'This is a training and trust problem, not an AI problem': Barely any workers are verifying information given to them by AI
THE SO WHAT
If under a quarter of employees are verifying AI output — and younger workers are asking the model before their manager — your risk surface is now cultural, not just technical. Treat AI verification like code review: design it into workflows, train for it explicitly, and instrument where it’s skipped.
READ THE SOURCE
MORE FROM THE WIRE
Applied AINvidia is set to work even closer with Meta's slightly creepy Muse AI models — and it's even investing $3bn in Meta to show its support
When your primary GPU supplier writes a $3 billion equity check, the relationship is no longer vendor–customer — it’s strategic alignment on roadmap and priority access. If you’re not a top-3 account for a major chip vendor, assume your pricing, allocation, and co-development leverage are structurally worse and plan capacity accordingly.
Applied AIOn Amazon's earnings call, Andy Jassy says outside AI shopping agents are "still not great, nor more convenient than coming to a retailer directly"
Amazon is telling the street it doesn’t see third-party shopping agents as an immediate threat — which buys incumbents a bit of time but also signals they’ll defend the retail surface aggressively. If you’re experimenting with commerce agents, assume limited cooperation from major marketplaces and design for scraping, email parsing, and direct brand integrations instead of official APIs.
Applied AINvidia in Talks to Offer Credit Support for Nscale Project
When Nvidia starts backstopping data center debt, GPUs are no longer the only choke point — financing and power become part of the bundle. For AI cloud buyers, this means more vertically integrated offerings and potentially better access to capacity, but also deeper lock-in to a single vendor’s financial ecosystem.
Applied AIAWS will be a $1T business, Jassy says. But first, it will spend big.
AWS is explicitly trading near-term free cash flow for a shot at owning AI cloud — that means more aggressive buildout of custom silicon, data centers, and higher-level services. As a customer, you’ll see richer AI primitives and better pricing in the near term, but you should also expect more opinionated platform moves that make multi-cloud harder in practice.