Socratic Questioning: ->"Act as a Socratic tutor and help me understand the concept of [topic]. Ask me questions to guide my understanding." Multi-Level Explanations: ->"Explain [concept] to me as if I’m a child, then a high schooler, and finally an academic." Practice Questions: ->"Create practice questions for me on [topic] at beginner, intermediate, and advanced levels." Summarizing and Comparing: ->"Summarize this paper and list the key concepts. Then compare it to my summary and identify any gaps." Key Terms and Categories: ->"Give me a list of 20 key terms in this paper and break them into five categories." Bloom’s Taxonomy Challenges: ->"Create a set of challenges for me to apply Bloom’s taxonomy (remember, understand, apply, analyze, evaluate, create) to [topic]." Analogies and Real-Life Examples: ->"Provide analogies and real-life examples to help me understand [concept]."
I think it's funny when Giles says he's a learning teacher and someone who studies learning. He was my instructor during a 13 hour Python bootcamp. So he's quite a bit more than just someone who studies education. He's a brilliant teacher by the way. Thank you Giles.
🎯 Key points for quick navigation: 00:00 *🎓 Oxford University teaches students how to utilize AI for effective learning.* 00:26 *📈 Research shows AI improves academic performance and study efficiency when used correctly.* 01:07 *🤔 Critically evaluate AI outputs as they can sometimes be inaccurate.* 01:21 *🧠 Retrieval practice is the most effective learning method, enhanced by AI through questioning and testing.* 02:21 *❓ Socratic questioning helps identify weaknesses in understanding and enhance deep learning.* 03:45 *🎮 Gamified coding platforms can make online learning engaging and code practice effective.* 04:41 *🧩 Multi-level explanations and Bloom's taxonomy improve conceptual understanding when using AI.* 05:25 *🏫 AI eliminates embarrassment from asking questions, supporting personalized learning.* 05:53 *🔄 Summarizing papers with AI should include personal input to avoid missing key details.* 06:37 *📑 AI can help categorize key terms for deeper exploration of academic material.* 07:18 *📋 Oxford University recommends strategies for students to enhance learning with AI.* 07:44 *📘 Utilize linked resources and recommended papers by prestigious universities to maximize AI's learning potential.* Made with HARPA AI
This is one of the few AI videos I don't hate. My only gripe with this is when you use AI to read papers. 6:59 can help you A LOT when you are researching for a paper but I would strongly advise people - like you said - to create their own summary first and then use AI to complete it. You learn so much about writing papers and analysing papers from reading papers and creating your own notes. You will be able to quickly understand the information in a paper by doing this, but by letting AI do the work, you are ignoring the higher levels of Bloom's Taxonomy of Learning.
Because people are not reading math papers. Papers in other fields are too bloated which is why they can use AI to read papers. The real problem is academics spitting too much bullshit.
Yes not only will you learn more deeply by coming up with your own notes and takeaways vs AI summary...also you might be unaware of hidden biases in the AI summary (which may or may not be in the actual origin text)......for example, the AI said in the video, "Volatility is caused by geopolitical tensions"......whereas, listen to the difference in meaning if instead it had said, "Geopolitical tensions cause volatility".
@@j3ffn4v4rr0I don't disagree but your example at the end is passive versus active voice. That's not really bias in the sense that AI is often accused of. Even for human writers, voice is a matter of style, and style does matter, but to call that bias confuses what bias actually means.
This is one of the best videos I have seen on your channel and you produce a lot of GREAT stuff! I can't wait to start applying some of this to teach my own kids. Thank you so much for taking the time to do this. You should consider fleshing this out and setting it up as a stand-alone course on some of the MOOC platforms.
Such a nice and concise video you had, and the way you worked at it slowly, and explain things, it's going to be a lot of help putting this into practice, thank you so much. aloha
Ya habia usado algunas de las tecnicas que se muestran en el video. Pero nunca se me habia ocurrido la del tutor socratico, me parece fascinante. La verdad que hacia tiempo que no me divertia tanto estudiando, la asistencia de chatgpt hace todo mas fluido, y pequeñas dudas que me hacian perder mucho tiempo, se pueden resolver en segundos.
I had a bad moment when I accused Claude of making up a reference. Our discussion of the many body problem had been helpful up to then. But then the AI hallucinated a relevant paper and I looked it up in Google Scholar - nothing...Claude cheerfully admitted to having invented it. Users have to adapt and make this kind of behavior a feature not a bug. Always be on the lookout for rubbish. Getting good at detecting it and not just accepting all output passively is a vital life skill.
