Weekly Review 19 January 2024

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

1. Reading a neural network's thoughts can be used to add an extra layer of safety to the operations of AI: https://www.science.org/content/article/artificial-intelligence-may-benefit-talking-itself 
2. Why open source AI is less secure, and what we can do about it: https://spectrum.ieee.org/open-source-ai-2666932122 
3. Some predictions about what is going to be happening with AI in 2024: https://www.technologyreview.com/2024/01/04/1086046/whats-next-for-ai-in-2024/ 
4. Of course AI can't replace human leaders and managers. Management, and leadership, is about people, not processes. It requires understanding and empathy, and so far only people can do that: https://hbr.org/2024/01/the-best-leaders-cant-be-replaced-by-ai 
5. Isn't this just the AI version of people copy-pasting the output of translation software, without understanding what it says? https://arstechnica.com/ai/2024/01/lazy-use-of-ai-leads-to-amazon-products-called-i-cannot-fulfill-that-request/ 
6. How businesses can take advantage of generative AI: https://hbr.org/2024/01/turn-generative-ai-from-an-existential-threat-into-a-competitive-advantage 
7. A more power-efficient chip for AI: https://spectrum.ieee.org/neuchips-low-power-ai 
8. Synthetic data can be used to overcome problems with bias and the cost of gathering data. But if it's not done carefully, biases can become even more entrenched: https://www.quantamagazine.org/neural-networks-need-data-to-learn-even-if-its-fake-20230616/ 
9. It is very dangerous for Wall Street to rely too heavily on AI: https://www.datanami.com/2024/01/12/consumer-watchdog-report-wall-street-ai-could-cause-financial-crisis/ 
10. Deep neural networks and other highly-parameterised AI models aren't necessary for many applications, a simpler model can work just as well: https://www.kdnuggets.com/are-we-undervaluing-simple-models 
11. The business forces driving the adoption of generative AI: https://hbr.org/2023/12/5-forces-that-will-drive-the-adoption-of-genai 
12. Using machine learning to construct chemical structures from X-ray images: https://spectrum.ieee.org/synchrotron-x-ray 
13. "A collision's a near miss! "Look, they nearly missed." "Yes, but not quite!"" George Carlin was a brilliant comic, I find it quite ghoulish that they've resurrected him with AI: https://dataconomy.com/2024/01/12/george-carlin-ai-criticism-daughter/
14. I don't agree that a large language model must be trained using copyright material. The more salient question is whether using that material is transformative and/or fair use: https://www.extremetech.com/internet/openai-claims-its-impossible-to-train-ai-models-without-copyrighted-materials 
15. Let me know when an AI robot can do something that a three-year-old couldn't do for the price of a lollipop: https://techcrunch.com/2024/01/15/elons-tesla-robot-is-sort-of-ok-at-folding-laundry-in-pre-scripted-demo/ 
16. Three ways to spot if an online article has been written by AI: https://gizmodo.com/how-to-spot-ai-written-content-1851106624 
17. The Australian government is also looking at restrictions on AI research: https://www.computerworld.com/article/3712166/australia-evaluates-compulsory-guardrails-to-ensure-safer-ai.html 
18. It looks like "AI washing" is still going on at breakneck pace: https://www.theverge.com/2024/1/13/24035152/ces-generative-ai-hype-robots 
19. Using neural networks to relate prints from one finger to prints on other fingers. But the whole thing seems to be based on an invalid assumption. AI is a set of algorithms, not magic: https://www.science.org/content/article/do-prints-two-different-fingers-belong-same-person-ai-can-tell 
20. Automating protein engineering with AI. I'm pretty sure the William Gibson short story New Rose Hotel involved something like that: https://www.nature.com/articles/d41586-024-00093-w 
21. Makes it sound like big tech firms haven't always recklessly pursued profits. How is it different if that profit comes from AI? https://www.theguardian.com/business/2024/jan/17/big-tech-firms-ai-un-antonio-guterres-davos 
22. When people think that a serial fraudster, sex offender, wannabe dictator is a good candidate for President of the USA, disinformation from AI is the least of our political worries: https://www.theregister.com/2024/01/17/ai_political_disinformation/
23. Should using generative AI be a part of your daily routine? Personally, I like working things out for myself: https://www.insidehighered.com/opinion/blogs/online-trending-now/2024/01/17/integrating-generative-ai-daily-work-and-personal-life 
24. Evaluating the risk of developing pancreatic cancer using AI: https://www.technologyreview.com/2024/01/17/1086730/a-new-ai-based-risk-prediction-system-could-help-catch-deadly-pancreatic-cancer-cases-earlier/