{"slug": "ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and", "title": "AI Weekly Issue #514: Applied AI Is Here: What's Working, What Got Pulled Back, and Why Now", "summary": "AI Weekly published a library of 159 real AI deployments across 21 industries, with reported outcomes on 77 and six halted projects. The library shows narrow, measurable tasks like OpenAI's Codex agent adoption (97.9% of employees), NHS chest X-ray AI (4 million patients), and Wonder's robot making 500 bowls per hour are working, while Ford rehired 350 inspectors after AI defects and Waymo paused robotaxis in four cities after a flood incident.", "body_md": "*Today we're publishing the AI Use-Case Library: 159 real, named AI deployments across 21 industries, with the tools, vendors, and reported outcomes on 77 of them. Six were halted or reversed, and those might be the most useful entries in the file. It's free, no signup, and built to search before you spend a budget or write a pitch. This edition is the tour: what's working, what got pulled back, and why a precedent file matters right now.*\n\n## Sponsor\n\n[Validate agents you build, buy, or integrate.](https://www.spec27.ai/?utm_source=ai_weekly&utm_medium=newsletter_sponsorship&utm_campaign=aiewf_2026&utm_content=2026_07_06_built_or_bought)\n\nSpec27 helps teams apply consistent behavioural validation across custom-built agents, third-party tools, and AI-powered workflows.\n## The receipts are live\n\nManagers don't need more AI news. They need precedent: has anyone in my industry actually done this, with what tools, and did it work.\n\nSo we built it. The [AI Use-Case Library](https://aiweekly.co/ai-use-cases) is live today: 159 documented deployments, 21 industries, reported outcomes on 77 of them, every entry linked to its source.\n\nNo signup. Search by industry, by function, by vendor, or by whether the project is still running. Below is a tour of what the file reveals.\n\n## What's actually working\n\nAdoption is real where the task is narrow and the outcome is measurable. A sample from the library:\n\n**OpenAI now runs on its own agents.**[97.9% of OpenAI employees use its Codex agent](https://www.theregister.com/ai-and-ml/2026/06/25/openai-says-employees-moving-beyond-chat-to-agents/5262499), up from about 40% last August, with Legal and Recruiting now on it too. (The Register)**The NHS is reading chest X-rays with AI at national scale.** A[£20M expansion has already helped over 4 million patients](https://www.gov.uk/government/news/ai-to-speed-up-cancer-diagnosis-for-millions-of-nhs-patients)get faster lung-cancer diagnoses or all-clears, cutting analysis on complex cases from eight days to four. (GOV.UK)**A robot is making 500 bowls an hour.** Wonder put[the bowl-making system it bought from Sweetgreen](https://fortune.com/2026/06/09/marc-lore-robots-make-500-burrito-bowls-an-hour-a-human-can-make-45/)into its first kitchen. A human line cook makes about 45. (Fortune)**Assisted driving shipped in close to a million cars.** Momenta's urban Navigate-on-Autopilot software is[in 680,000 to 900,000-plus production vehicles](https://www.reuters.com/business/autos-transportation/chinas-momenta-seeks-up-751-million-hong-kong-ipo-boost-autonomous-driving-rd-2026-06-29/)from Toyota, Mercedes-Benz, BYD, GM and Audi. (Reuters)**Pinterest's AI ad engine hit its first $1B quarter.** Its[AI-powered Performance+ campaigns now drive about 30% of lower-funnel revenue](https://www.cnbc.com/2026/05/04/pinterest-pins-q1-earnings-report-2026.html), and advertisers who adopted them grew that spend at nearly twice the rate of those who didn't. (CNBC)**China Post sorts mail with humanoids.** Robots at its Guangzhou hub[process up to 1,200 parcels an hour](https://interestingengineering.com/ai-robotics/china-deploys-humanoid-robots-in-postal-hub)at a site that moves 6.5 million pieces a day. (Interesting Engineering)\n\n## What companies pulled back\n\nThe six halted entries are the cheapest lessons in the library. They show where AI quietly fails: institutional knowledge, edge cases, and anything a model will confidently get wrong. A few:\n\n**Ford rehired 350 quality inspectors.** Its[automated quality systems produced defects](https://www.bloomberg.com/news/articles/2026-06-25/ford-has-been-rehiring-quality-inspectors-after-ai-fell-short)that only experienced human inspectors reliably caught. (Bloomberg)**Waymo paused robotaxis in four cities.** A robotaxi[drove into an Atlanta flood and sat stuck for about an hour](https://techcrunch.com/2026/05/21/waymo-pauses-service-in-four-cities-as-robotaxis-keep-driving-into-floods/); Waymo then paused service across Atlanta, San Antonio, Dallas and Houston, the last two as a precaution for forecast weather. (TechCrunch)**Meta dropped automated hate-speech moderation.** After it switched to a Community Notes model,[abusive and racist posts targeting US legislators tripled](https://www.wired.com/story/threats-against-politicians-skyrocketed-after-meta-changed-its-speech-rules/)within six months. (Wired)**A federal judge threw out DOGE's AI grant cuts.** DOGE used ChatGPT to flag about $100M in NEH grants as \"DEI-related.\" The[model labeled a Holocaust literature anthology that way](https://www.bloomberg.com/news/articles/2026-05-07/doge-slammed-by-judge-for-using-ai-to-find-100-million-in-cuts), and a 143-page ruling found unconstitutional viewpoint discrimination. (Bloomberg)**Wake County schools banned AI detectors.