs Right Now?

In a recent podcast episode, Silicon Valley investor Elad Gil revealed that researchers at multiple top-tier AI labs are seriously discussing whether marriage makes sense given their belief that artificial general intelligence (AGI), or AI systems matching human-level reasoning across all domains, could arrive within 18 months. The concern isn't about relationship problems; it's existential. If AI achieves recursive self-improvement, a process where AI systems improve themselves without human intervention, these researchers worry they'll no longer be needed as intellectual workers.

This anxiety has triggered what the industry calls a "wave of departures." Unlike typical job changes driven by better offers, these resignations come with heavy-toned open letters. In February 2026, Mrinank Sharma, head of the safety research team at Anthropic, posted a resignation letter stating "the world is in danger" that received 14.7 million views. Days later, Zoë Hitzig, a researcher at OpenAI, published her resignation in The New York Times. Earlier departures included Jan Leike, former head of OpenAI's Superalignment team, and Ilya Sutskever, OpenAI co-founder, who founded Safe Superintelligence Inc. and raised $3 billion without releasing any products to date.

Hieu Pham, a former OpenAI and xAI employee who helped build some of the world's most powerful AI systems, offered perhaps the most candid account. He wrote: "I used to scoff at the idea of deteriorating mental health. But it is real, painful, terrifying and dangerous." He returned to Vietnam with his family seeking healing.

Nathan Lambert, a senior research scientist at AI2, described the work culture at OpenAI and Anthropic as equivalent to "996" culture, a term referring to working from 9 a.m. to 9 p.m., six days a week. The logic driving this intensity is chillingly precise: if programming problems are solved by end of 2026 and mild recursive self-improvement appears by end of 2027, one researcher calculated that each remaining week represents 2% of their remaining productive career, necessitating 16-hour workdays.

Is This Anxiety Based on Realistic Timelines?

Sarah Guo, Elad Gil's partner and fellow investor, raised a critical counterpoint: the prediction that AGI will arrive in 18 months has been repeated every 18 months for the past five years. Yet this historical pattern of missed timelines hasn't stopped top AI talent from reorganizing their lives around the belief that the timeline is real this time.

The anxiety isn't confined to AI labs. It's spreading to mathematics, a discipline that has existed for thousands of years. In 2026, large language models (LLMs), or AI systems trained on vast amounts of text to predict and generate language, began proving professional mathematician-level theorems almost weekly. OpenAI's internal models released batches of mathematical breakthroughs simultaneously. A phrase circulating on social media declared: "This year's Fields Prize will be the last one".

The Fields Prize was indeed awarded to four mathematicians in July 2026, but media coverage carried an ominous undertone. The AFP used the phrasing "as AI reshapes the mathematics industry." Stanford University held a seminar inviting three Fields Prize winners and researchers from OpenAI and DeepMind to discuss "the future of mathematics." Terence Tao, a renowned mathematician, proposed that people need to move beyond asking "whether AI can generate proofs" and instead rethink the mathematical workflow itself.

A New Clinical Diagnosis: AI Replacement Dysfunction

When enough people exhibit similar psychological symptoms, the psychiatric community names the condition. In September 2025, researchers Stephanie McNamara and Joseph Thornton from the University of Florida College of Medicine published a paper in the journal Cureus formally proposing a new clinical concept: Artificial Intelligence Replacement Dysfunction, or AIRD.

AIRD differs fundamentally from traditional unemployment anxiety. It attacks the core of personal identity. Thornton called it an "invisible disaster." The condition manifests through several interconnected symptoms:

• Anxiety and Insomnia: Persistent worry about professional relevance disrupts sleep patterns and daily functioning.
• Paranoia and Loss of Identity: Sufferers question their fundamental worth and unique value as thinking beings.
• Sense of Worthlessness: The existential threat creates feelings that decades of training and professional accumulation may be meaningless.
• Appearance in Previously Healthy Individuals: AIRD can emerge in people with no prior history of mental illness, suggesting it's driven by external threat perception rather than internal pathology.

The paper emphasizes that AIRD pain "is not rooted in traditional psychopathology, but in the existential threat of professional obsolescence".

This distinction matters. When textile workers, carriage drivers, and typists were displaced by previous technological waves, their pain was primarily economic; they lost income and struggled to support families. AIRD patients experience existential pain: "If AI can do everything I do, what do my 20 years of training and accumulation mean? What unique value does my mind still have?".

How Are Different Professions Responding to AI Threat Perception?

Research from the University of Mannheim examined how doctors react to AI and found that perceptions of "threat to professional competence" and "threat to professional recognition" directly lead to resistance toward AI adoption. Interestingly, medical students showed stronger identity threat and resistance than experienced doctors. Young professionals who haven't fully established their professional identity are more easily shaken by AI capabilities.

Broader workforce data reveals the scale of this anxiety. A KPMG survey found that 52% of employees worry that AI will threaten their job security. Quantum Workplace's analysis of voices from more than 700,000 employees discovered that workers who use AI frequently report a burnout rate of 45%, significantly higher than the 35% burnout rate among people who don't use AI. The pattern is clear: the more employees use AI, the more anxious they become.

What Should Organizations and Individuals Know About This Emerging Crisis?

The emergence of AIRD and widespread anxiety among elite knowledge workers signals a fundamental shift in how technological disruption affects human psychology. Unlike previous waves of automation that primarily threatened manual labor and routine work, AI threatens the cognitive and creative domains where many professionals have invested their identities.

The fact that this anxiety is most acute among the world's smartest people, those closest to AI development, suggests it's not irrational fear but informed concern. These researchers have direct knowledge of AI capabilities and timelines. Their departures from prestigious positions, often accompanied by statements about existential risk, represent a form of professional whistleblowing about the pace and direction of AI development.

The mathematics field's sudden confrontation with AI capability offers a preview of what other knowledge-intensive professions may face. When Terence Tao, one of the world's greatest living mathematicians, publicly acknowledges that AI is solving important problems in his field, it signals that no domain of human expertise is immune to disruption.

Whether the 18-month timeline to AGI proves accurate or repeats the pattern of previous missed predictions, the psychological impact is already real. Thousands of highly talented people are reorganizing their lives around existential uncertainty, experiencing clinical-level anxiety, and questioning fundamental life decisions. This represents a new category of AI risk, one that operates not through direct harm but through the psychological weight of perceived obsolescence.