Legal and medical professionals are wrestling with how to use AI responsibly. At Legal Innovators UK, SSE's Emily Bull described a framework where legal teams decide whether a task can be eliminated, enabled, engineered, or truly requires a lawyer's expertise. Meanwhile, a survey found 94% of legal professionals use AI, yet only 29% have embedded it in their organization's strategy. Most worry about inaccurate information from AI tools, and 18% say no one owns AI governance within their firm.
Medical education is also adapting. Residents in Uganda lost access to an AI tool for reading heart scans during a power outage, highlighting concerns about over-reliance on technology. Medical schools now design exercises using paper and pencil to ensure students build foundational clinical judgment. A Pew Research survey found 47% of U.S. adults believe AI will worsen the patient-doctor relationship, and most want human review of AI-driven medical decisions.
In aerospace, Shield AI will build its X-BAT uncrewed combat aircraft in the Puget Sound region starting March 2027. The U.S. Navy has provided $200 million in development funding for the vertically-takeoff-and-landing jet. Shield AI plans to demonstrate mission capability by 2028 and begin full production by 2029, hiring thousands of workers for its Des Moines facility.
On the education front, professors show divided opinions. One Washington University professor uses AI for lecture slides and research analysis, while another avoids the technology entirely, fearing students will skip deep learning. Korea Aerospace Industries is taking a different approach, announcing AI training programs for its suppliers starting in November, with regular aerospace-focused courses beginning in 2027.
Anthropic's S-1 filing ahead of its IPO reveals that the company depends on consumer use of Claude.ai to drive enterprise revenue. The filing acknowledges that long-term effects of AI on mental health and social behavior remain poorly understood. Dave Lempers, CEO of Mainecode, warns that Australia faces AI security risks due to a lack of local software skills and over-reliance on foreign technology, arguing the country needs its own AI infrastructure.
Key Takeaways
- 94% of legal professionals use AI, but only 29% have it embedded in organizational strategy
- SSE's Emily Bull uses a 4 Es framework to guide AI adoption in legal teams
- Medical schools use paper-based exercises to ensure students build skills beyond AI tools
- 47% of U.S. adults believe AI will worsen the patient-doctor relationship
- Shield AI receives $200 million in U.S. Navy funding for X-BAT production starting 2027
- Anthropic's S-1 reveals Claude.ai consumer use drives enterprise revenue model
- Mainecode CEO warns Australia lacks local AI security software expertise
- KAI begins AI training for 42 suppliers in November with aerospace courses in 2027
- Professors show divided views on AI in college teaching
- 18% of legal firms have no clear ownership for AI governance
SSE Legal Uses the 4 Es Framework to Guide AI Adoption
Emily Bull, Director of Legal at SSE, uses a 4 Es test to decide how AI can help her legal team. The test asks whether a task can be Eliminated, Enabled for someone else, Engineered into a better process, or truly needs a lawyer's Expertise. She spoke at Legal Innovators UK in London on November 4 and 5. Bull stresses that AI helps lawyers work faster but human judgment must stay in charge for decisions involving accountability and context.
Law Firms Must Strengthen AI Governance and Accountability
Karen Waldron reports that 94% of legal professionals now use AI, but only 29% have it embedded in their organization's strategy. 18% say no one has clear ownership for AI governance. 83% worry about inaccurate or fabricated information from AI tools. Large firms and in-house teams have clearer structures than smaller firms. The article calls for clear accountability and rules about which tools lawyers may use and what information they can enter.
Medical Schools Adapt Teaching as AI Enters the Classroom
A team of medical residents in Uganda lost power and their AI tool for reading heart scans stopped working, leaving them unsure of their manual skills. Educators say AI will assist physicians but not replace them. Some schools design exercises using paper and pencil to ensure students build foundational clinical judgment. Faculty members stress that preserving human oversight is essential so doctors can spot AI errors and hallucinations.
