Deutsche Bank Experiment: Three Major AI Models Agree AI Is More Likely to Drive Inflation in the Short Term Rather Than Reduce It

Deep Tide TechFlow reports, on April 2, according to Fortune, Deutsche Bank's Chief U.S. Economist Matthew Luzzetti and his team released a research report on March 30, testing the market consensus that 'AI will significantly reduce inflation' by posing questions to three major AI systems. The subjects of the experiment included Deutsche Bank's proprietary tool dbLumina, OpenAI's ChatGPT-5.2, and Anthropic's Claude Opus 4.6. The results were surprising: in a one-year outlook, all three models unanimously agreed that AI's impact on inflation is most likely 'negligible,' and all models assessed that the probability of AI driving inflation is higher than significantly reducing it. Specifically, dbLumina estimated the probability of AI driving inflation at 40%, while the probability of significantly reducing inflation was only 5%; Claude provided probabilities of 25% and 5%, respectively; ChatGPT estimated 20% and 5%. The primary reason cited by all three models was the demand-pull inflation pressure caused by the AI investment boom itself—such as large-scale expansion of data centers, surging demand for semiconductors, and the sharp increase in electricity consumption driven by AI workloads.