![Sony WH-1000XM6 Unveiled With Smarter Noise Canceling and Studio-Tuned Sound [Video]](https://www.iclarified.com/images/news/97341/97341/97341-640.jpg)
![Trump Tells Cook to Stop Building iPhones in India and Build in the U.S. Instead [Video]](https://www.iclarified.com/images/news/97329/97329/97329-640.jpg)
![[The AI Show Episode 147]: OpenAI Abandons For-Profit Plan, AI College Cheating Epidemic, Apple Says AI Will Replace Search Engines & HubSpot’s AI-First Scorecard](https://www.marketingaiinstitute.com/hubfs/ep%20147%20cover.png)
![How to Enable Remote Access on Windows 10 [Allow RDP]](https://bigdataanalyticsnews.com/wp-content/uploads/2025/05/remote-access-windows.jpg)
![Artist Shocked To Find Her Poster Designs From 2017 In Bungie's Marathon: 'A Major Company Has Deemed It Easier To Pay A Designer To Imitate Or Steal My Work Than To Write Me An Email' [Update]](https://i.kinja-img.com/image/upload/c_fill,h_675,pg_1,q_80,w_1200/4ce7afff77473c3cccca9cc349c42790.jpg)
Weekly Review 16 May 2025
Some interesting links that I Tweeted about in the last week (I also post these on Mastodon, Threads, Newsmast, and Bluesky):The US tax service wants to use AI for collections enforcement. This is not going to end well: https://www.theregister.com/2025/05/08/the_irs_plans_to_replace/Not content with crashing half the planet in 2024, CrowStrike now wants to lean even more on AI: https://www.theregister.com/2025/05/07/crowdstrike_trims_workforce_ai/The major problem facing organisations wanting to roll out generative AI is a lack of a skilled workforce: https://www.bigdatawire.com/2025/05/06/these-are-the-top-chadllenges-to-genai-adoption-according-to-aws/Google's AI tools that protect against scammers: https://techcrunch.com/2025/05/08/google-rolls-out-ai-tools-to-protect-chrome-users-against-scams/It is not safe to rely on AI chatbots for therapy: https://www.theguardian.com/technology/2025/may/07/experts-warn-therapy-ai-chatbots-are-not-safe-to-useUniversity instructors want to use AI, but want more guidance and support in doing so: https://www.insidehighered.com/news/student-success/academic-life/2025/05/08/professors-using-ai-need-more-administrative-supportAI generated vulnerability reports are overwhelming open source projects: https://arstechnica.com/gadgets/2025/05/open-source-project-curl-is-sick-of-users-submitting-ai-slop-vulnerabilities/AI companies want to use AI to evaluate therapeutic drugs: https://dataconomy.com/2025/05/08/openai-and-xai-talking-ai-drug-evaluation-with-fda/Three free courses on generative AI: https://www.kdnuggets.com/3-excellent-practical-generative-ai-coursesAmazon's AI tool to improve product listings. I'm not sure "improve" is the right word here: https://techcrunch.com/2025/05/08/amazons-newest-ai-tool-is-designed-to-enhance-product-listings/A sci-fi conference used AI to screen potential panellists. This has upset a lot of people: https://www.theregister.com/2025/05/07/worldcon_uses_ai/Chain of Thought Reasoning to improve the performance of AI: https://spectrum.ieee.org/chain-of-thought-promptingAI are not neutral third parties, they will be used to manipulate people: https://www.theregister.com/2025/05/06/schneier_ai_models/Do not rely on AI chatbots for health advice: https://techcrunch.com/2025/05/05/people-struggle-to-get-useful-health-advice-from-chatbots-study-finds/AI training causes spikes in power demand. Ways of hardening the grid against these spikes are needed: https://spectrum.ieee.org/supercapacitor-2671883490If you want an AI that doesn't need prompt engineering, start with the data: https://www.datasciencecentral.com/how-to-design-llms-that-dont-need-prompt-engineering/People who use AI more in their work have poorer professional reputations: https://arstechnica.com/ai/2025/05/ai-use-damages-professional-reputation-study-suggests/Securing AI applications: https://www.kdnuggets.com/securing-machine-learning-application-with-authentication-and-user-managementAsking an AI to be more concise makes it more likely to hallucinate: https://techcrunch.com/2025/05/08/asking-chatbots-for-short-answers-can-increase-hallucinations-study-finds/It is better to be more discerning about which data you use to build an AI, rather than just shovelling everything in: https://www.bigdatawire.com/2025/05/07/how-to-build-a-lean-ai-strategy-with-data/Why current cloud governance practices are hindering the growth of AI: https://www.informationweek.com/it-infrastructure/cloud-governance-in-the-age-of-aiOpen source AI models discriminate against women when used for recruitment: https://www.theregister.com/2025/05/02/open_source_ai_models_gender_bias/LinkedIn's AI powered job search tool: https://www.theverge.com/news/662490/linkedin-ai-job-search-tool-availabilitySome AI companies will certainly fail. But those companies that provide useful services will survive and thrive: https://www.computerworld.com/article/3980239/will-genai-businesses-crash-and-burn.html