Paul Knoepfler Examines AI's Risk to Biomedical Research
Paul Knoepfler, a professor at UC Davis School of Medicine, worries that overreliance on AI could make biomedical researchers less capable. He is excited about how useful AI can be but is concerned about its effects on new and future biologists who grow up using it. He also raises the possibility of job losses in science due to AI. The article notes that this is a STAT+ exclusive story requiring a subscription to read in full.
5 Examples Show How AI Reasoning Differs From Human Thinking
Researchers found that chatbots built on large language models do not truly think or understand like people. AI systems only process words called tokens and use math to predict which words appear together, while humans learn from real-world experiences. After training ends, an LLM stops learning and changing. One example showed early AI models confusing a camel on grass with a cow in a desert because they memorized image backgrounds instead of understanding animals. Human minds still outperform AI where understanding and context matter.
Americans support AI in healthcare but want human oversight
A Pew Research survey found that 47% of U.S. adults think AI will worsen the patient-doctor relationship. Most Americans want humans to review AI decisions, especially for diagnoses and insurance claims. Only 5% support AI making insurance decisions without human input. Younger adults are more worried about AI causing poor health outcomes. Americans also have low confidence in insurance and AI companies to protect patient interests.
KAI to train suppliers in AI skills
Korea Aerospace Industries will train employees and executives at its suppliers in artificial intelligence. The company held a special lecture on October 28 with about 60 attendees from 42 suppliers. Yonam Institute of Technology and Korea Testing Laboratory helped organize the event. KAI will start training programs for working-level employees in November. Regular customized courses focused on aerospace manufacturing will begin in 2027.
AI-piloted X-BAT fighter jet production coming to Seattle area
Shield AI will build its X-BAT fighter jet in the Puget Sound region starting in March 2027. The X-BAT is an uncrewed combat aircraft that takes off and lands vertically. The U.S. Navy has provided $200 million in development funding. Shield AI plans to demonstrate mission capability by 2028 and begin full production by 2029. The company is hiring thousands of workers for the Des Moines facility.
Two professors show different views on AI in college teaching
Professor Mitchell Sommers at Washington University uses AI to create colorful lecture slides and analyze large research datasets. He believes AI is a valuable tool that improves teaching and research. Professor David Schuman is skeptical and worries AI shortcuts prevent students from deeply learning material. Schuman ignores the university's AI initiatives and does not feel pressure to adopt the technology. The two professors show how colleges are handling the rise of AI in different ways.
Australia faces AI security risks without local software skills
Mainecode CEO Dave Lempers warns that Australia lacks software expertise in AI security. He says the country depends too much on foreign technology and has no core compute capabilities. Lempers argues that without local AI models, Australia could lose access to crucial technology during a crisis. He also points out legal and ethical risks when using foreign AI services. He calls for Australia to invest in building its own software infrastructure.
Anthropic S-1 Reveals Consumer AI Risks Ahead of IPO
Anthropic's S-1 filing shows the company relies on consumer use of Claude.ai to drive its enterprise revenue model. The filing admits that long-term effects of AI on mental health, thinking, and social behavior are not fully understood. It also notes that the Long-Term Benefit Trust cannot control day-to-day business decisions. Anthropic acknowledges risks to children and potential harms to its reputation. The article argues that regulators need to focus on consumer AI products, not just frontier AI systems.
Sources
- The 4 Es Inhouse AI Framework
- AI Governance Must Keep Pace with Legal Practice
- Schooled by AI: The evolving role of artificial intelligence in medical education
- Is AI more a partner, competitor, or crutch for biomedical scientists?
- These 5 quirks reveal how AI doesn’t think like we do
- Americans Are Open to AI in Healthcare, but Want Humans Making Final Calls
- KAI to Launch AI Training for Suppliers
- Production of AI-piloted X-BAT fighter jet coming to Puget Sound in 2027
- As AI sweeps into higher education, two professors take divergent approaches
- Australia AI security: national risk and future strategy
- Enterprise AI with a Consumer Funnel: What Anthropic’s S-1 Reveals About the AI IPO Era
Comments
Please log in to post a comment.