Jeff Bezos is sharing his long-term vision for both AI and space. In a recent interview, he said the U.S. must beat China to land humans on the moon by 2028 and build a sustainable lunar base. Bezos has put $28 billion into Blue Origin and believes AI could significantly boost productivity, possibly leading some Americans to work fewer days. Blue Origin plans to return its New Glenn rocket to flight in December, carrying the Blue Ring payload vehicle.
On the AI research front, NVIDIA teamed up with Princeton and the University of Maryland to develop PivotOPD, a method that helps multi-turn AI agents recover from early errors. The approach outperformed 13 baselines on benchmarks like ALFWorld and WebShop using Qwen3-1.7B and Qwen3-8B models. Separately, JetBrains launched Mellum2.1, an open 12B mixture-of-experts coding model available on Hugging Face under the Apache 2.0 license. It activates 2.5B parameters per token and can explore code repositories, edit files, and verify its own changes.
Meanwhile, practical concerns about AI are surfacing in everyday life and local government. Michael Geisler of Ace Handyman Services warns that AI cannot physically assess a home and may lead homeowners into projects they cannot finish. San Joaquin County paused new data center approvals for 45 days, citing gaps in rules around noise, energy use, and emergency response. The Rapid City Council also delayed a proposed AI policy after critics called its principles too vague.
Public opinion on AI regulation is shifting too. A global survey found strong support for an independent authority to regulate AI, especially when job losses occur without new job creation. Political ideology shapes these views. In the workforce, investors and employers are adapting as well. Matthew Lilling of ClearBridge says the AI boom makes it harder to separate quality companies from cyclical ones. At the College of Charleston, instructor Anurag Tiwari turns student AI projects into job assets, with one student landing work at the Naval Information Warfare Center after building a deployable card-game rules judge.
Key Takeaways
- Jeff Bezos has invested $28 billion in Blue Origin and wants humans on the moon by 2028
- NVIDIA's PivotOPD helps AI agents recover from early mistakes, outperforming 13 baselines
- JetBrains released Mellum2.1, an open 12B coding model on Hugging Face
- San Joaquin County paused new data center approvals for 45 days over regulatory gaps
- Rapid City delayed its AI policy vote after critics called rules too vague
- Survey shows voters want AI regulation focused on jobs, not just safety
- AI investment boom blurs quality and cyclicality for investors, per ClearBridge
- College instructor Anurag Tiwari turns student AI projects into job assets
- Expert warns AI cannot physically assess homes and may worsen repair projects
- Blue Origin plans to fly New Glenn rocket again in December carrying Blue Ring
Jeff Bezos shares AI future and moon plans
Jeff Bezos discussed his vision for Blue Origin and AI in a recent interview. He said the U.S. must beat China to land humans on the moon by 2028. Blue Origin plans to return its New Glenn rocket to flight in December. Bezos has invested $28 billion in Blue Origin and believes AI could boost productivity greatly. He also said AI might lead some Americans to work fewer days.
Jeff Bezos details moon and AI future plans
Jeff Bezos said the U.S. must win the race against China to put humans on the moon. He wants to build a sustainable lunar base, not just visit. Blue Origin is fixing its launch pad and plans to fly the New Glenn rocket again in December. The next flight will carry Blue Ring, a vehicle for moving payloads in space. Bezos wants Blue Origin to eventually launch rockets every minute.
Expert warns on using AI for home repairs
Michael Geisler of Ace Handyman Services says AI can help with home repair ideas and checklists. But he warns that AI cannot see the home in person and may lead to bigger problems. Homeowners sometimes start projects they cannot finish and make things worse. Geisler recommends calling a professional for work involving heights, electrical systems, or unclear damage. He also suggests fall tasks like cleaning gutters and sealing gaps.
San Joaquin County pauses new AI data centers
San Joaquin County stopped approving new data center projects for 45 days. The Board of Supervisors said the county lacks rules for data center noise, energy use, and emergency response. Officials will study whether local power infrastructure can support multiple large data centers. No proposals are currently pending in the unincorporated areas. Sacramento leaders also debate data center rules, with some calling for limits and others for guardrails.
Voters want AI rules for jobs not just safety
A survey found that people in many countries want an independent authority to regulate AI. Support is highest in the United Kingdom and lowest in Italy and South Korea but still high. People favor stricter rules when AI takes jobs without raising company profits. Support drops when job losses come with new job creation. The survey also found that political ideology shapes views on AI regulation.
NVIDIA PivotOPD helps AI agents learn from big mistakes
NVIDIA researchers worked with Princeton and the University of Maryland to create PivotOPD. This method trains multi-turn AI agents to recover from their worst early errors. It outperformed 13 baselines on ALFWorld, WebShop, and Search-based QA using Qwen3-1.7B and Qwen3-8B models. The research shows that learning to recover is possible with this new approach.
Rapid City delays AI policy vote
The Rapid City Council voted to delay a proposed artificial intelligence policy. The policy covers four pages and six principles including human oversight and privacy. Critics said the rules are too vague and do not define key terms. City staff were directed to create a more detailed document, but no date was set for a new vote.
College instructor turns student AI projects into job assets
Anurag Tiwari teaches AI courses at the College of Charleston. He uses a Greek method of why, what, and how to guide students. Students build real AI tools like chatbots and hiring aids for their final projects. One student built a rules judge for a card game and landed a job at the Naval Information Warfare Center. Employers value these deployable projects more than traditional class assignments.
Networks become the key focus for AI systems
AI is now central to how organizations operate, making networks a critical dependency. Future AI-native networking will rely on stealth, multipath transport, and Zero Trust principles. Networks will hide endpoints, split traffic across many paths, and verify every session. They will also self-optimize, self-heal, and scale automatically to meet AI demands.
AI investing blurs quality and cyclicality
Matthew Lilling of ClearBridge says the AI investment boom makes it hard to tell good companies from cyclical ones. AI tools help investors analyze data and predict market trends. But these tools can be swayed by sentiment and economic conditions. This blurring of quality and cyclicality creates new challenges for investors.
JetBrains launches Mellum2.1 open coding model
JetBrains released Mellum2.1, an open 12B mixture-of-experts model designed for coding agents. The model activates 2.5B parameters per token and uses reinforcement learning in real software environments. It is available under Apache 2.0 license on Hugging Face. Mellum2.1 can explore code repositories, edit files, and verify its own changes.
Sources
- Jeff Bezos reveals AI vision that could transform the American workweek
- Jeff Bezos reveals his vision for America's next frontier — from AI to the moon
- Using AI for home repairs? Expert says know when to call a professional
- San Joaquin County orders temporary halt on new AI data center projects
- Voters Want AI Regulated for Jobs, Not Just Safety
- NVIDIA PivotOPD Teaches Multi-Turn AI Agents to Recover From Pivotal Mistakes
- Rapid City council delays vote on artificial intelligence policy
- Computer Science Instructor Has Students on (and in) the Right Course
- The Network Is Now the Critical Dependency for AI
- Commentary: AI is blurring the line between quality and cyclicality
- JetBrains Releases Mellum2.1: A 12B MoE Open Model for Coding Agents
Comments
Please log in to post a comment.