Some interesting links that I Tweeted about in the last week (I also post these on Mastodon, Threads, Newsmast, and Bluesky):
- The US tax service wants to use AI for collections enforcement. This is not going to end well: https://www.theregister.com/2025/05/08/the_irs_plans_to_replace/
- Not content with crashing half the planet in 2024, CrowStrike now wants to lean even more on AI: https://www.theregister.com/2025/05/07/crowdstrike_trims_workforce_ai/
- The major problem facing organisations wanting to roll out generative AI is a lack of a skilled workforce: https://www.bigdatawire.com/2025/05/06/these-are-the-top-chadllenges-to-genai-adoption-according-to-aws/
- Google's AI tools that protect against scammers: https://techcrunch.com/2025/05/08/google-rolls-out-ai-tools-to-protect-chrome-users-against-scams/
- It is not safe to rely on AI chatbots for therapy: https://www.theguardian.com/technology/2025/may/07/experts-warn-therapy-ai-chatbots-are-not-safe-to-use
- University instructors want to use AI, but want more guidance and support in doing so: https://www.insidehighered.com/news/student-success/academic-life/2025/05/08/professors-using-ai-need-more-administrative-support
- AI generated vulnerability reports are overwhelming open source projects: https://arstechnica.com/gadgets/2025/05/open-source-project-curl-is-sick-of-users-submitting-ai-slop-vulnerabilities/
- AI companies want to use AI to evaluate therapeutic drugs: https://dataconomy.com/2025/05/08/openai-and-xai-talking-ai-drug-evaluation-with-fda/
- Three free courses on generative AI: https://www.kdnuggets.com/3-excellent-practical-generative-ai-courses
- Amazon's AI tool to improve product listings. I'm not sure "improve" is the right word here: https://techcrunch.com/2025/05/08/amazons-newest-ai-tool-is-designed-to-enhance-product-listings/
- A sci-fi conference used AI to screen potential panellists. This has upset a lot of people: https://www.theregister.com/2025/05/07/worldcon_uses_ai/
- Chain of Thought Reasoning to improve the performance of AI: https://spectrum.ieee.org/chain-of-thought-prompting
- AI are not neutral third parties, they will be used to manipulate people: https://www.theregister.com/2025/05/06/schneier_ai_models/
- Do not rely on AI chatbots for health advice: https://techcrunch.com/2025/05/05/people-struggle-to-get-useful-health-advice-from-chatbots-study-finds/
- AI training causes spikes in power demand. Ways of hardening the grid against these spikes are needed: https://spectrum.ieee.org/supercapacitor-2671883490
- If you want an AI that doesn't need prompt engineering, start with the data: https://www.datasciencecentral.com/how-to-design-llms-that-dont-need-prompt-engineering/
- People who use AI more in their work have poorer professional reputations: https://arstechnica.com/ai/2025/05/ai-use-damages-professional-reputation-study-suggests/
- Securing AI applications: https://www.kdnuggets.com/securing-machine-learning-application-with-authentication-and-user-management
- Asking an AI to be more concise makes it more likely to hallucinate: https://techcrunch.com/2025/05/08/asking-chatbots-for-short-answers-can-increase-hallucinations-study-finds/
- It is better to be more discerning about which data you use to build an AI, rather than just shovelling everything in: https://www.bigdatawire.com/2025/05/07/how-to-build-a-lean-ai-strategy-with-data/
- Why current cloud governance practices are hindering the growth of AI: https://www.informationweek.com/it-infrastructure/cloud-governance-in-the-age-of-ai
- Open source AI models discriminate against women when used for recruitment: https://www.theregister.com/2025/05/02/open_source_ai_models_gender_bias/
- LinkedIn's AI powered job search tool: https://www.theverge.com/news/662490/linkedin-ai-job-search-tool-availability
- Some AI companies will certainly fail. But those companies that provide useful services will survive and thrive: https://www.computerworld.com/article/3980239/will-genai-businesses-crash-and-burn.html
