Weekly Review 29 December 2023

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

1. Why it is important to encounter failure in research: https://www.science.org/content/article/why-i-teach-my-students-about-scientific-failure What kinds of failure can researchers in AI encounter?
2. Art generated by AI, based on stolen sources, is everywhere on Facebook and people are lapping it up: https://www.404media.co/facebook-is-being-overrun-with-stolen-ai-generated-images-that-people-think-are-real/ 
3. There is a lot of opportunity in the New Zealand IT sector, but like a lot of the world, diversity is still lacking: https://www.nzherald.co.nz/kahu/david-downs-nzs-tech-industry-is-worth-over-17b-but-lacks-diversity/O4RRFFSP4ZEAREDO22HXZ5S7JM/
4. The impact generative AI has already made on science: https://www.nature.com/articles/d41586-023-03930-6 
5. It is important to support people returning to research after taking time out to be parents: https://www.science.org/content/article/how-i-returned-academia-after-stepping-away-be-family
6. The six big questions around generative AI: https://www.technologyreview.com/2023/12/19/1084505/generative-ai-artificial-intelligence-bias-jobs-copyright-misinformation/ 
7. All the ways in which AI in general, and generative AI in particular, create liabilities for businesses: https://www.theregister.com/2023/12/21/artificial_intelligence_is_a_liability/
8. How AI can help students transfer between institutions: https://www.insidehighered.com/opinion/blogs/beyond-transfer/2023/12/21/how-can-ai-help-transfer-let-us-count-ways
9. Trends in AI that we will likely see in 2024: https://www.informationweek.com/machine-learning-ai/key-ai-trends-to-look-for-in-2024 
10. Kirin's AI for drink development: https://www.theregister.com/2023/12/19/kirin_ai_drink_development/ Bound to come up with some interesting mixtures.
11. When you shamelessly scrape the entire internet for data to feed your AI, you shouldn't be surprised if that data contains some nasty stuff: https://www.theregister.com/2023/12/20/csam_laion_dataset/ 
12. It seems that like most technologies that get rushed to market, large language model AI have issues with safety and reliability: https://www.datanami.com/2023/12/20/patronus-ai-finds-alarming-safety-gaps-in-leading-llms/ 
13. The large language model AI bubble hasn't popped yet. It's going to be pretty messy when it does: https://www.datanami.com/2023/12/12/genai-hype-bubble-refuses-to-pop/
14. An AI for interior design: https://dataconomy.com/2023/12/21/collov-ai-design-features-and-pricing/ 
15. Data biases in AI have real-world legal consequences for companies that use them: https://www.informationweek.com/machine-learning-ai/ftc-prescribes-ban-on-rite-aid-s-ai-facial-recognition-use 
16. A good AI needs good data. That means capturing data at the edge not from the cloud: https://www.theregister.com/2023/12/08/what_are_you_feeding_your/ 
17. This ruling makes it seem that if an AI attempts to claim to be an inventor, that will fail in a patent application. But if a person uses an AI as a tool to invent something, that will succeed? https://www.computerworld.com/article/3711782/uk-supreme-court-no-your-ai-can-t-hold-a-patent.html 
18. A high schooler's thoughts on the use of ChatGPT and generative AI in education: https://www.technologyreview.com/2023/04/14/1071194/chatgpt-ai-high-school-education-first-person/ 
19. The best way to avoid bias in a generative AI is to tell it not to be biased? https://www.techrepublic.com/article/anthropic-claude-discrimination-bias-mitigation/ 
20. The myths and realities of automating security with AI: https://www.kdnuggets.com/ai-automated-cybersecurity-what-to-automate
21. More thoughts on AI governance and justice: https://www.datasciencecentral.com/ai-and-justice-in-a-brave-new-world-part-3-ai-governance/ 
22. Want an AI to be a better driver? Make it scared. I'm sure there are many other applications of anxious AI as well: https://spectrum.ieee.org/autonomous-vehicle-safety-defensive-driving