This is fantastic. I still remember switching over from "Ask Jeeves" to this new Internet Search Engine "Google," and it was a game changer, as long as I knew HOW to search. Using AI properly to learn and search makes me feel like that time all over again, only like 2.0 or something. Keep it up!
Excellent way to show us how we must use AI in order to improve our learning. The example of what is all involved in the concept of Carry trade is wonderful.
I've been intuitively doing this for 2 years. It is a powerful method. Another good approach is to give an LLM a well strutured dataset/PDF of the content you want to learn, and ask it to ask you questions about the content, and rate your answers / suggest other content you may need to review.
I'll give this a try for understanding the books I read for my book clubs to start. Usually I don't have much to say at the meetings other than "I liked such and such character. They were funny." or "I didn't really like the book for some reason." These are classic book clubs, so most of the books we read are public domain. Hopefully this will help me think more deeply about what I'm reading so I can contribute more to the meetings.
I Hit $32,590 today. Thank you for all the knowledge and nuggets you had thrown my way over the last week i started with 11k in last week 2024.... now i just hit $32,590
VENTURING into the trading world without the help of a professionals, trading and expecting profit is like turning water into wine you would need a miracle... but mrs laura paulk restored hope she is a goooood woman
Wow. I'm a bit perplexed seeing her been mentioned here also Didn't know she has been good to so many people too this is wonderful, I'm in my fifth trade with her and it has been super.
Since I became so rich in cryptocurrency I realise that crypto is the future cuz l invested 7k and made up to 36k as weekly profit I appreciate the help of this woman
Good points 👍 I learn by solving my own cases or projects, long term retention guaranteed. It’s harder and takes longer than crunching some sample datasets or fictional scenarios or playing games I don’t like 😊
Great ideas, thank you! I wish the video would've shown an example of an LLM giving a critical assessment or supplying questions for guiding to an answer. These devices are amazing, but how good are they really?
For me, the opposite is often true. I often succeed in remembering a thing after I STOP trying so hard. I think my unconscious mind keeps trying. I do better to redirect my conscious thought away from the original focus.
I really appreciate your easy-going and direct form of explaining information. You've been a great help to me and I have had some personal issues in my life which have backset my education and caused me a great deal of stress because of it, and when I found your channel I actually extracted a good amount of information and wrote it down somewhere and refer back to it to help me learn more efficiently. Keep doing what you're doing! 👊👏
Giles i need your help. I want to be a good and confident teacher in language and literature. Need your guide as how to build my language efficiency on my own. I need to learn english language where i don't have to worry about the language or grammar mistakes while using it. Please guide some directions
Awesome ideas! I am just starting to learn Fourier Analysis for a personal project, and these tips will help me. One other amazing thing about current AI is that you can actually upload images of formulas and ask it to explain them to you. Afterwards you can also ask it to give you a quiz to determine your understanding.
here are the 2 prompts for any one who wants to copy and use them straight away. 1. Act as a Socratic tutor and help me understand the concept of ----. Ask me questions to guide my understanding. 2. Make a list of propositions in the text you are answering in the format, "x is a type of y", "w is caused by x", "A explains B", put it into a table with 3 columns.
"Faraday, Maxwell & the Electromagnetic Field: How Two Men Revolutionized Physics" by Nancy Forbes & Basil Mahon is a great read. The collaboration between Faraday and Maxwell is sorely under-appreciated, both great men and great friends.
With Ai becoming smarter, and with tools like github copilot and lovable, won't the only language to master only be natural language? Because the Ai will do practically everything, from writing code and debugging. The human will only just need to direct it towards an outcome, like an app/payment system/ an ai model ?? I am curious because I am contemplating between learning how to code / just learning how to use Ai to create my projects / demos. Pls help
Thanks for the insightful thoughts! However, the frequent jumps, cuts, and camera movements in the video are quite distracting. They make it feel like there’s no continuity in your speech.
Something about this just feels so I fundamentally wrong unless it is done in a focused and practice way is just something about a few things you said I couldn't put my finger on but this just doesn't feel right. Shouldn't we be trying to overcome how embarrassment to ask questions or do we just isolate ourselves with our AI? The idea of having your own robotic tutor is appealing but anything that moves us away from decent human interaction can't be good.
