Past tech like movies, radio, TV, even MOOCs were supposed to revolutionise education and... didn't, really Derek Muller -> points out that maybe it's because education isn't really about just getting information to people (we've had books for ages, right?) but about the effortful learning process and the human connection Like a personal trainer – the gym is there, but you need someone to push you, hold you accountable, and maybe a group doing it together So, if AI isn't necessarily replacing things entirely, what is it? Mostly, people agree it's a TOOL But how to think about that tool matters heaps! Jeremy Utley at Stanford says ditch the 'tool' mindset. Think of AI as a TEAMMATE If it's a tool, you get a 'good enough' or mediocre result and just move on or give up If it's a teammate, you give it feedback, you coach it, you help it get better HOT TIP : Get AI to ask YOU questions! Instead of just asking it how to do something, tell it you want its help and ask it to ask you questions until it understands your context This helps it give way better, non-obvious recommendations Utley even had AI interview him about someone else to build a psychological profile and then roleplay a difficult conversation This points to a key idea: HUMAN AGENCY YOU are in charge. YOU decide what tasks AI does, YOU evaluate the results, YOU provide feedback You have to be able to DESCRIBE what AI should do Current human effort that is required Curiosity & Tackling Hard Problems: Anima Anandkumar says these roles won't be replaced , Be curious, go after difficult things Use AI to fuel your curiosity, not kill it Asking Questions & Critical Thinking: Ask questions about everything AI can give answers fast, but you need to check them Po-shen Loh at CMU says the new skill is learning "how to grade the homework" – vetting AI's output You need to be a quick and analytical thinker to know if what AI says is correct Creativity: Utley says AI can hugely boost creativity. But sometimes it made people less creative in his studies! -> Again, treating it as a teammate helps Creativity is NOT just magic; it's a discipline! We gotta feed your brain with inspiration and different inputs The definition of creativity? Doing more than the first thing you think of. AI makes getting 'good enough' super easy, but getting exceptional still requires effort , What you bring (experience, perspective, inspiration) is what makes your AI output unique Understanding People & Their Needs: Loh stresses this for creating value. You can't just think about value theoretically; you need to interact with people from different backgrounds, understand their pain points, needs, and what drives them , This means getting out there, maybe even giving talks in parks! Understanding Physical Behaviour / Science: Anandkumar's team developed neural operators – AI trained to understand physics, not just text patterns -> Revolutionised weather forecasting, being tens of thousands of times faster than traditional methods and surprisingly accurate Showed AI isn't limited by what others think is difficult if you have data and methods Programming: AI is getting better, but a great programmer who can check and fix AI's code will be more valuable than ever. You still need to understand what AI is doing under the hood. Handling Tedious Tasks: This is a great win! AI can automate repetitive stuff. Example: A ranger in the US National Park Service built a simple AI tool in 45 minutes to save two days of paperwork per job . This tool is now saving the service an estimated 7000 days of labour annually! -> Non-technical people CAN have huge impact with basic AI collaboration skills . Okay, digging a bit deeper into the LEARNING PROCESS , which Muller focuses on heavily: We have two systems in our brain: System 1: Fast, automatic, intuitive, relies on long-term memory/patterns. Where that "10 cents" answer for the bat/ball problem comes from System 2: Slow, effortful, methodical, uses working memory, can check mistakes. The voice in your head System 2 has limited capacity (maybe only ~4 things at once!) Learning involves using System 2 repeatedly and effortfully to build up those patterns in long-term memory that System 1 uses , This is how experts develop "chunking" – seeing complex things as one unit (like chess masters seeing board configurations or physicists seeing equations) BIG CONCERN: If AI does the hard work for us, we might not do the effortful practice needed to build those System 1 structures. This could stop us developing deep expertise or skills like writing effectively In education, we need to manage cognitive load 49 : Extraneous load (distractions) -> BAD Intrinsic load (task difficulty) -> Manage it! Don't overload working memory with too much novel stuff at once Break things down, slow down Germaine load (thinking about thinking/patterns) -> GOO Teaching methods that help: Scaffolding ! Don't just throw hard problems at students and say "figure it out" (discovery learning can be risky!) Provide worked examples and gradually fade assistance This reduces intrinsic load while System 2 builds understanding Need to reach MASTERY so skills become automatic in System 1. Making a task slightly harder to process (like using a difficult font) can force System 2 to engage and improve accuracy! (Though maybe not practical advice!) AI in Education Specifics: AI tutors can provide timely feedback AI can offer scaffolding and hints AI can be good for drill and