Weekly Review 29 November 2024

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

1. Looks like everyone is jumping on the lawsuit bandwagon. This will likely continue until a definitive ruling is made about whether using data to train an AI constitutes copying that data: https://techcrunch.com/2024/11/18/indian-news-agency-sues-openai-alleging-copyright-infringement/
2. While most people are optimistic about generative AI, security experts aren't keen on it: https://www.techrepublic.com/article/ciso-generative-ai-problems-ntt-data/
3. An overview of AI agents: https://spectrum.ieee.org/ai-agents
4. From experimenting to rolling out AI in the organisation: https://www.datanami.com/2024/11/20/four-steps-to-go-from-experimentation-to-embedding-ai-across-the-enterprise/
5. Large scale AI needs large scale nuclear power, other energy sources emit too much carbon: https://www.theregister.com/2024/11/20/hokkaido_electric_power_nuclear_datacenter_ambition/
6. More progress in medical applications of AI: https://www.extremetech.com/science/microsofts-new-ai-model-can-spot-cancer-chest-infections-and-more
7. AI agents are just like any other software project, you need good documentation around them or people forget what they're for: https://www.theregister.com/2024/11/21/gartner_agentic_ai/
8. Quantization, a technique that makes AI more energy efficient, has a downside: https://techcrunch.com/2024/11/17/a-popular-technique-to-make-ai-more-efficient-has-drawbacks/
9. An analog AI chip that is much more power efficient than digital: https://spectrum.ieee.org/analog-ai-2669898661
10. Large language model AI are now being applied to military tasks: https://spectrum.ieee.org/ai-used-by-military
11. A high level API layer that abstracts away a lot of the donkey work with building AI: https://techcrunch.com/2024/11/21/lightning-ai-looks-to-make-managing-ai-a-piece-of-cake/
12. AI agents are going to become more and more prevalent in business: https://www.computerworld.com/article/3609764/ai-agents-are-coming-to-work-heres-what-businesses-need-to-know.html
13. AI generated advertisements are still not quite ready for mainstream usage: https://www.nzherald.co.nz/entertainment/coca-cola-christmas-ad-2024-cokes-ai-generated-commercials-slammed-by-consumers/YHGCFVAHY5HNPNLVJERVGUNA2E/
14. Think of the hilarity if an AI Jesus started to hallucinate answers: https://www.theguardian.com/technology/2024/nov/21/deus-in-machina-swiss-church-installs-ai-powered-jesus
15. How AI can help clean up plastic pollution: https://www.informationweek.com/machine-learning-ai/ai-and-the-war-against-plastic-waste
16. The role of AI in automating the global supply chain: https://www.techrepublic.com/article/ai-supply-chain-automation-global/
17. Ben Affleck is right that AI can't produce good movies. But they don't need to, they only need to produce profitable movies: https://techcrunch.com/2024/11/18/ben-affleck-tells-actors-and-writers-not-to-worry-about-ai/
18. AI that improve the energy efficiency of factories: https://techcrunch.com/2024/11/18/juna-ai-wants-to-use-ai-agents-to-make-factories-more-energy-efficient/
19. I feel that people shouldn't have to be told to not upload their medical data into AI training sets: https://techcrunch.com/2024/11/19/psa-you-shouldnt-upload-your-medical-images-to-ai-chatbots/
20. Spotting scam calls in real-time. But scammers are using AI to make the calls now, so it's really AI against AI: https://www.extremetech.com/mobile/google-introduces-ai-scam-detection-for-pixel-phones
21. AI continue to improve at finding bugs: https://www.theregister.com/2024/11/20/google_ossfuzz/
22. Just because a policy doesn't explicitly ban the use of AI to cheat in an assignment doesn't mean it's not cheating: https://arstechnica.com/tech-policy/2024/11/school-did-nothing-wrong-when-it-punished-student-for-using-ai-court-rules/
23. Even if variables like race are explicitly removed, an AI can still infer them from other factors and act in a biased manner: https://www.theregister.com/2024/11/21/ai_hiring_test_bias/
24. The idea of ad-supported AI gives me all sorts of misgivings: https://techcrunch.com/2024/11/21/marissa-mayer-just-laid-out-a-possible-business-model-for-ad-supported-ai-chatbots/