AI is changing how people write, assess and make decisions, but the effects depend on the setting. A November 12 article on university grading describes two possible futures: students submit AI-generated work that AI then grades, or instructors use AI to reduce grading time. A college essay coach also reports that this year’s application essays sound more formal and less like students’ own voices, which the freelancer attributes to AI tools. Architect Rajiv Kathpalia says AI will reshape the economics of architecture firms, though the article provides no further detail.
Research on chatbot relationships finds that people rate emotional and intellectual intimacy similarly with chatbots, long-distance human partners and in-person partners. But chatbot relationships differ in loneliness, satisfaction and well-being, with outcomes tied to the kind of intimacy and a person’s broader social support. In a separate argument about AI consciousness, David Weinberger says language models can manipulate symbols and appear to reason or feel without understanding as humans do. He links human understanding to embodied experience, personal projects and awareness that life is at risk.
AI’s role in high-stakes decisions is also under scrutiny. Workday has asked a San Francisco federal judge to reject nationwide class certification in Mobley v. Workday, which alleges its hiring tools discriminate based on race, sex, age and disability. Workday argues that customers make hiring decisions and their different settings leave no common evidence. The case, filed in February 2023, is led by Derek L. Mobley and three other applicants. An age-discrimination group was provisionally certified in May 2025; more than 15,000 people had joined by October 2026. Workday filed motions November 10 and 11, replies are due December 8, and a certification hearing is set for March 9, 2027. Separately, intensive-care researchers propose training clinical AI on carefully selected, high-performing physicians rather than routine data alone, while accounting for differences in patient case mix. They warn that routine-data models can reproduce existing biases; their example shows how an average ICU admission rate of 12% could become a target even when top clinicians achieve non-inferior outcomes at 8%.
Security concerns are prompting AI and technology companies to recruit former intelligence officials. OpenAI, Anthropic and Nvidia are among firms with growing hiring activity in counterintelligence. Many roles pay $200,000 to about $500,000, while top corporate security chiefs can earn $1 million to $4 million a year, including stock awards. OpenAI added former NSA director Paul Nakasone to its board and safety and security committee in June 2024. Anthropic added former CIA deputy director David Cohen and former NSA cybersecurity director Dave Luber to its advisory council in August 2025. The report connects the hiring push to threats against trade secrets and a Nvidia scandal it describes.
Google researcher Pietro Miotti used 48 agents from different AI apps to search NASA LiDAR data collected across Peru. In 72 hours, the agents identified several possible structures, including a site not yet discussed in scientific literature, and drafted a Spanish email to Mexico’s National Institute of Anthropology and History. INAH representatives are investigating, but the site has not been confirmed as an ancient Maya site. Miotti says researchers should use LiDAR across the planet to look for more sites.
Infrastructure and hardware developments offer a counterpoint to AI’s expanding reach. The Limits to AI Act article calls for controls on data-center growth, citing heavy energy and water use, noise and air pollution, and infrastructure costs that utilities can pass to households. Residential electricity rates rose 30% from March 2021 to March 2026, and the largest hyperscale data centers can use up to five million gallons of water daily for cooling. Senator Bernie Sanders has introduced a federal budget amendment to limit data-center growth. Meanwhile, Qualcomm CEO Cristiano Amon says AI will not replace smartphones soon. Handsets remained Qualcomm’s largest semiconductor business in fiscal 2026, generating $24.07 billion of $37.21 billion in semiconductor revenue despite a 13% year-over-year decline in handset sales. Automotive chip revenue rose 89% to a record $1.99 billion.
Key Takeaways
- University grading may involve AI evaluating AI-generated work, while instructors may also use AI to reduce grading workload.
- A college essay coach says application essays have become more formal and less reflective of students’ own voices, possibly because of AI tools.
- Chatbot relationships can receive intimacy ratings similar to human relationships, but loneliness, satisfaction and well-being differ.
- Workday is contesting nationwide class certification in a hiring-discrimination case; more than 15,000 people had joined the provisionally certified age-discrimination group by October 2026, with a hearing set for March 9, 2027.
- Intensive-care researchers advocate training clinical AI on carefully selected high-performing physicians and accounting for patient case mix to avoid reproducing biases in routine decisions.
- OpenAI, Anthropic and Nvidia are among companies hiring former intelligence personnel; top corporate security chiefs can earn $1 million to $4 million annually, including stock awards.
- Google researcher Pietro Miotti’s 48-agent LiDAR search identified a possible site in Mexico, but INAH has not confirmed it as an ancient Maya site.
- The Limits to AI Act article cites data-center energy, water and pollution impacts; residential electricity rates rose 30% from March 2021 to March 2026, and some large facilities use up to five million gallons of water daily.
- Qualcomm CEO Cristiano Amon says smartphones will not be replaced by AI soon; handset revenue was $24.07 billion in fiscal 2026, while automotive chip revenue climbed 89% to $1.99 billion.
- David Weinberger argues that language models lack human consciousness because human understanding is tied to embodied experience, personal concerns and awareness of risk.
AI grading could leave university work unread by people
A November 12, 2026 article by Margaret Bearman explores how artificial intelligence could change university grading. One possible outcome is that students submit AI generated work that AI systems then grade, leaving no human to read it. Another possibility is that teachers use AI to reduce the time and work involved in grading. The article presents these as contrasting possible futures for university assessment.
Study finds AI chatbot intimacy can feel like human connection
A study co-authored by University of Michigan School of Information researcher Lionel Robert examined intimacy in relationships with chatbots and people. It found similar ratings for emotional and intellectual intimacy in chatbot, long-distance human and in-person human relationships. However, chatbot relationships showed different patterns of loneliness, satisfaction and well-being. The researchers say the experience depends on the type of intimacy and a person's wider social support.