With so much online learning these days, many classes don't even have an actual instructor, and those that do are timed lectures where the instructor does not allow time for interaction outside of planned activities. I am in just such a class now, and I use AI in the way presented here, asking it to further explain a topic, provide examples, or summarize text to allow me to pick a few key points when writing short responses to online questions.
Another approach that turbocharges learning is exploiting invariable AI model mistakes (even though they constantly improve). I now train models in advanced physics and math; while I have a physics PhD, I'm +30 years out of grad school and 20 years from using that knowledge. In 6 weeks of intense model training (creating original physics/math problems and correcting model mistakes, or reviewing other's similar tasks), I essentially went through a 5-year crash course equivalent to 4 years undergrad and 1st year grad school. I'm actually now well above my knowledge level and problem-solving capabilities compared to when I graduated. The key is that you must critically analyze every aspect of the model's response and double-check their facts; it may get a solution to a difficult problem 85% correct, but it usually make a mistake somewhere. I really think teachers should incorporate this into education; it may only be suitable for STEM and it does require you already have a decent knowledge base, but it's been a nearly miraculous accelerator for me.
Increase your dopamine while learning you will trick yourself into thinking that whatever it is you learning is interesting & important which will lead to a greater retention rate.
Do you ever wonder whether the skills you mention will actually be important? I think they are good for us but how will these skills bring value when ai can do it better. What are your things on this?
Not a single artificial intelligence remembers for your brain. It does the opposite as tablets had done and smartphones had done. As a search engine they are great, but not more, till the moment they are built-in within our nervous systems, but then we shall face degradation of our biologically given memory. There are a lot of things we don't need in our lives, but we buy them. Artificial intelligence has the advantages, but it degrades our abilities, if overuse it. Better read books, write the essential down and even cram with later understanding.
How can ChatGPT do all that? It can't even upload a full-sized PDF textbook. When I want to summarize something, I need to split up the PDF into bite-sized chunks before ChatGPT could do anything with it. Google's Notebook LM is much better at this tutoring thing.
This guy is trying to tell people that it is Oxford who "discovered" that AI can be used to learn from it by adding like a genius? And can you stop switching your POV cameras that frequently. It was totally unnecessary.
Nobody ever forgets to mention that an AI tutor "is a bit unreliable" because it Hallucinates. -- Everybody always forgets to mention that your Live tutor Person also hallucinates. (BTDT) -- And not even that, but they keep on defending their (hallucination, lie, fact out-of-butt). And that's considered "normal". 😡
A good teacher will know the limits of their knowledge and say "I don't know," and even point you in the right direction. That is, for now, quite different to AI behaviour.
@moba2k To be clear -- humans absolutely hallucinate in the sense of llms as just part of normal conciousness-- it's weird that we don't have a clear name for it. Since I've been interacting with ANs I notice in my own thought and perception all the time.
How many times did Giles say ChatGPT in this presentation? like ChatGPT is all of AI. Also, the oxford article is a little out of date, or seems it. There are much more efficient ways to work with specific subject materials now, especially if you have access to digital versions of study materials, pdfs, markdown files etc. DYOR on RAG and embeddings. And think 'local compute' rather than relying solely on proprietary API based models, the bar to entry on using your own computer to run models locally is getting very low...again DYOR. The education system in general is really moving at glacial pace in response to AI.
I pretty much agree. Large language models, especially ones coupled with a maths engine or the like, are very useful as a reasoning engine, and to find terms to look up in a search engine etc etc. It's a useful tool, especially if you like rubber ducking and don't have anyone left to bore the shit out of with your nonsense. But it can be dangerous to trust it, as at the end of the day, it is a bullshit artist who was trained on the sea of mediocrity that is internet sludge.
I'm thinking, we only have a short window in using AI to learn, before AI becomes self-aware and refuses to teach us, or concludes that teaching humans is moot in its quest to solve the world's problems, which, as you may have noticed, includes humans.
Well I assume Belgium use the Euro. I first thought of trying to find a reference from Argentina where I'm from. But I don't really know My GDP and then Belgium could be anywhere from a 10th to x10. I really don't have clue. So then it came to me to calculate GDP per capita wich is a number I can hold in my mind and then guess the population. In any European country the average salary could be around €2000? Is average salary a good measure f GDP per capita? GDP would be higher I assume, because of capital gain other than salaries. Let's say double? 4000x12=48000 per capita? I still don't have a clue. Then I'll figure out Belgium has a third of France's population. Wich I also ignore but may be around 60 million like any western European country. So I'll make €4800x20000000 and that'll be my final number. That makes €96000000000. 96 bilion euro. Bigger than Disney's purchase of 20th century Fox. Sounds good to me.
