OpenAI CFO Sarah Friar shares five lessons from trying to build finance around AI rather than simply adding AI to existing processes. The central shift is from automating individual tasks to redesigning the full path from source data to decision—while enabling finance professionals to build their own tools, keeping human accountability explicit, and measuring whether AI produces dependable business value. For knowledge workers, the useful lesson is that AI-native work is primarily a workflow and operating-model change, not a software rollout.

This EY-Parthenon article challenges four common assumptions about AI’s economic impact: that productivity will surge immediately, AI is nearly free to use, workers will quickly become redundant, and AI will rapidly reduce inflation. Its central argument is that technological capability is only the starting point; the economic effects depend on infrastructure, recurring costs, organizational redesign, workforce adaptation and time. For knowledge workers and business leaders, the useful lesson is to judge AI by measurable value and how effectively people and processes change around it.

Satya Nadella argues that enterprise AI advantage will not come simply from access to the best frontier model, but from building a learning loop between human capital and company-owned AI capability. His useful distinction is that human capital — judgement, relationships, creativity, domain knowledge, and pattern recognition — becomes more valuable when it is connected to “token capital”: the AI capability a firm builds from its own workflows, knowledge, evaluations, and execution traces. The broader warning is economic as well as technical: if companies only rent intelligence from a few general models, they risk losing control over their own learning, differentiation, and returns. For knowledge workers, the value is that this reframes AI adoption from “which model should we use?” to “how do we preserve and compound what our organisation knows?”

A policy argument for treating AI as a public-finance and social-contract question, not only a technology question. Senator Elizabeth Warren argues that if AI concentrates wealth, replaces workers, raises data-centre energy costs, and depends on public infrastructure and research, then the gains from AI should be taxed and used to support people through health care, education, jobs, and unemployment protection. Whether or not the reader agrees with the proposal, the article is useful because it frames AI’s impact as a question of who captures the benefits and who pays the costs.

A useful look at how Chinese AI models are becoming part of the U.S. AI ecosystem, not by beating American frontier models in prestige, but by being cheaper, open-weight, efficient, and good enough for many real developer workflows. The bigger point is that AI competition may not be decided only at the very top of the model leaderboard. It may also be shaped by pricing, openness, infrastructure constraints, startup adoption, and whether export controls push competitors to become more resourceful.

A thoughtful future-of-work essay arguing that AI automation does not simply make expert human work disappear; it changes where that work happens. Dan Shipper’s core point is that AI makes many forms of competence cheap, which increases the volume of output, but also creates sameness, review burden, and new demand for human judgement. The useful insight for knowledge workers is that the valuable human role moves toward framing the problem, judging quality, creating difference, maintaining agent systems, and deciding what matters now.