Weekly Review 24 October 2025

Some interesting links that I Tweeted about in the last week (I also post these on Mastodon, Threads, Newsmast, and Bluesky):

1. AI elevator inspectors: https://www.extremetech.com/computing/ai-is-coming-for-elevator-inspectors-jobs
2. I don't think the military using commercial AI to help make decisions is a very good idea: https://arstechnica.com/ai/2025/10/army-general-says-hes-using-ai-to-improve-decision-making/
3. AI is hitting graduate recruitment in tech and pharma the most: https://www.theregister.com/2025/10/16/uk_tech_grad_jobs/
4. A bespoke AI chatbot for university students: https://www.insidehighered.com/news/student-success/health-wellness/2025/10/14/dartmouth-builds-its-own-ai-chatbot-student-well
5. The Three Laws of Robotics updated (tongue in cheek) for generative AI: https://www.computerworld.com/article/4070466/asimovs-three-laws-updated-for-the-genai-age.html
6. Once again, AI detectors do not work well enough to be the sole basis of accusations of student misconduct: https://dataconomy.com/2025/10/13/how-a-universitys-ai-witch-hunt-derailed-a-students-career/
7. Now AI-generated ads are coming to device screen savers: https://arstechnica.com/gadgets/2025/10/directv-screensavers-will-show-ai-generated-ads-with-your-face-in-2026/
8. When a medical AI makes a mistake, who is liable? https://www.theguardian.com/technology/2025/oct/13/ai-tools-medical-health-liability-artificial-intelligence
9. I'm not sure a MAGA minion is the right person to advise anyone on bias: https://www.theguardian.com/technology/2025/oct/12/meta-ai-adviser-robby-starbuck
10. What will collapse first? The AI bubble, or the electricity grids? https://www.theregister.com/2025/10/13/ai_power_bills/
11. Using bacteria to build artificial neurons that can connect to biological neurons: https://spectrum.ieee.org/artificial-neuron
12. AI can identify earthquakes that wouldn't otherwise be detected: https://arstechnica.com/science/2025/10/like-putting-on-glasses-for-the-first-time-how-ai-improves-earthquake-detection/
13. British actors' union is taking action over AI-generated "actors": https://www.theguardian.com/technology/2025/oct/13/equity-threatens-mass-direct-action-over-use-of-actors-images-in-ai-content
14. Where is the electricity to run these chips going to come from? https://dataconomy.com/2025/10/14/openai-bets-big-on-energy-hungry-custom-chips-to-scale-chatgpt-and-sora/
15. All the electricity going into AI data centres is coming out as heat, and handling that heat is becoming more difficult as chips become more powerful: https://spectrum.ieee.org/data-center-liquid-cooling
16. When I was growing up the phrase was "the camera never lies". My photographer father showed me that wasn't true. It's even less true now we have AI-driven image editors: https://arstechnica.com/google/2025/10/googles-nano-banana-ai-image-editor-is-coming-to-search-photos-and-notebooklm/
17. Bridging the AI skills gap: https://www.computerworld.com/article/4055753/5-ideas-to-help-bridge-the-genai-skills-gap.html
18. The company that is pushing everyone to use their AI is now complaining that workers are using AI: https://www.theregister.com/2025/10/14/microsoft_warns_of_the_dangers/
19. In an about-face, New Zealand teachers may now use AI to mark student assessments, but must monitor the results: https://www.nzherald.co.nz/nz/ministry-revises-ai-marking-rules-as-teachers-warn-of-confusion/E6AHUPGGGJHPTJ2S4GOT5RCGVA/
20. AI generated erotic material? What took them so long? https://www.theregister.com/2025/10/14/openai_chatgpt_ai_erotica/
21. The majority of articles on the web are now generated by AI: https://graphite.io/five-percent/more-articles-are-now-created-by-ai-than-humans
22. A wellness council that doesn't stop AI from causing harm. Lovely. https://arstechnica.com/tech-policy/2025/10/openai-unveils-wellness-council-suicide-prevention-expert-not-included/
23. Google's AI will use email contents to help you find a meeting slot: https://www.theverge.com/news/799160/google-gmail-gemini-ai-help-me-schedule
24. The pushback against Google's AI summaries: https://arstechnica.com/ai/2025/10/inside-the-web-infrastructure-revolt-over-googles-ai-overviews/
25. Energy is now the biggest bottleneck in AI rollouts: https://www.bigdatawire.com/2025/10/10/powering-data-in-the-age-of-ai-part-1-energy-as-the-ultimate-bottleneck/
26. As AI do more and more online, the internet will need to be redesigned to accommodate them: https://spectrum.ieee.org/agentic-web
27. Any AI that is trained on human-generated material is going to have some political bias: https://www.theregister.com/2025/10/10/openai_gpt5_bias/
28. 71% of workers are using external AI in their work, and it's a huge security risk. The solution? Give them access to the tools: https://dataconomy.com/2025/10/14/71-of-workers-are-using-rogue-ai-tools-at-work-microsoft-warns/
29. AI-generated podcasts are a threat to the industry: https://www.rnz.co.nz/news/world/575817/mass-produced-ai-podcasts-disrupt-a-fragile-industry
30. Kids are using AI to prank their parents, and police are freaking out about it: https://www.theverge.com/news/798681/police-stop-pulling-ai-homeless-man-tiktok-prank