practice BUT , the huge risk is students using AI to do the work for them , avoiding essential practice This means assessments might need to change Maybe assignments need to be done in controlled settings like exams? Or have parts where AI isn't allowed. It's a real challenge Loh's point about needing to "grade the homework" is crucial here -> students must be able to verify AI's answers, which requires understanding the material deeply themselves Entrepreneurship & Value Creation in the AI Age (Loh's perspective): The world needs people who can solve problems and create value Building something scalable often needs a business model Loh built an ecosystem by identifying pain points for different groups and creating a win-win-win model -> Connecting math students with math experts + drama coaches to teach communication skills Drama people get jobs, math kids learn to think AND communicate Success = not just money, but convincing people to enjoy being thoughtful 57 . The AI world is a "Wild West" of opportunities for people good at creating value , Random Interconnections / Takeaways: The guy who saved the National Park Service 7000 days was non-technical. -> Don't need to be a coder to use AI for big impact. Creativity requires discipline, just like getting good at anything requires effortful practice Maybe the "revolution" is less about the tech replacing humans, and more about how it changes what humans need to be good at – directing, vetting, being creative, understanding others, doing the deep thinking that AI can't (yet?) This segment showcases the professor's teaching methodology, emphasizing critical thinking and intuitive problem-solving skills over rote memorization. He uses the example of designing a fire alarm to illustrate how practical, intuitive approaches can complement mathematical training. This segment uses a humorous anecdote to illustrate the interplay between System 1 and System 2 thinking, demonstrating how quickly intuitive responses can be overridden by deliberate thought. This highlights the importance of critical thinking and self-awareness in learning.This segment discusses the limited capacity of working memory (around four items), using the example of a digit span test and pupil dilation to illustrate cognitive load. This is crucial for understanding how to design learning experiences that don't overwhelm learners' cognitive resources. This segment presents a thoughtful discussion on using AI to answer questions, weighing the benefits of efficiency against the potential loss of deeper understanding gained through independent research and problem-solving. This segment discusses the potential roles of AI in education. While acknowledging AI's potential for providing timely feedback, it also highlights the concern that AI could reduce the need for effortful practice, potentially hindering the development of deep understanding and mastery. This segment presents a balanced view of AI's potential benefits and risks in education.This segment focuses on the potential negative impact of AI on learning, specifically its ability to reduce the need for effortful practice. The speaker expresses concern that readily available AI tools could prevent students from engaging in the essential process of struggling, practicing, and iteratively improving their skills. This segment underscores the importance of effortful engagement in learning.This segment addresses the question of why educational revolutions haven't materialized despite technological advancements. It suggests that the problem isn't access to information but rather the lack of engagement, the absence of a supportive learning environment, and the crucial role of social interaction in learning. This segment offers a nuanced perspective on the challenges of educational reform. Important Moments from Videoes This segment introduces a novel approach to using AI: instead of directly asking questions, use AI to help formulate better questions. The example of using AI to refine queries before submitting them to AI itself showcases a powerful, less intuitive method of leveraging the technology.This segment presents a compelling case study of how a non-technical employee in the National Park Service used AI to automate tedious paperwork, saving thousands of hours of work. It demonstrates the practical, impactful applications of AI for everyday professionals.This segment introduces the core concept of treating AI as a collaborative teammate rather than a mere tool. It highlights the significant difference in outcomes based on this mindset shift and introduces the "realization gap" found in research.This segment presents research findings showing that while AI can increase productivity, it doesn't always enhance creativity. It explains that the key difference lies in how people approach AI: as a tool or as a teammate. The focus on coaching and feedback is crucial.This segment provides practical examples of how treating AI as a teammate can unlock its potential. Roleplaying difficult conversations and using AI to generate questions are highlighted, showcasing creative applications that extend beyond simple task automation. This segment uses a simple matchstick puzzle to illustrate the importance of thinking outside the box and approaching problems from unconventional angles, a crucial skill in today's world. It highlights the difference between rote learning and creative problem-solving. Will AI change the Education????