Workday asks court to block class action over AI hiring tools
Workday asked a federal judge in San Francisco to reject nationwide class certification in a case alleging its AI hiring tools discriminated based on race, sex, age and disability. The company says its customers make hiring decisions and that differences in customer settings mean the claims have no common evidence. The case, Mobley v. Workday Inc., was filed in February 2023 and is led by Derek L. Mobley and three other applicants, who allege repeated automated rejections. An age discrimination group was provisionally certified in May 2025, and more than 15,000 people had joined by October 2026. Workday filed its motions on November 10 and 11, 2026, replies are due December 8, and a certification hearing is scheduled for March 9, 2027.
AI companies hire former intelligence officers to guard against spying
AI and technology companies are hiring former intelligence officers to protect against espionage, with hiring activity growing in recent years at firms including OpenAI, Anthropic and Nvidia. The article reports that many counterintelligence jobs pay between $200,000 and about $500,000, while top corporate security chiefs can earn $1 million to $4 million a year, including stock awards. OpenAI appointed former National Security Agency director Paul Nakasone to its board and safety and security committee in June 2024, and Anthropic added former CIA deputy director David Cohen and former NSA cybersecurity director Dave Luber to its advisory council in August 2025. The report links the industry's focus on security to growing concerns about threats to trade secrets and the Nvidia scandal described in the article.
Qualcomm CEO says AI will not replace smartphones soon
Qualcomm CEO Cristiano Amon said AI will not replace smartphones anytime soon. Smartphones remained Qualcomm's largest semiconductor business in fiscal 2026, bringing in $24.07 billion of its $37.21 billion total semiconductor revenue, though handset sales fell 13% from the prior year. Automotive chip revenue rose 89% to a record $1.99 billion, helping offset the decline. Amon also said Qualcomm had survived major changes in the mobile industry while expanding into automotive technology and advanced AI chips.
AI tools are taking over college essay coaching
A freelancer who helps students with college application essays says this year’s essays became more formal and rigid. The writer believes students used AI tools to create them, making the essays sound less like their own voices. In past years, the freelancer helped about 10 students get into their top-choice schools each year. The writer plans to explore what AI-generated writing could mean for writers and readers in future posts.
Rajiv Kathpalia says AI will reshape architecture firms
The article presents Rajiv Kathpalia’s view that artificial intelligence will change the economics of architectural practice. The provided article text does not include further details about how AI may affect architecture firms or their work.
AI agents help uncover a possible ancient Maya site in Mexico
Google researcher Pietro Miotti asked 48 agents from different AI apps to search NASA LiDAR data collected across Peru for possible structures. The agents found several possible buildings and identified a site not yet discussed in scientific literature. They also suggested contacting Mexico’s National Institute of Anthropology and History, or INAH, and drafted an email in Spanish with information about the location. INAH representatives are investigating, but the site has not yet been confirmed as an ancient Maya site. The process took 72 hours, and Miotti said researchers should use LiDAR across the planet to help find more sites.
Why AI may never have the kind of consciousness humans experience
David Weinberger argues that language models are not conscious, even when they can reason, remember, and appear curious or emotional. He uses philosopher John Searle’s Chinese Room thought experiment to explain that computers can manipulate symbols without understanding their meaning. Weinberger contrasts this with human experience, which is tied to a living body that feels pain, tastes food, and moves through the world. He also draws on Martin Heidegger’s idea that people understand things through their projects and what matters to them. The article links that sense of what matters to the knowledge that humans are at risk and care about their lives.
Limits to AI Act calls for controls on data center growth
An article published November 12, 2026, introduces the Limits to AI Act as a way to limit harms to people and the natural world. It says data centers use large amounts of energy and water and create noise and air pollution, while utilities can pass infrastructure costs on to households. Average residential electricity rates rose 30% between March 2021 and March 2026. The largest hyperscale data centers can use up to five million gallons of water per day for cooling, and some facilities operate near homes. The article also notes that Senator Bernie Sanders introduced a federal budget amendment to limit data center growth.
Training AI on Top Physicians Could Reduce Bias in Intensive Care
Daniel Ostrovsky, Daniel Talmor, and Victor Novack argue in a perspective published in Intensive Care Medicine that clinical AI should learn from carefully chosen, high-performing physicians. They say models trained on routine data can reduce differences in recommendations but may also repeat biases in existing clinical decisions. For example, if ICU admission rates range from 0 to 24 percent and average 12 percent, an algorithm could push everyone toward 12 percent, even when top-performing clinicians achieve non-inferior outcomes with an 8 percent rate. The authors suggest selecting training data from clinicians identified through measures such as complication rates, ICU use, peer rankings, patient assessments, experience, and knowledge tests. They also warn that developers must account for differences in patient case mix when judging clinicians’ performance.
Sources
- AI grading means university assessments could theoretically never be read by a human
- Finding romance with AI: Study explores benefits, limits of chatbot intimacy
- Workday asks judge to block class action over AI hiring screening
- AI Firms Hire Ex-Spies Amid Nvidia Security Scandal
- Qualcomm CEO: AI won't replace your smartphone anytime soon
- I’ve Been AI’d Out of a Job
- Artificial intelligence will change the economics of architectural practice - Rajiv Kathpalia on the future of architectural practice
- 'I think we're on the brink of a revolution': Google researcher hails AI agents after they found what looks like an ancient Maya site in Mexico
- Why AI and Consciousness May Never Meet
- Beyond a moratorium: Introducing the limits to AI act
- Training Artificial Intelligence on Top Physicians May Reduce Intensive Care Bias
Comments
Please log in to post a comment.