A crisis of meaning
The future is already here. It is simply not evenly distributed.
On 25 July 2026, mathematician Kirwin Hampshire published The Dark Night of Mathematics. He was not simply expressing concern about the performance of artificial intelligence. He described the distress of seeing a technology approach the very heart of what gave meaning to his profession: searching, discovering, creating and being recognised for patiently built expertise.
“I’m going through a profound spiritual crisis because of these developments. I’ve been screaming internally for days. I feel like I’m living in a nightmare.”
Kirwin Hampshire
Seven days later
On 1 August, OpenAI announced ten results in mathematics and theoretical computer science produced by an internal version of Astra, its next major model. Each resolves a long-standing open problem or delivers a substantial advance. According to OpenAI, the compute required to search for these solutions would represent about USD 2,000 at Sol API prices.
The manuscripts were then prepared by humans using the same model, and the arguments formalised as Lean certificates. Some of these problems had occupied mathematical communities for a long time. Their resolution illustrates a rapid shift in the boundary between research assistance and the direct production of knowledge.
The contrast is striking: on one side, an announcement of unprecedented scientific capability; on the other, a professional describing what its arrival is already doing to the meaning he gives his work.
Transformation can begin before jobs disappear
What is happening in mathematics today may foreshadow what other professionals experience tomorrow. Anxiety about obsolescence, loss of recognition, a sense of dispossession from one’s occupation, conflicts of values and reduced autonomy are not abstract side effects of AI.
They can become genuine psychosocial risk factors when technology transforms activity faster than workers and teams can appropriate it. One person’s account cannot be generalised or establish a causal health effect. It can, however, reveal a tension before employment or productivity indicators show it.
The AI wave will only be sustainable, effective and socially acceptable if work organisation evolves with it. This means discussing the place left for human discovery, criteria for recognition, career pathways, room for manoeuvre and the role of teams when AI enters the heart of expertise.
The future is already here. Occupational-health professionals must be here too.
Point to watch
Meaning can be weakened before employment itself. Monitoring job losses alone would leave loss of recognition, dispossession from an occupation, conflicts of values and professional uncertainty invisible.