Weekly Review 8 December 2023

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

1. More cuts in New Zealand universities, and this new right-wing Frankenstein's Monster of a government isn't going to do a damn thing to help: https://www.stuff.co.nz/national/education/301017113/massey-university-entering-academic-equivalent-of-an-ice-age 
2. A better way of developing AI? Fighting against the biases and ethical issues in standard data sets: https://spectrum.ieee.org/joy-buolamwini 
3. Seven different points of view on how ChatGPT and generative AI has impacted science and science education: https://www.nature.com/articles/d41586-023-03798-6
4. AI written articles attributed to authors with AI generated head shots. When will they come up with an AI editor that fixes the egregious errors in the generated articles? https://futurism.com/sports-illustrated-ai-generated-writers 
5. I would not trust AI with such life-and-death decisions as generating target lists: https://www.theguardian.com/world/2023/dec/01/the-gospel-how-israel-uses-ai-to-select-bombing-targets 
6. An AI model claims that volcanic eruptions - the formation of the Deccan Traps - was enough to doom the dinosaurs, with the Chicxulub meteorite not being necessary: https://www.extremetech.com/science/ai-model-decides-volcanoes-not-an-asteroid-killed-the-dinosaurs 
7. One of the better explanations of local optima I've come across, and a problem that everyone who works with AI needs to be wary of: https://poisonedminds.com/d/20230816.html
8. The Turing test is not very useful for determining whether an artificial intelligence is actually intelligent. Dolphins and elephants are undeniably intelligent, and self-aware, but they would all fail the Turing test: https://spectrum.ieee.org/turing-test
9. On how Deep Learning and other AI can be used to segment and recognise cells in microscope images: https://www.nature.com/articles/d41586-023-03722-y 
10. Artificial intelligence cannot replace the real world: https://www.thepress.co.nz/a/nz-news/350120085/why-i-think-ai-may-change-real-world-will-never-replace-it
11. As a certain young law student is fond of telling me, international law is basically a bunch of pinky promises. The regulations around AI are just the same: https://www.informationweek.com/cyber-resilience/new-secure-ai-development-rules-are-historic-but-do-they-matter- 
12. Focused, single-purpose AI models are still relevant: https://techcrunch.com/2023/12/01/good-old-fashioned-ai-remains-viable-in-spite-of-the-rise-of-llms/ 
13. I'm quite certain that there will be executives in Hollywood who think that using AI to do everything with movies is a great way to go: https://poisonedminds.com/d/20230515.html
14. The first (local) law written by ChatGPT has been passed. What's the quote from Frank Herbert about men with machines making other men slaves?https://www.theregister.com/2023/12/02/chatgpt_law_brazil/
15. Five major issues with, and proposed fixes for, generative AI: https://www.datasciencecentral.com/generative-ai-five-major-issues-and-how-to-fix-them/ 
16. I know pika are animals, but when I read about this I think of the disorder pica, not an AI video production system: https://dataconomy.com/2023/11/29/pika-ai-video-tool-pika-art-pika-1-0/ 
17. More research on the energy usage, and carbon footprint, of generative AI models: https://www.technologyreview.com/2023/12/01/1084189/making-an-image-with-generative-ai-uses-as-much-energy-as-charging-your-phone/ 
18. I teach A* to my first year algorithms students, and Q-learning to the second years. I an skeptical that combining them is going to produce meaningful breakthroughs in AI: https://www.computerworld.com/article/3711369/the-fear-and-hype-around-ai-is-overblown.html 
19. How can we tell that an artificial intelligence is self-aware? We think it thinks, therefore it probably is? https://poisonedminds.com/d/20230526.html
20. Modelling near-ground air temperature from ground temperate, as measured by satellite infrared cameras: https://spectrum.ieee.org/satellite-weather-map 
21. A guide to choosing a large language model AI: https://www.datanami.com/2023/11/30/data-and-algorithms-the-building-blocks-of-artificial-intelligence/ 
22. AI text summarisation tools. No, I don't use these to compose my posts: https://dataconomy.com/2023/11/29/10-best-summary-generators-to-summarize-any-text-online/ 
23. How AI will impact supply chains and procurement management: https://www.technologyreview.com/2023/11/28/1083628/procurement-in-the-age-of-ai/ 
24. Using an AI to posit new inorganic molecules, and a robot to make them: https://www.nature.com/articles/d41586-023-03745-5
25. An attack technique to get ChatGPT to expose its training data: https://www.404media.co/google-researchers-attack-convinces-chatgpt-to-reveal-its-training-data/ 
26. I don't think it's remote work that has led to fewer transformational papers, so much as the issues with the peer-review process: https://www.stuff.co.nz/science/301018006/remote-working-is-stifling-the-eureka-moment-in-science
27. Some limits and the way forward for generative AI: https://innovate.ieee.org/innovation-spotlight/a-closer-look-at-generative-ai/ 
28. Some ways in which artificial intelligence will impact scientific research: https://www.technologyreview.com/2023/07/05/1075865/eric-schmidt-ai-will-transform-science/