• THE AI WAY
  • Posts
  • AI Missteps in Alaska Policy Draft: A Call for Accuracy in AI-Generated Content

AI Missteps in Alaska Policy Draft: A Call for Accuracy in AI-Generated Content

Author

Niel de Kock
November 07, 2024

Welcome to the latest news in AI and AI in educational news.

Education AI News

AI Hallucinations Mislead Alaska Policy, Sparking Concerns Over AI Reliability in Decision-Making

Alaska recently faced a setback in using AI for policymaking when fabricated AI-generated citations were included in a proposed policy draft on cellphone restrictions in schools. Alaska's Education Commissioner, Deena Bishop, used generative AI to draft the policy, which led to the inclusion of nonexistent academic references that went unverified and reached the State Board of Education. Although these citations were intended as placeholders, their presence in the final draft highlights the risks of unchecked AI content in sensitive areas like education policy. This incident underscores the importance of fact-checking AI outputs to prevent misinformation in decision-making and maintain public trust in AI technologies.

Why AI Won't Ruin Children's Education

As AI tools like ChatGPT evolve, concerns over cheating and plagiarism in education have led some institutions to revert to pen-and-paper exams or outright ban AI. However, Vaughan Couillault, president of the Secondary Principals Association, argues that banning AI won’t solve the problem. He emphasizes that AI is a tool with a wide range of applications, from beneficial study aids to potentially enabling plagiarism. Couillault believes that, similar to calculators or search engines, AI can be integrated thoughtfully into learning when students first understand foundational concepts. Using AI as an educational aid—like one student’s approach of having ChatGPT break down complex lecture slides—demonstrates its potential to deepen understanding. Couillault advocates for a balanced approach, monitoring AI’s impact carefully rather than banning it outright, as education continues to adapt to these evolving technologies.

Opinion: AI in Education – a Disservice to Teachers or a True Assistant?

In this opinion piece, Byron Scaf critiques the EdTech industry's approach to integrating AI in education, arguing that many tools are implicitly designed to replace rather than support teachers. Scaf observes a troubling trend of dismissing teachers' expertise and the irreplaceable human connection they bring to student learning. AI tools, which often provide automated feedback, lack the personal insight that teachers use to meet individual student needs effectively. Scaf emphasizes that effective learning requires trust and context, elements best provided by real teachers, not algorithms. He advocates for using AI as an assistive tool, enhancing rather than supplanting teachers’ roles, and enabling students to become informed, skillful AI users without sacrificing foundational knowledge.

Survey Highlights Surge in AI Use Among Educators, with Optimism and Caution

A new report from HMH’s 2024 Educator Confidence Report shows a significant increase in AI integration in classrooms, as educator confidence in technology grows post-pandemic. Generative AI use among teachers has increased fivefold, with many teachers finding it valuable for time-saving and instructional enhancement. Key findings include that 97% of educators are confident in their tech skills, while 76% of those using generative AI believe it positively impacts their work. However, challenges remain, with concerns about plagiarism, student over-reliance on AI, and a need for professional development to safely incorporate AI tools. Despite these challenges, teachers continue to show resilience and openness to technology’s role in transforming education, though issues like mental health and compensation continue to affect job satisfaction.

AI News

Could AI Help Develop New Treatments for Children’s Brain Tumors

Researchers at the Cancer Research UK Children’s Brain Tumour Centre of Excellence in Cambridge are exploring how AI might advance treatments for pediatric brain tumors, a rare but aggressive condition affecting about 420 children annually in the UK. The team, led by Dr. Elizabeth Cooper and Professor Richard Gilbertson, aims to leverage AI to create digital models of brain tumors. These models would enable virtual clinical trials, helping to identify less toxic and more effective therapies, especially as immunotherapy has shown limited success due to the unique immune system of the brain. The project aims to improve survival rates and reduce the severe side effects associated with current treatments like radiotherapy.

AI-Driven Virtual Trials Could Revolutionize Cancer Treatment Research

A £5.9 million Cancer Research UK grant is funding an innovative project led by the University of Manchester and The Christie cancer hospital, aimed at improving cancer treatments through AI-simulated clinical trials. This research will create "virtual" patient groups, generated from real patient data, allowing scientists to test the effectiveness of treatments like radiotherapy and proton beam therapy in a faster, safer, and more cost-effective way. By focusing on patient-specific genetics and tumor characteristics, the AI-driven trials could accelerate advancements in radiotherapy, ultimately enhancing survival rates for patients like Martin Storey, a lung cancer survivor who benefited from earlier clinical trials.

Why AI Models Like GPT-4 May Not Truly Understand the World

A new study reveals that large language models (LLMs) like GPT-4 may not form a true understanding of the world and its rules, despite their impressive performance on various tasks. Researchers at MIT and Harvard used new metrics to analyze how well LLMs could model real-world environments, such as navigation in New York City and the board game Othello. They found that while LLMs could provide accurate directions or moves, they didn’t truly understand the rules, resulting in significant errors when minor changes, like street closures, were introduced. This highlights the limitations of LLMs in forming coherent world models and suggests that further innovation is needed to build AI with a deeper, rule-based understanding of real-world contexts.

Anthropic Advocates for AI Regulation to Prevent Catastrophic Risk

AI company Anthropic is urging structured regulation to prevent potential misuse of advanced AI, particularly in cybersecurity and CBRN (chemical, biological, radiological, and nuclear) fields. Warning that the next 18 months are crucial for establishing AI safeguards, Anthropic has proposed a Responsible Scaling Policy (RSP) to elevate safety measures as AI capabilities advance. The organization advocates for transparent, adaptive regulations that strike a balance between safety and innovation, suggesting federal legislation in the U.S. as a comprehensive solution. Anthropic emphasizes that regulations should target foundational safety practices, allowing flexibility to support innovation while mitigating risks.

AI Tools

Sapio AI: Your Personalized AI Coach for Lifelong Growth

Sapio AI is a personal growth app that transforms conversations into a pathway for self-improvement. By using AI-driven practice sessions, Sapio provides real-time feedback to help users build confidence and emotional intelligence in a private, judgment-free space. Tailored to each individual’s goals and unique style, Sapio allows users to work on everything from professional skills to relationship-building. With a gamified journey that rewards progress, Sapio AI makes personal development an engaging adventure. Download now on Google Play or the App Store to start your journey toward becoming your best self.

Todof: Where Reminders Feel Human, Not Robotic

Todof is an innovative reminder app that doesn’t just notify—it compels. Designed to resonate on a psychological level, Todof combines sensory gating with dynamic features to keep you engaged. With Aware Reminders that evolve and feel personal, Auto Icons that create visual cues for each task, and Progressive Sounds that get louder if ignored, Todof adapts to keep you on track. It's not just about reminding you but making you feel your tasks. Available on Android and iOS, Todof is the new way to make sure nothing slips through the cracks.

Enjoy your week till next week.