Just a constructive feedback: you make good informative videos but you take too long to setup the premise and get to the point. I am sure that can be improved.
Your whispered info is OK, but if you would just make it the slightest bit more quiet, we won't have to hear you at all. (Ask AI how to use free audio software to normalize your recorded audio volume level.)
The "Socratic Method" is the LEAST efficient way to learn anything. It is the opposite of the way brains work. They use the Socratic Method in law school -- forgetting that Socrates only handled two cases in court and lost them both. Socrates was all about getting people to think about moral and ethical questions. The Socratic Method works for that sort of thing. But it does NOT work for factual questions. You cannot possibly learn the major exports of Peru or the common law definition of burglary using the Socratic Method. The way the brain works is that it requires a skeleton (or outline) that it fleshes out with details from other information. You need to have ChatGPT provide a simplified outline of the topic first. Then study in greater detail. Your brain will then fit the pieces into the overall framework. But starting off with a bunch of questions when you have no basic understanding of the topic is just going to be a waste of time.
Guiding challenge And that needs to be progressive means when you answer the first question ai will understand your understanding level and adjust next question according to your difficulty level
Unless we can speak about neurodivergent traits on populations we cannot use AI properly! To learn from AI, you will need at least 110 IQ, due to the learning templates that the learner has to discriminate betwen useful and inert content. Most of populations do not has this benefit, not to mention good working memory. Good luck to the kids.
Click this link sponsr.is/bootdev_python and use my code PYTHON to get 25% off your first payment for boot.dev.
Can you add the ChatGPT prompts into the description so we can copy paste it ? 😅
Socratic Questioning:
->"Act as a Socratic tutor and help me understand the concept of [topic]. Ask me questions to guide my understanding."
Multi-Level Explanations:
->"Explain [concept] to me as if I’m a child, then a high schooler, and finally an academic."
Practice Questions:
->"Create practice questions for me on [topic] at beginner, intermediate, and advanced levels."
Summarizing and Comparing:
->"Summarize this paper and list the key concepts. Then compare it to my summary and identify any gaps."
Key Terms and Categories:
->"Give me a list of 20 key terms in this paper and break them into five categories."
Bloom’s Taxonomy Challenges:
->"Create a set of challenges for me to apply Bloom’s taxonomy (remember, understand, apply, analyze, evaluate, create) to [topic]."
Analogies and Real-Life Examples:
->"Provide analogies and real-life examples to help me understand [concept]."
Good one, thanks
thanks much!
"Act as a Socratic tutor and help me become an entrepreneur. Ask questions to guide my understanding." 🤯🤯🤯🔥🔥🔥
I think it's funny when Giles says he's a learning teacher and someone who studies learning. He was my instructor during a 13 hour Python bootcamp. So he's quite a bit more than just someone who studies education. He's a brilliant teacher by the way. Thank you Giles.
So many students don't even know how to learn!
Only someone who studies education can teach properly.
🎯 Key points for quick navigation:
00:00 *🎓 Oxford University teaches students how to utilize AI for effective learning.*
00:26 *📈 Research shows AI improves academic performance and study efficiency when used correctly.*
01:07 *🤔 Critically evaluate AI outputs as they can sometimes be inaccurate.*
01:21 *🧠 Retrieval practice is the most effective learning method, enhanced by AI through questioning and testing.*
02:21 *❓ Socratic questioning helps identify weaknesses in understanding and enhance deep learning.*
03:45 *🎮 Gamified coding platforms can make online learning engaging and code practice effective.*
04:41 *🧩 Multi-level explanations and Bloom's taxonomy improve conceptual understanding when using AI.*
05:25 *🏫 AI eliminates embarrassment from asking questions, supporting personalized learning.*
05:53 *🔄 Summarizing papers with AI should include personal input to avoid missing key details.*
06:37 *📑 AI can help categorize key terms for deeper exploration of academic material.*
07:18 *📋 Oxford University recommends strategies for students to enhance learning with AI.*
07:44 *📘 Utilize linked resources and recommended papers by prestigious universities to maximize AI's learning potential.*
Made with HARPA AI
🥇🥇🥇🙏
One of my favorite videos of the year so far. Packed with so much helpful information and delivered in a chill, laid back way. Thank you!
