Martin Fowler's link roundup covers three topics: an Anthropic study of 80,000 users revealing that most people hold simultaneous hope and fear about AI (with optimism correlating to less-developed countries), a piece arguing that spec-driven LLM development is incomplete without automated test suites to enforce those specs, and a Lawfare article warning that US national security capacity to counter Iranian covert operations has been weakened by recent agency firings.
Sort: