Weekly Review 14 March 2025

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

1. An AI that can grade maths problems like a teacher: https://dataconomy.com/2025/03/07/can-ai-grade-math-like-a-teacher-research-says-yes/
2. At least some tech executives recognise the dangers to privacy that AI pose: https://techcrunch.com/2025/03/07/signal-president-meredith-whittaker-calls-out-agentic-ai-as-having-profound-security-and-privacy-issues/
3. Looks like a bit of a messy free for all in terms of legislation on AI in the US: https://techcrunch.com/2025/03/07/us-lawmakers-have-already-introduce-hundreds-of-ai-bills-in-2025/
4. Many years ago we'd use ensembles of simple AI to boost performance. Turns out that using multiple AI in concert is still a way to get better results: https://www.informationweek.com/machine-learning-ai/why-using-multiple-ais-is-trending-now
5. AI summaries of app store reviews. I feel that this will not end well: https://dataconomy.com/2025/03/06/ai-is-now-summarizing-apple-app-store-reviews/
6. If it's human artists using AI as a tool, I don't think it's a problem, but purely AI generated art shouldn't really be worth much at all: https://techcrunch.com/2025/03/06/christies-ai-art-auction-reportedly-exceeds-expectations/
7. Some tips for getting the most out of Copilot: https://www.computerworld.com/article/1611598/microsoft-copilot-tips-how-to-use-copilot-right.html
8. Why an AI arms race is a bad idea: https://theregister.com/2025/03/06/schmidt_ai_superintelligence/
9. AI is just a tool, it will soon be useful as an aid to scientists, but won't replace scientists themselves: https://techcrunch.com/2025/03/05/experts-dont-think-ai-is-ready-to-be-a-co-scientist/
10. Fear of AI is already leading to violence: https://www.theguardian.com/global/ng-interactive/2025/mar/05/zizians-artificial-intelligence
11. An AI to assist physicians. But how well does it guard patient privacy? https://dataconomy.com/2025/03/04/microsoft-new-ai-assistant-aims-to-fix-doctor-burnout/
12. Don't use AI to replace connections with real people: https://techcrunch.com/2025/03/07/health-expert-warns-of-leaning-too-heavily-on-ai-for-social-connections/
13. Now propagandists are poisoning AI training data: https://techcrunch.com/2025/03/07/russian-propoganda-is-reportely-influencing-ai-chatbot-results/
14. AI don't really innovate, so the fear that they'll become software yes-men is well founded: https://techcrunch.com/2025/03/06/hugging-faces-chief-science-officer-worries-ai-is-becoming-yes-men-on-servers/
15. How AI can be applied to project management: https://dataconomy.com/2025/03/06/artificial-intelligence-in-product-management-how-al-eases-the-life-of-a-product-manager-tools-overview-and-personal-experience/
16. An overview of agentic AI: https://www.kdnuggets.com/what-data-scientists-need-to-know-about-ai-agents-and-autonomous-systems
17. The problems with AI transparency, especially in healthcare: https://www.extremetech.com/science/the-paradox-of-trust-and-transparency-in-ai
18. Avoiding mistakes when building an AI governance platform: https://www.informationweek.com/machine-learning-ai/how-to-build-a-reliable-ai-governance-platform
19. Turns out certain advanced AI can also get stuck in the trap of overthinking: https://spectrum.ieee.org/reasoning-in-ai Next they'll be getting procrastination anxiety
20. Another tech industry giant warns about the dangers of AI: https://dataconomy.com/2025/03/04/steve-wozniak-ai-warning-it-has-a-dark-side/
21. A PhD-level AI isn't that impressive to me, I know lots of people with PhDs and some of them aren't that smart: https://arstechnica.com/ai/2025/03/what-does-phd-level-ai-mean-openais-rumored-20000-agent-plan-explained/
22. It's not all bad news regarding AI in education, a lot of applications are really helpful: https://www.theguardian.com/education/2025/mar/06/the-english-schools-looking-to-dispel-doom-and-gloom-around-ai
23. The BBC intends to use AI to better serve its readers and viewers: https://www.theguardian.com/media/2025/mar/06/bbc-news-ai-artificial-intelligence-department-personalised-content
24. Using AI to process a vast (1M+) number of data sources in real time: https://www.bigdatawire.com/2025/02/27/the-ai-firm-turning-1m-real-time-data-sources-into-actionable-intelligence/