Weekly Review 5 July 2024

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

1. At this point it would be more effective to replace managers with AI, rather than frontline fast food workers: https://futurism.com/the-byte/wendys-ai-drive-thru-orders
2. Three mistakes to avoid when analysing data with large language model AI: https://www.datanami.com/2024/06/25/top-three-pitfalls-to-avoid-when-processing-data-with-llms/
3. California seems to be taking a leading role in AI regulation: https://www.computerworld.com/article/2503219/ai-regulation-while-congress-fiddles-california-gets-it-done.html
4. I'm not sure that using AI to monitor the truthfulness of another AI is a good idea: https://spectrum.ieee.org/openai-rlhf
5. Exam answers generated by AI are now not only undetectable but also get better marks: https://www.theguardian.com/education/article/2024/jun/26/researchers-fool-university-markers-with-ai-generated-exam-papers
6. Some ways to make AI more energy efficient: https://www.theregister.com/2024/06/26/ai_model_fpga/
7. AI do not have any conception of what is true or false: https://futurism.com/perplexity-webpage-mushrooms
8. The steps to a career in AI: https://www.kdnuggets.com/building-a-career-in-ai-from-student-to-professional
9. I expect Meta is using an AI to detect if an image is generated by AI, and so it's churning out false positives: https://futurism.com/the-byte/facebook-real-photos-ai
10. Zuckerberg claims that closed source AI is inferior. How open is Meta with their models, though? https://techcrunch.com/2024/06/27/zuckerberg-disses-closed-source-ai-competitors-as-trying-to-create-god/
11. It appears that a lot (most?) of the data used to train AI was scraped by bots that ignored the robots.txt file: https://www.datasciencecentral.com/computing-where-should-ai-safety-for-superintelligence-and-agi-start/
12. AI is using more and more energy. I disagree that renewables are the best solution for this, nuclear is more stable and reliable: https://www.theregister.com/2024/06/26/datacenter_electricity_demand_study/
13. There's already a shortage of women in IT, changes that further reduce the pipeline should be avoided: https://www.theguardian.com/technology/article/2024/jun/27/number-of-girls-in-england-taking-computing-gcse-plummets-study-finds
14. When a generative AI like ChatGPT doesn't know something, it makes an answer up, a process called confabulation. This can now be detected: https://arstechnica.com/ai/2024/06/researchers-describe-how-to-tell-if-chatgpt-is-confabulating/
15. The problem with testing AI generated code with an AI is that neither the generator nor tester have any clue about what the code actually means: https://devclass.com/2024/06/25/can-you-trust-ai-to-test-ai-generated-code-devclass-talks-ai-coding-with-tabnine/
16. Is ChatGPT's voice chatbot badly misbehaving? https://futurism.com/the-byte/openai-chatgpt-voice-mode-trouble
17. So Meta is rolling out AI chatbots, and is labelling them as such: https://techcrunch.com/2024/06/27/meta-starts-testing-user-created-ai-chatbots-on-instagram/
18. How AI is changing the business world: https://bigthink.com/business/stanford-report-how-ai-is-actually-transforming-the-business-world/
19. An influencer's AI clone went rogue: https://futurism.com/the-byte/influencer-ai-clone-dark-fantasies
20. An essay on why keeping the output of AI weird is a good thing: https://qaswa.com/keep-ai-weird
21. It's clear that the immediate winners of the generative AI boom will be the lawyers: https://www.technologyreview.com/2024/06/27/1094379/ai-music-suno-udio-lawsuit-record-labels-youtube-licensing/
22. How quantum computing can make AI more energy efficient: https://www.computerworld.com/article/2497022/ai-is-starving-for-more-power-can-quantum-computing-help.html
23. Well that's not concerning. Is the giant robot at least a Timberwolf/Madcat? https://futurism.com/anthropic-ceo-dario-amodei-giant-robot
24. Six ways that AI can help with mental health treatment: https://dataconomy.com/2024/06/27/the-impact-of-artificial-intelligence-on-mental-health-interventions/