The experiment, detailed in a report titled “What AI Says About AI Eating Itself and the World,” utilized Google’s Gemini 2.5 Pro model to generate a deep analysis of global sectors. The findings suggest that data-rich industries with repetitive tasks are standing on a precipice, while those requiring human empathy or manual dexterity in unpredictable environments remain safe—for now.
(And Fortune Intelligence, the wing of the Fortune newsroom that uses generative AI as a research tool, conducted a meta-meta-experiment to expedite the publishing of this news article about it.)
Perhaps the most ironic conclusion for Silicon Valley is that the sector most exposed to disruption may be the one building the disrupters: information technology and software. The AI found the sector to be particularly susceptible because software development is built on logic and patterns—the very qualities AI systems are designed to automate.
The report notes that over 85% of developers are already using AI coding assistants, with productivity gains of up to 60%. That efficiency boost may help corporations, but it also raises concerns about the long-term sustainability of traditional software licensing models. The recent $2 trillion selloff in software stocks over two weeks, dubbed the “SaaSpocalypse,” underscores investor anxiety and the evaporation of entry-level coding roles.
Beyond tech, the AI set its sights on the financial sector. It identified wealth management as a primary target, predicting an even greater shift toward “robo-advisors.” The report projects that by 2027, AI-driven tools could be the primary source of advice for nearly 80% of retail investors, fundamentally challenging the role of human financial advisors.
However, the robot was humble about its limitations. The report outlined “sectors of resilience” where human traits remain premium currency. Jobs requiring “deep empathy,” such as nursing, therapy, and early childhood education, were insulated from the algorithm’s reach, ceded the AI about its own future dastardly impact.
Additionally, the AI admitted that it struggles with the physical world. Skilled trades like plumbing, carpentry, and construction—which require manual dexterity in messy, unpredictable environments—were deemed the least digitized and least vulnerable sectors. High-level strategic leadership also remains a “human-only” zone, as AI lacks the intuition required for complex corporate negotiation.
Deutsche Bank’s human analysts, Jim Reid and Adrian Cox, noted that the AI’s self-assessment was a “faithful reflection of the current consensus.” However, they cautioned, the machine likely underestimated the physical obstacles to its own takeover, such as the massive energy requirements for data centers and data quality governance.
Ultimately, the AI views its rise as a transformation rather than an apocalypse. While it foresees displacing 92 million jobs by 2030, it also predicts the creation of 170 million new roles, resulting in a net gain for the global workforce. “However, this transition will be disruptive,” Reid and Cox wrote, with estimates suggesting that activities accounting for up to 30% of hours currently worked in the U.S. could be automated by 2030, “necessitating as many as 12 million occupational transitions.”