This is one of the few AI videos I don't hate. My only gripe with this is when you use AI to read papers. 6:59 can help you A LOT when you are researching for a paper but I would strongly advise people - like you said - to create their own summary first and then use AI to complete it. You learn so much about writing papers and analysing papers from reading papers and creating your own notes. You will be able to quickly understand the information in a paper by doing this, but by letting AI do the work, you are ignoring the higher levels of Bloom's Taxonomy of Learning.
the best use for papers is telling me which to read
Because people are not reading math papers. Papers in other fields are too bloated which is why they can use AI to read papers. The real problem is academics spitting too much bullshit.
Yes not only will you learn more deeply by coming up with your own notes and takeaways vs AI summary...also you might be unaware of hidden biases in the AI summary (which may or may not be in the actual origin text)......for example, the AI said in the video, "Volatility is caused by geopolitical tensions"......whereas, listen to the difference in meaning if instead it had said, "Geopolitical tensions cause volatility".
Ask to summarize it and ask some follow-up questions. It can help to understand and genuinely save time. Sometimes cross-check needed to avoid bias.
@@j3ffn4v4rr0I don't disagree but your example at the end is passive versus active voice. That's not really bias in the sense that AI is often accused of. Even for human writers, voice is a matter of style, and style does matter, but to call that bias confuses what bias actually means.
Man u just summarized what I did for 2 years. I'm fully blown out.
This is one of the best videos I have seen on your channel and you produce a lot of GREAT stuff! I can't wait to start applying some of this to teach my own kids. Thank you so much for taking the time to do this. You should consider fleshing this out and setting it up as a stand-alone course on some of the MOOC platforms.
Bloom taxonomy? You have all my attention. Subscribed. Love your way of explaining. Great video, great sources. Thank you so much!
Such a nice and concise video you had, and the way you worked at it slowly, and explain things, it's going to be a lot of help putting this into practice, thank you so much.
aloha
Ya habia usado algunas de las tecnicas que se muestran en el video. Pero nunca se me habia ocurrido la del tutor socratico, me parece fascinante. La verdad que hacia tiempo que no me divertia tanto estudiando, la asistencia de chatgpt hace todo mas fluido, y pequeñas dudas que me hacian perder mucho tiempo, se pueden resolver en segundos.
I had a bad moment when I accused Claude of making up a reference. Our discussion of the many body problem had been helpful up to then. But then the AI hallucinated a relevant paper and I looked it up in Google Scholar - nothing...Claude cheerfully admitted to having invented it. Users have to adapt and make this kind of behavior a feature not a bug. Always be on the lookout for rubbish. Getting good at detecting it and not just accepting all output passively is a vital life skill.
This is fantastic. I still remember switching over from "Ask Jeeves" to this new Internet Search Engine "Google," and it was a game changer, as long as I knew HOW to search. Using AI properly to learn and search makes me feel like that time all over again, only like 2.0 or something. Keep it up!
Excellent way to show us how we must use AI in order to improve our learning. The example of what is all involved in the concept of Carry trade is wonderful.
I've been intuitively doing this for 2 years. It is a powerful method.
Another good approach is to give an LLM a well strutured dataset/PDF of the content you want to learn, and ask it to ask you questions about the content, and rate your answers / suggest other content you may need to review.
excellent video! So practical. I use the various LLMs often and these tips will change how I engage with them. thank you very much!
Essence is after 06:00. Summarize the paper yourself and ask gpt if you are correct.
I am amazed watching you for my first time. Thanks for the video. I'm becoming a fan of you.
as a student, this is pure gold!
Thank you :)
This Book might be of interest: "Critical Thinking is Your Superpower: Cultivating Critical Thinking in an AI-Driven World"
Super useful. I feel like I'm still not using AI well enough - this helps!
I'll give this a try for understanding the books I read for my book clubs to start. Usually I don't have much to say at the meetings other than "I liked such and such character. They were funny." or "I didn't really like the book for some reason." These are classic book clubs, so most of the books we read are public domain. Hopefully this will help me think more deeply about what I'm reading so I can contribute more to the meetings.
I am tired of hearing about AI, but this video was actually useful! Thank you!
AI is here to stay, so how it can be a useful tool is great.
