Some interesting links that I Tweeted about in the last week (I also post these on Mastodon, Threads, Newsmast, and Bluesky):
- You can't trust an AI shopping agent to not screw things up: https://futurism.com/artificial-intelligence/target-ai-agent-tos
- Concepts you need to understand to use LLM AI effectively: https://www.kdnuggets.com/10-llm-engineering-concepts-explained-in-10-minutes
- Using AI means designers can evaluate designs orders of magnitude faster than before: https://spectrum.ieee.org/large-physics-models-design-engineering
- I understand that doing a literature review is time-consuming and dull, but using AI to do it for you just means you end up with hallucinated references: https://www.nature.com/articles/d41586-026-00969-z
- China regulates to prevent harm caused by AI: https://futurism.com/artificial-intelligence/china-usa-ai-regulations
- Instead of building more and larger data centres to train AI, distribute the training instead: https://spectrum.ieee.org/decentralized-ai-training-2676670858
- Adapting your hiring processes to include AI: https://dataconomy.com/2026/04/06/enhancing-recruitment-processes-with-ai-strategies-for-modern-hiring-teams/
- Using AI to model and manage water flow in the Colorado river: https://spectrum.ieee.org/colorado-river-water-shortage
- Google's AI overviews are not reliable: https://arstechnica.com/google/2026/04/analysis-finds-google-ai-overviews-is-wrong-10-percent-of-the-time/
- Businesses plan to keep spending money on AI, even if it's not generating real returns: https://www.theregister.com/2026/04/10/ai_roi_kpmg/
- AI have a lot of security risks of their own, but they are also becoming powerful tools for finding security flaws: https://www.theguardian.com/technology/2026/apr/10/us-summoned-bank-bosses-to-discuss-cyber-risks-posed-by-anthropic-latest-ai-model
- Turns out AI really will destroy a lot of jobs: https://futurism.com/future-society/economist-ai-job-forecast
- If you're going to use AI tools in a clinical setting, informed consent of the patients is really important: https://arstechnica.com/tech-policy/2026/04/californians-sue-over-ai-tool-that-records-doctor-visits/
- AI tend to break in ways that engineers aren't used to or trained to recognise: https://spectrum.ieee.org/ai-reliability
- The inaccuracies of Google's AI overviews is creating a crisis of misinformation: https://futurism.com/artificial-intelligence/google-ai-overviews-misinformation
- If you are integrating AI into your business, build as if you are starting from scratch: https://www.theregister.com/2026/04/07/aws_garman_humanx_ai_underhyped/
- By this point nothing said by the CEO of an AI company should be trusted or taken seriously: https://arstechnica.com/tech-policy/2026/04/what-the-heck-is-wrong-with-our-ai-overlords/
- Spending on AI data centres is projected to hit $7T. Does anyone honestly think that that is sustainable? https: https://www.hpcwire.com/bigdatawire/2026/04/07/ai-is-running-into-a-7-trillion-wall/
- Only a minority of AI projects pay off: https://www.theregister.com/2026/04/07/ai_returns_gartner/
- The job market for older workers is so dire that people are having to train their own AI replacements just to make ends meet: https://www.theguardian.com/technology/ng-interactive/2026/apr/07/ai-training-work-jobs
- The challenges of moving AI data centres into space: https://www.hpcwire.com/bigdatawire/2026/04/10/could-space-become-the-next-frontier-for-ai-data-centers/
- AI has taken over most of the operations in currency trading: https://dataconomy.com/2026/04/10/ai-powered-trading-bots-and-the-evolution-of-forex-automation/
- AI trained on patient X-rays will happily hallucinate a diagnosis even if not shown an X-ray: https://futurism.com/artificial-intelligence/frontier-models-medical-advice-x-rays-cant-see
- AI still have a while to go before they can organise a good party: https://www.theguardian.com/technology/2026/apr/05/ai-bot-party-manchester-gaskell
- Jobseekers now need to know how to get through AI interviews: https://www.stuff.co.nz/nz-news/360961113/company-interviewed-more-1000-kiwis-ai-month
- Improving efficiency with AI requires effective leadership: https://www.hpcwire.com/bigdatawire/2026/04/06/the-ai-productivity-opportunity-bridging-the-technology-divide-starting-with-your-leadership/
