https://aiimpactlab.net/impact/ai-job-forecast-needs-a-bridge-between-jobs AIImpactLab en 2026-10-25T14:00:00.000Z An AI jobs forecast is not a plan for the people displaced https://aiimpactlab.net/impact/treasury-proposes-wider-ai-investment-boundaries AIImpactLab en 2026-10-25T14:00:00.000Z The AI race is also a decision about where capital is allowed to go https://aiimpactlab.net/impact/medical-ai-brings-the-test-closer-to-patients AIImpactLab en 2026-10-25T14:00:00.000Z Medical AI can matter before it becomes a better doctor https://aiimpactlab.net/impact/surgeons-want-ai-uncertainty-they-can-interpret AIImpactLab en 2026-10-25T14:00:00.000Z Surgeons want AI to show where it might be wrong https://aiimpactlab.net/impact/claude-formal-proofs-are-strong-because-their-limits-are-explicit AIImpactLab en 2026-10-25T14:00:00.000Z AI-generated software proofs are impressive because they make a limited promise https://aiimpactlab.net/impact/amex-job-cuts-need-a-work-redesign-receipt AIImpactLab en 2026-10-24T14:00:00.000Z Amex plans a 5% cut. What work disappears? https://aiimpactlab.net/impact/openai-containment-is-a-tested-boundary-not-a-promise AIImpactLab en 2026-10-24T14:00:00.000Z OpenAI rebuilt the cage. The test is what it contains. https://aiimpactlab.net/impact/clinical-ai-human-review-must-catch-the-error AIImpactLab en 2026-10-24T14:00:00.000Z A doctor in the loop is not a safety guarantee https://aiimpactlab.net/impact/synthetic-ads-and-court-errors-test-public-trust AIImpactLab en 2026-10-24T14:00:00.000Z An ad can fake a scene. A court cannot fake the record. https://aiimpactlab.net/editorials/a-smaller-ai-promise-can-be-a-bigger-breakthrough AIImpactLab en 2026-10-25T14:00:00.000Z Editorial: A smaller AI promise can be a bigger breakthrough https://aiimpactlab.net/editorials/the-cheapest-answer-can-be-the-most-expensive-decision AIImpactLab en 2026-10-24T14:05:00.000Z Editorial: The cheapest answer can be the most expensive decision