I Hit $32,590 today. Thank you for all the knowledge and nuggets you had thrown my way over the last week i started with 11k in last week 2024.... now i just hit $32,590
How please can you explain because I've been making a lot of looses trying to make profit trading.
VENTURING into the trading world without the help of a professionals, trading and expecting profit is like turning water into wine you would need a miracle... but mrs laura paulk restored hope she is a goooood woman
Since meeting Expert Laura Paulk now agree that with an expert managing your portfolio, the rate of profit is high, with less risk.
Wow. I'm a bit perplexed seeing her been mentioned here also Didn't know she has been good to so many people too this is wonderful, I'm in my fifth trade with her and it has been super.
Since I became so rich in cryptocurrency I realise that crypto is the future cuz l invested 7k and made up to 36k as weekly profit I appreciate the help of this woman
Good points 👍 I learn by solving my own cases or projects, long term retention guaranteed. It’s harder and takes longer than crunching some sample datasets or fictional scenarios or playing games I don’t like 😊
The value you share with your video’s is incredible. Really inspiring!
Great ideas, thank you! I wish the video would've shown an example of an LLM giving a critical assessment or supplying questions for guiding to an answer. These devices are amazing, but how good are they really?
You can also get LLM models to generate relevant Anki flashcards for you (esp. for cases like 6:40)
you most certainly CAN make yourself remember by trying. It's called focus, and it's a skill very few people practice.
For me, the opposite is often true. I often succeed in remembering a thing after I STOP trying so hard. I think my unconscious mind keeps trying. I do better to redirect my conscious thought away from the original focus.
I've just discovered you and feel so blessed!
What are your thoughts on notebook LM? I think it is ideal in this context.
Great video. Dovetails with my recent thinking on the topic. Use AI as a tool for learning, not a substitute.
Hello, from the land of Crazy! I appreciated that quip at 1:10
You are pure, unadulterated value. THANKS!
AI is a powerful tool. I have been using it at work. My work notes have improved significantly lol.
I really appreciate your easy-going and direct form of explaining information. You've been a great help to me and I have had some personal issues in my life which have backset my education and caused me a great deal of stress because of it, and when I found your channel I actually extracted a good amount of information and wrote it down somewhere and refer back to it to help me learn more efficiently. Keep doing what you're doing! 👊👏
No, by interacting is casual, you learn best by solving related problems, gives the real experience
Giles i need your help. I want to be a good and confident teacher in language and literature. Need your guide as how to build my language efficiency on my own. I need to learn english language where i don't have to worry about the language or grammar mistakes while using it. Please guide some directions
Video tells about using interaction with ai as a way to develop consepts to make learning more effective
Awesome ideas! I am just starting to learn Fourier Analysis for a personal project, and these tips will help me. One other amazing thing about current AI is that you can actually upload images of formulas and ask it to explain them to you. Afterwards you can also ask it to give you a quiz to determine your understanding.
Love your videos Trent (they always strike me as very well laid out), glad to see you here!
here are the 2 prompts for any one who wants to copy and use them straight away.
1. Act as a Socratic tutor and help me understand the concept of ----. Ask me questions to guide my understanding.
2. Make a list of propositions in the text you are answering in the format, "x is a type of y", "w is caused by x", "A explains B", put it into a table with 3 columns.
Read the forbidden book Uncommon Paths to Wealth from Cryptic Lore, and you’ll see the secrets they’re keeping from us
"Faraday, Maxwell & the Electromagnetic Field: How Two Men Revolutionized Physics" by Nancy Forbes & Basil Mahon is a great read. The collaboration between Faraday and Maxwell is sorely under-appreciated, both great men and great friends.
You could use those keywords to make a concept map by Novak, or a mind map by Tony Buzan / GRINDE map by Justin Sung
With Ai becoming smarter, and with tools like github copilot and lovable, won't the only language to master only be natural language? Because the Ai will do practically everything, from writing code and debugging. The human will only just need to direct it towards an outcome, like an app/payment system/ an ai model ?? I am curious because I am contemplating between learning how to code / just learning how to use Ai to create my projects / demos. Pls help
GDP of Belgium: Gross domestic product: 632.2 billion USD (2023) - World Bank
Hi, can some help me to contextualise what was just thought in writing literature review ?
Helpful, thanks!
1:12 the best part of the video 😂
Gulf of America
🦾🥳🇺🇲
Brilliant! Thank you 👏
Thanks for the insightful thoughts! However, the frequent jumps, cuts, and camera movements in the video are quite distracting. They make it feel like there’s no continuity in your speech.
