Key points:
- Sam Altman now claims AI will not cause a jobs apocalypse, contradicting his 2023 statements.
- OpenAI is preparing for a $1 trillion IPO, creating potential motives for reassuring language.
- Altman admits his company was “pretty wrong” on social and economic implications.
- Anthropic CEO Dario Amodei still predicts AI could eliminate 50% of entry-level office jobs.
- AI technology has outpaced economic adoption, creating a dangerous lag in understanding real impacts.
The psychological calculus behind Altman’s shifting narrative
Altman’s evolving position reveals a pattern that deserves psychological analysis rather than blind acceptance. When he told The Atlantic in 2023 that “jobs are definitely going to go away, full stop,” he was speaking from a position of technological determinism. He saw the raw power of large language models and extrapolated immediate labor displacement. Now, in 2025, he tells Commonwealth Bank CEO Matt Comyn that “we’ve been roughly right on technological predictions and pretty wrong on the social and economic implications.” This admission of error, however convenient, masks a fundamental tension.
The OpenAI CEO is navigating a psychological landscape where he must maintain credibility with both technologists who fear disruption and business leaders and employees who need stability and assurances that their livelihoods are not going to be taken away. His statement that “we can’t afford to take the psychological convenient option of pretending that the changes were all way off in the future” suggests he understands the cognitive dissonance inherent in his position. He wants transparency about risk while simultaneously downplaying immediate consequences. This is not dishonesty necessarily, but it is strategic ambiguity designed to keep all stakeholders engaged without triggering panic that could slow AI adoption or invite regulatory crackdowns.
Altman’s personal boundary, where he refuses to outsource his own email and Slack communications to AI, reveals a telling contradiction. He will not trust the technology for his most intimate human interactions, yet he asks society to trust his assessment that job displacement is not imminent. This psychological split between personal caution and public reassurance deserves careful attention from anyone questioning the official narrative here.
The gap between technological capability and economic reality
The core tension Altman identifies, that AI models have grown “incredibly smart” while economic adoption remains “very early,” points to a deeper structural problem. The technology is advancing at quadratic time, where doubling model size quadruples training time and cost, but these constraints are rapidly dissolving. The cost of building and distributing large language models is expected to drop to $20,000 within two years, and once scientists replicate the brain’s efficient backpropagation methods, retraining could happen on smartphone CPUs.
This means the current calm before the jobs apocalypse may be temporary, not permanent. Altman’s delight at being wrong could prove premature. His own executives have noted that AI agents are being pushed through communication channels designed for people, but “that was unlikely to be the long-term model.” When persistent, always-running AI systems replace today’s prompt-response tools, the economic impact could accelerate dramatically.
The timing of Altman’s reversal, coinciding with OpenAI’s confidential IPO filing, cannot be ignored. A company seeking $1 trillion valuation cannot simultaneously tell investors that its technology will destroy half of white-collar jobs. The psychological pressure to soften the message is immense. Yet Anthropic’s Dario Amodei, whose company develops the Claude chatbot, continues to predict 50% job elimination. We have two AI CEOs, with two different messages, and the public is left to discern which narrative serves whose interests.
Sources include:
Please contact us for more information.
