Gimcheon City in South Korea launched an on-site AI training program for residents on November 6, teaching practical skills like using ChatGPT and voice search on smartphones. The sessions, held alongside neighborhood head meetings, aim to expand AI literacy through the K-High Tech Platform, with a second session planned for November 24 in Yulgok-dong.
Meanwhile, a survey from the Archbridge Institute published November 7 found that 56% of Americans believe AI will undermine meaningful living, while 44% think it will support it. People who view AI as a human-controlled tool tend to be more optimistic, whereas concerns about autonomous AI are especially high among women and older adults.
In the US, voters rejected AI data center construction proposals during midterm elections. Only one ballot measure approved a data center project out of more than 40 local areas. Residents opposed the facilities over concerns about electricity and water use, noise, and limited economic returns. Democrats taking control of Congress and new governors in Texas and Georgia calling for moratoriums add political headwinds to the AI building boom.
Apple confirmed that deeper Siri app control will not arrive until Spring 2027, likely landing in iOS 27.5 or iOS 27.6. Third-party app support for Siri is also planned for a preview in spring 2027 as Apple works toward a more agentic assistant that can handle tasks with less oversight.
Tech companies including Apple, Google, Meta, and Samsung are investing heavily in AI hardware beyond smartphones, such as smart glasses, wearables, AI pins, and pendants. These devices aim to make digital assistance more natural through voice commands, but success depends on addressing battery life, privacy, and affordability challenges.
On the technical front, a homelabber built a system using second-generation Optane memory sticks to run a 2.8-trillion-parameter AI model on a CPU alone. The setup runs hot and requires blower fans, with plans to add GPUs and use DRAM as a faster data tier in the future.
Research also shows AI coding agents still need human help on nearly three out of four tasks. The Co-Design Bench, developed by Alibaba and Asian university researchers, found that even the best models fully complete only about one in four real-world coding requests, especially when multiple types of changes are combined.
Key Takeaways
- Gimcheon City, South Korea launched on-site AI training for residents, teaching ChatGPT and smartphone voice search skills starting November 6.
- Archbridge Institute survey finds 56% of Americans believe AI will undermine meaningful living, while human-controlled AI views generate more optimism.
- US voters rejected AI data center proposals in midterm elections, with only one measure approved across more than 40 local ballot areas.
- Apple delays deeper Siri app control until Spring 2027, with third-party app support also planned for a spring preview.
- Apple, Google, Meta, and Samsung are investing in AI hardware beyond smartphones, including smart glasses, wearables, and AI pins.
- A homelabber ran a 2.8-trillion-parameter AI model on a CPU using second-generation Optane memory sticks.
- AI coding agents fully complete only about 25% of real-world tasks, according to the Co-Design Bench from Alibaba and Asian universities.
- Democrats taking control of Congress and moratorium calls from Texas and Georgia governors create political challenges for AI data center expansion.
- Library series in Solana Beach explores AI ethics, covering transparency, accountability, and bias in AI development.
- GM patented automated launch control for drag racing that uses car cameras to recognize Christmas tree lights.
Gimcheon City launches on-site AI training for residents
Gimcheon City in South Korea started a pilot AI training program on November 6. Residents learned practical skills like using ChatGPT and voice searches on smartphones. The training was held alongside neighborhood head meetings. A second session is planned for November 24 in Yulgok-dong. The city aims to expand on-site AI training to more areas through the K-High Tech Platform.
Homelabber runs massive AI model using old Optane memory
A homelab builder created a system using second-generation Optane memory sticks to run a 2.8-trillion-parameter AI model on a CPU alone. The build uses Memory Mode and App Direct to manage where data is stored. The system runs hot and requires blower fans to prevent nearby RAM from melting. The builder plans to add GPUs and use DRAM as a faster data tier in the future.
Voters reject AI data center construction in midterms
Voters across the United States rejected proposals for AI data center construction during midterm elections. Only one ballot approved a data center project nationwide. Residents opposed the centers because they use large amounts of electricity and water, create noise, and offer limited economic benefits. NBC's Allie Canal reported on the growing pushback against these facilities.
Survey finds AI views depend on how much control humans keep
A survey from the Archbridge Institute found that 56% of Americans believe AI will undermine meaningful living while 44% think it will support it. People who see AI as a human-controlled tool are more optimistic than those who expect it to become autonomous. Women and older adults were especially concerned about AI threatening life meaning. The survey was published on November 7, 2026.
Apple delays deeper Siri app control until Spring 2027
Apple says Siri will not gain more control over Apple apps until Spring 2027. The feature will likely arrive as part of iOS 27.5 or iOS 27.6, ahead of the 2027 WWDC. Third-party app support is also planned for a preview in spring 2027. Apple is working toward a more agentic Siri that can help users with tasks without constant oversight.
US voters reject AI data centres in midterm elections
Voters in the United States rejected AI data centres in Tuesday's midterm elections. The measures were on ballots in more than 40 local areas. With Democrats taking control of Congress and new governors in Texas and Georgia calling for moratoriums, the AI building boom faces political challenges. The article questions why similar backlash has not occurred in France, where cheap nuclear power and fast-track laws support data centre growth.
GM patents automated launch control for drag racers
General Motors has patented automated launch control technology for drag racing. The system uses car cameras to recognize Christmas tree lights at drag strips. Once activated, the car takes over acceleration without driver input. The technology aims for precise timing but has drawn criticism from racing enthusiasts who value human skill. The patent describes a system that initiates launch processes before traffic lights turn green.
Library series explores AI moral dilemmas and ethics
The Friends of the Solana Beach Library hosts a four-part series on AI ethics. The second session features Shelly Skinner from Salience AI and AI engineer Tulay Muezzinoglu. They discuss transparency, accountability, and bias in AI development. The talk raises questions about whether just because something can be done, it should be. The series continues with topics on AI regulation and education in 2027.
Tech companies build AI devices beyond smartphones
Technology companies are developing AI hardware that operates outside of smartphones. The products include smart glasses, wearables, AI pins, and pendants. These devices aim to make digital assistance more natural through voice commands. Major companies like Apple, Google, Meta, and Samsung are investing in this hardware. The success of these products depends on solving issues with battery life, privacy, and affordability.
AI coding agents need human help on most tasks
A real-world test found that AI coding agents need human cleanup on nearly three out of four tasks. Researchers from Alibaba and Asian universities built the Co-Design Bench to test this. The benchmark uses real user requests that require rephrasing or rolling back changes. Even the best models fully complete only about one in four tasks. The test shows that combining multiple types of changes overwhelms the AI models.
Sources
- Gimcheon City Strengthens Residents' Digital Skills with On-Site AI Training
- Homelabber uses second-gen Optane memory sticks to run a 2.8-trillion-parameter AI model on CPU alone — warns that Optane's extreme heat can 'burn out' nearby memory DRAM, used blower fans to prevent melting RAM
- What Do the Midterm Results Mean for Future of AI Data Centers?
- Will AI make life less meaningful? A survey finds optimism hinges on whether humans keep control.
- Siri AI won't get more control of Apple apps until Spring 2027
- Tech 24 - US voters reject AI data centres: could the backlash spread to France?
- AI Comes For Drag Racers' Jobs As GM Patents Automated Launch Control
- ‘Let’s Talk’ series to hold session 2: ‘AI’s Moral Dilemma: AI and Ethics’
- The Future of AI Hardware: Why Tech Companies are Building Devices Beyond Smartphones
- AI coding agents still need human cleanup on nearly three out of four tasks in a real-world test
Comments
Please log in to post a comment.