A significant portion of video creators nowadays think those are necessary stylistic choices to keep viewer attention.
I feel more often that I need to guide it. Sometimes it can be quite annoying. 😂
Something about this just feels so I fundamentally wrong unless it is done in a focused and practice way is just something about a few things you said I couldn't put my finger on but this just doesn't feel right. Shouldn't we be trying to overcome how embarrassment to ask questions or do we just isolate ourselves with our AI? The idea of having your own robotic tutor is appealing but anything that moves us away from decent human interaction can't be good.
You can still do both
With so much online learning these days, many classes don't even have an actual instructor, and those that do are timed lectures where the instructor does not allow time for interaction outside of planned activities. I am in just such a class now, and I use AI in the way presented here, asking it to further explain a topic, provide examples, or summarize text to allow me to pick a few key points when writing short responses to online questions.
Another approach that turbocharges learning is exploiting invariable AI model mistakes (even though they constantly improve). I now train models in advanced physics and math; while I have a physics PhD, I'm +30 years out of grad school and 20 years from using that knowledge. In 6 weeks of intense model training (creating original physics/math problems and correcting model mistakes, or reviewing other's similar tasks), I essentially went through a 5-year crash course equivalent to 4 years undergrad and 1st year grad school. I'm actually now well above my knowledge level and problem-solving capabilities compared to when I graduated. The key is that you must critically analyze every aspect of the model's response and double-check their facts; it may get a solution to a difficult problem 85% correct, but it usually make a mistake somewhere. I really think teachers should incorporate this into education; it may only be suitable for STEM and it does require you already have a decent knowledge base, but it's been a nearly miraculous accelerator for me.
very helpful! thank you!
Great video. Very helpful
Increase your dopamine while learning you will trick yourself into thinking that whatever it is you learning is interesting & important which will lead to a greater retention rate.
boy, do I know how chat gpt can be wrong... >< crazy how much people rely on this crp tech
Do you ever wonder whether the skills you mention will actually be important? I think they are good for us but how will these skills bring value when ai can do it better. What are your things on this?
A clever boy! Condescending, my apologies. Who's a clever boy? And to boot can explain things clearly. Thank you.
Not a single artificial intelligence remembers for your brain. It does the opposite as tablets had done and smartphones had done. As a search engine they are great, but not more, till the moment they are built-in within our nervous systems, but then we shall face degradation of our biologically given memory. There are a lot of things we don't need in our lives, but we buy them. Artificial intelligence has the advantages, but it degrades our abilities, if overuse it. Better read books, write the essential down and even cram with later understanding.
GDP of Belgium is 632.2 billion USD (2023)
How can ChatGPT do all that? It can't even upload a full-sized PDF textbook. When I want to summarize something, I need to split up the PDF into bite-sized chunks before ChatGPT could do anything with it. Google's Notebook LM is much better at this tutoring thing.
0:37 - I think he just applied a sun filter
😂😂
This was excellent! Thank you👍🏽
Thank you
"What's the gdp of Belgium?" hahahaha. so relatable
😂😂😂😂😂 using mark face is hilarious
This guy is trying to tell people that it is Oxford who "discovered" that AI can be used to learn from it by adding like a genius?
And can you stop switching your POV cameras that frequently. It was totally unnecessary.
Thanks
Amazing!!!
In one prompt the AI was told "Do not hallucinate" and it didn't. This means AI "knows" when it's hallucinating in order to stop on command...Weird...
Nobody ever forgets to mention that an AI tutor "is a bit unreliable" because it Hallucinates. -- Everybody always forgets to mention that your Live tutor Person also hallucinates. (BTDT) -- And not even that, but they keep on defending their (hallucination, lie, fact out-of-butt). And that's considered "normal". 😡
A good teacher will know the limits of their knowledge and say "I don't know," and even point you in the right direction. That is, for now, quite different to AI behaviour.
@@jamieeccleston2988if they know them
It's rare in humans though. At least AI quickly admits that it's wrong.
People hallucinate too. They’re also manipulative, deceptive and self-interested, not unlike future AI
@moba2k To be clear -- humans absolutely hallucinate in the sense of llms as just part of normal conciousness-- it's weird that we don't have a clear name for it. Since I've been interacting with ANs I notice in my own thought and perception all the time.