** A[student given a zero on a detector flag appealed](https://www.govtech.com/education/k-12/wake-county-n-c-schools-revise-ai-policy-with-more-caution)and had the grade changed to 100 when a second teacher found no AI use. (GovTech)\n\n## The Audit Era, With Receipts\n\nHere's why a precedent file matters this week. The AI market just moved from \"can it do this\" to \"what did it cost, and did it work.\"\n\nAmazon CTO Werner Vogels [told Fortune](https://fortune.com/2026/07/10/amazon-cto-companies-shifting-toward-cheaper-opensource-ai-models-werner-vogels/) that enterprises are shifting to cheaper open-source models because the bills got real. He cited Uber burning through its entire 2026 AI budget in four months, and another company running through half a billion dollars in a single month before it capped employee usage.\n\n[Bloomberg reported](https://www.bloomberg.com/news/articles/2026-07-12/openai-meta-spacexai-compete-for-more-cost-efficient-ai-models) OpenAI, Meta and xAI are racing to undercut Anthropic on price as buyers scrutinize their invoices. [404 Media documented](https://www.404media.co/companies-are-throttling-employees-ai-use-because-its-too-expensive/) Amazon, Adobe, Atlassian and Citi throttling employee AI use, with one firm's monthly spend tripling past $15 million.\n\nWhen budgets get audited, \"we think it'll work\" stops being enough. Buyers want to see who already tried it and what happened. That's the whole reason we built the library, and why the halted entries carry as much weight as the wins.\n\n## How to use it before you spend\n\n- Filter by your industry and function first. If someone in your seat already ran the project, the outcome is right there.\n- Read the halted entries before the wins. They mark where AI still misses the judgment a human carries.\n- Every entry links its source, so you can take the primary reporting straight into a budget meeting or a pitch.\n\n## Key Takeaways\n\n- The wins are concrete and uneven. OpenAI runs on agents, the NHS reads X-rays, a robot makes 500 bowls an hour. Adoption lands where the task is narrow and measurable.\n- The failures rhyme. Ford, Waymo and Wake County all pulled back where AI missed something a human still catches.\n- The audit era is here. With Vogels flagging blown budgets and buyers scrutinizing invoices, precedent beats hype in a procurement conversation.\n- Free and searchable beats another slide deck. The library is built to be used before you commit a budget.\n\n## Worth Reading\n\n[Anthropic committed $10M CAD to eight Canadian research institutions](https://www.anthropic.com/news/canadian-ai-research), including Mila, Vector, Amii and the University of Toronto, with Claude credits and startup access. (Anthropic)[The FTC opened public comment on a policy statement on AI accuracy](https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy). If you sell or deploy AI to consumers, this is the docket to read. (FTC)[Knight Columbia's \"AI as Social Technology\"](https://knightcolumbia.org/content/ai-as-social-technology)argues for treating AI as a social technology, with civil-rights and regulatory implications if the frame sticks. (Knight Columbia)\n\n## Wait, What?\n\n**A rural upstate New York school district is putting a humanoid robot in a classroom this fall.**[The Salamanca City Central School District](https://www.edtechinnovationhub.com/news/ai-gets-a-body-as-humanoid-robot-enters-new-york-classroom), on the Seneca Nation reservation, is deploying a Realbotix M-Series humanoid and an AI teacher's assistant called Optio for about 500 high-school students, as part of its own AI and robotics curriculum. EdTech Innovation Hub calls it a landmark moment while noting it's still a single-district test. (EdTech Innovation Hub)\n\n## Worth Watching\n\nThe videos AI practitioners are passing around right now — curated on [AI TV](https://aiweekly.co/ai-tv).\n\n404 Media |\n\n[Sundar Pichai on A.I. Backlash, the Future of Work and Google’s Next Era](https://aiweekly.co/ai-tv?v=RgV57kDzcng)Hard Fork\n\n## This week's poll\n\n**Twelve months from now, your organization's AI spend is:**\n\nLast week, 208 of you voted:\n\n**Regulators called it systemic risk this week. Where is the AI trade 12 months from now?**\n\n**Twelve months from now, your organization's AI spend is:**\n\nBack Friday.\n\nAlexis", "url": "https://wpnews.pro/news/ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and", "canonical_source": "https://aiweekly.co/issues/applied-ai-is-here-whats-working-what-got-pulled-back-and", "published_at": "2026-07-15 00:00:00+00:00", "updated_at": "2026-07-15 09:54:32.563429+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-tools", "ai-infrastructure", "ai-ethics"], "entities": ["OpenAI", "NHS", "Wonder", "Momenta", "Pinterest", "Ford", "Waymo", "AI Weekly"], "alternates": {"html": "https://wpnews.pro/news/ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and", "markdown": "https://wpnews.pro/news/ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and.md", "text": "https://wpnews.pro/news/ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and.txt", "jsonld": "https://wpnews.pro/news/ai-weekly-issue-514-applied-ai-is-here-what-s-working-what-got-pulled-back-and.jsonld"}}