America's top companies keep talking about AI – but can't explain the upsides (www.ft.com)

Top U.S. companies are loudly proclaiming AI as central to future strategy, but public communications — earnings calls, investor decks and press releases — often stop at buzzwords rather than quantifiable outcomes. Analysts and journalists note a rising cadence of “AI” mentions tied to promises of automation, personalization and efficiency gains, yet firms rarely present concrete metrics (revenue uplift, cost savings, productivity improvement), model details (type, size, training data), or deployment outcomes (A/B test results, latency, failure rates). That gap leaves investors and the public guessing whether these initiatives are pilot projects, production systems, or marketing spin. For the AI/ML community this matters: vague claims hinder reproducibility, risk assessment and responsible deployment. Without standardized reporting—model cards, dataset provenance, evaluation metrics, monitoring for drift and fairness—organizations and regulators can’t judge real impact or systemic risks. The practical implication is a potential misallocation of capital and talent, and slower progress on safety, auditability and governance. The story underscores a growing demand for transparency: rigorous metrics and operational details will be necessary to separate genuine AI-driven value from hype and to enable accountable scaling of machine-learning systems.