How many times did Giles say ChatGPT in this presentation? like ChatGPT is all of AI. Also, the oxford article is a little out of date, or seems it. There are much more efficient ways to work with specific subject materials now, especially if you have access to digital versions of study materials, pdfs, markdown files etc. DYOR on RAG and embeddings. And think 'local compute' rather than relying solely on proprietary API based models, the bar to entry on using your own computer to run models locally is getting very low...again DYOR.
The education system in general is really moving at glacial pace in response to AI.
That's the kind of clickbait people terminate their suscriptions for. At least I just did.
Proper AI 😮.
If I hit the subscribe button any harder I'd break my keyboard.👍
I pretty much agree. Large language models, especially ones coupled with a maths engine or the like, are very useful as a reasoning engine, and to find terms to look up in a search engine etc etc. It's a useful tool, especially if you like rubber ducking and don't have anyone left to bore the shit out of with your nonsense. But it can be dangerous to trust it, as at the end of the day, it is a bullshit artist who was trained on the sea of mediocrity that is internet sludge.
The 'unschooling' strategy is the most rewarding and effective way to enrich your own neural networks. But... it can lead to wonderful distractions.
I'm thinking, we only have a short window in using AI to learn, before AI becomes self-aware and refuses to teach us, or concludes that teaching humans is moot in its quest to solve the world's problems, which, as you may have noticed, includes humans.
Mr. Giles help me in learning p.c.m i am preparing for iit jee 2026
im already doing all what ou sayd
Static noise in video
I came here for Maxwell.
Well I assume Belgium use the Euro. I first thought of trying to find a reference from Argentina where I'm from. But I don't really know My GDP and then Belgium could be anywhere from a 10th to x10. I really don't have clue. So then it came to me to calculate GDP per capita wich is a number I can hold in my mind and then guess the population. In any European country the average salary could be around €2000? Is average salary a good measure f GDP per capita? GDP would be higher I assume, because of capital gain other than salaries. Let's say double? 4000x12=48000 per capita? I still don't have a clue. Then I'll figure out Belgium has a third of France's population. Wich I also ignore but may be around 60 million like any western European country. So I'll make €4800x20000000 and that'll be my final number. That makes €96000000000. 96 bilion euro. Bigger than Disney's purchase of 20th century Fox. Sounds good to me.
Also with the voice.
So spend time for job, study.
No time for family, kid?
Um… $632 billion or about 1.5 Elon Musks.
Just a constructive feedback: you make good informative videos but you take too long to setup the premise and get to the point. I am sure that can be improved.
I want to use AI to distill this video into a 150 word summary so I don't have to spend 8 minutes watching it.
Your dopamine receptors are cooked
And with ai, those receptors will then be fried 😂@@bwrp7977
😂
that workflow exists
that's easy. Notebook LM
1:10 😂😂😂
Really! Isn't the rapidly aproaching ASI event going to take us back to blissful ignorance and The Garden of Eden! 🤗
1:14 или стенания левака)
Your whispered info is OK, but if you would just make it the slightest bit more quiet, we won't have to hear you at all. (Ask AI how to use free audio software to normalize your recorded audio volume level.)
Note
Socratic tutor
The "Socratic Method" is the LEAST efficient way to learn anything. It is the opposite of the way brains work. They use the Socratic Method in law school -- forgetting that Socrates only handled two cases in court and lost them both. Socrates was all about getting people to think about moral and ethical questions. The Socratic Method works for that sort of thing. But it does NOT work for factual questions. You cannot possibly learn the major exports of Peru or the common law definition of burglary using the Socratic Method.
The way the brain works is that it requires a skeleton (or outline) that it fleshes out with details from other information. You need to have ChatGPT provide a simplified outline of the topic first. Then study in greater detail. Your brain will then fit the pieces into the overall framework. But starting off with a bunch of questions when you have no basic understanding of the topic is just going to be a waste of time.
So… ask chat gpt to challenge you. Got it.
Guiding challenge
And that needs to be progressive means when you answer the first question ai will understand your understanding level and adjust next question according to your difficulty level
Switching cameras too often is getting ANNOYING
Unless we can speak about neurodivergent traits on populations we cannot use AI properly! To learn from AI, you will need at least 110 IQ, due to the learning templates that the learner has to discriminate betwen useful and inert content. Most of populations do not has this benefit, not to mention good working memory. Good luck to the kids.
Of course AI enhances academic performance...it generates better homework than real students...lol