What do you think?
Rate this book
Brave New Words: How AI Will Revolutionize Education
From the founder of Khan Academy, the first book written for general audiences on the AI revolution in education, its implications for parenting, and how we can best harness its power for good.
Whether we like it or not, the AI revolution is coming to education. In Brave New Words , Salman Khan, the visionary behind Khan Academy, explores how artificial intelligence and GPT technology will transform learning, offering a roadmap for teachers, parents, and students to navigate this exciting (and sometimes intimidating) new world.
An insider in the world of education technology, Khan explains the ins and outs of these cutting-edge tools and how they will revolutionize the way we learn and teach. Rather than approaching the sea of change brought on by ChatGPT with white-knuckled fear, Khan wants parents and teachers to embrace AI and adapt to it (while acknowledging its imperfections and limitations), so that every student can complement the work they're already doing in profoundly new and creative ways, to personalize learning, adapt assessments, and support success in the classroom.
But Brave New Words is not just about technology—it's about what this technology means for our society, and the practical implications for administrators, guidance counselors, and hiring managers who can harness the power of AI in education and the workplace. Khan also delves into the ethical and social implications of AI and GPT, offering thoughtful insights into how we can use these tools to build a more accessible education system for students around the world.
Whether we like it or not, the AI revolution is coming to education. In Brave New Words , Salman Khan, the visionary behind Khan Academy, explores how artificial intelligence and GPT technology will transform learning, offering a roadmap for teachers, parents, and students to navigate this exciting (and sometimes intimidating) new world.
An insider in the world of education technology, Khan explains the ins and outs of these cutting-edge tools and how they will revolutionize the way we learn and teach. Rather than approaching the sea of change brought on by ChatGPT with white-knuckled fear, Khan wants parents and teachers to embrace AI and adapt to it (while acknowledging its imperfections and limitations), so that every student can complement the work they're already doing in profoundly new and creative ways, to personalize learning, adapt assessments, and support success in the classroom.
But Brave New Words is not just about technology—it's about what this technology means for our society, and the practical implications for administrators, guidance counselors, and hiring managers who can harness the power of AI in education and the workplace. Khan also delves into the ethical and social implications of AI and GPT, offering thoughtful insights into how we can use these tools to build a more accessible education system for students around the world.
272 pages, Hardcover
Published May 14, 2024
About the author
Salman Khan
137 books51 followersLibrarian Note: There is more than one author by this name in the Goodreads database.
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 30 of 213 reviews
January 23, 2024
Much of the material written about the impact of artificial intelligence on higher education has been filled with overly pessimistic doom and gloom. Brave New Words presents a breath of air in that sense, clearly explaining how the Khan Academy non-profit has partnered with OpenAI in order to offer the best possible integration of AI and education. How they're approaching it is definitely admirable and I'm happy to see someone is taking a positive approach to this.
If all you've seen in the headlines are negatives about AI and education, this book will provide a very welcome alternative view. In short, I actually think every educator should read this book, considering the educational role that Khan Academy plays now and will likely continue to play in the future as it adopts AI.
Nonetheless, this book is so optimistic about how widespread AI use can be and the number of problems it will tackle, that this actually raised new concerns for me. These go far beyond the typical concerns about plagiarism and replacing teachers that have been written about most frequently. I'll give a few examples.
1. Khan sees this technology being used not just in the classroom, but as a way to help parents monitor their children's progress, offer therapy, and even facilitate conversation among families. These are helpful, it's argued, because of the anxiety of speaking in class, or the difficulty of getting our kids to tell us about their day. The book only briefly notes the negative of more screen time here, but there has already been serious research that raises concerns far beyond that. Sherry Turkle, for example, has written in the book Alone Together about how when we converse with machines precisely because it is easier, we lose the important skills that are needed to communicate face-to-face with others. If we can't converse with our own kids without AI intervention, we have lost something significant, not gained something. Meta (and other researchers) are already working on designing glasses that can listen to your conversation and then tell you how to respond to those with whom you're speaking.
2. This book also sets up some straw men when discussing why people don't like standardized testing. The main concern isn't about the actual standardization, it's the way it has led to explicitly teaching to the test, and how student promotion to the next grade level, teacher evaluations (and their associated employment), and school district funding have all been linked to the outcomes of these tests. Despite that oversight, Khan does suggest some approaches to adopting AI that would mitigate at least some of these challenges.
3. The vision presented for the future of the job application and hiring process is rosier than I could possibly ever imagine, and doesn't note the ways AI has already made this process dystopian in the present. Already, we have big data and AI finding absurd correlations, such as the fact that people who submit applications using a web browser installed by default tend to leave a job more quickly than those who install a different browser. We also have AI programs that analyze applicants' social media presence and the content of that to create reports about their predicted personality and leadership skills. This goes beyond the advice to avoid posting pictures of yourself drinking or even avoiding being on social media (that was so 2010s). Instead, it means that to have a better chance in being hired, you must have a social media presence, you must positively interact with a network of others in a publicly visible way, and your writing patterns must match some personality profile a black-boxed AI has determined will mean you're a team player (this review probably doesn't match that). But even these already existing uses of AI are not addressed.
4. Near the end, Khan argues the invention of generative AI means we're either heading to a Star Trek like utopia or a populist hell. I've been re-reading Fahrenheit 451 with my kids. I'll juxtapose two quotes.
"It [generative AI] could reformulate the news article they are reading closer to their grade level, potentially leaving out age-inappropriate details." -- Brave New Words, Salman Khan
"Classics cut to fifteen-minute radio shows, then cut again to fill a two-minute book column, winding up at last as a ten- or twelve-line dictionary resume... Many were those whose sole knowledge of Hamlet... was a one-page digest." - Fahrenheit 451, Ray Bradbury
To conclude, I do think this is a book everyone should read. I'm in awe of Khan's optimism, and it's worth hearing how he describes his vision. At the same time, it worries me that someone who spends so much time thinking about this has either skipped over or chosen not to discuss some of the serious challenges (not fluffy think pieces) that are already present in our world due to AI.
If all you've seen in the headlines are negatives about AI and education, this book will provide a very welcome alternative view. In short, I actually think every educator should read this book, considering the educational role that Khan Academy plays now and will likely continue to play in the future as it adopts AI.
Nonetheless, this book is so optimistic about how widespread AI use can be and the number of problems it will tackle, that this actually raised new concerns for me. These go far beyond the typical concerns about plagiarism and replacing teachers that have been written about most frequently. I'll give a few examples.
1. Khan sees this technology being used not just in the classroom, but as a way to help parents monitor their children's progress, offer therapy, and even facilitate conversation among families. These are helpful, it's argued, because of the anxiety of speaking in class, or the difficulty of getting our kids to tell us about their day. The book only briefly notes the negative of more screen time here, but there has already been serious research that raises concerns far beyond that. Sherry Turkle, for example, has written in the book Alone Together about how when we converse with machines precisely because it is easier, we lose the important skills that are needed to communicate face-to-face with others. If we can't converse with our own kids without AI intervention, we have lost something significant, not gained something. Meta (and other researchers) are already working on designing glasses that can listen to your conversation and then tell you how to respond to those with whom you're speaking.
2. This book also sets up some straw men when discussing why people don't like standardized testing. The main concern isn't about the actual standardization, it's the way it has led to explicitly teaching to the test, and how student promotion to the next grade level, teacher evaluations (and their associated employment), and school district funding have all been linked to the outcomes of these tests. Despite that oversight, Khan does suggest some approaches to adopting AI that would mitigate at least some of these challenges.
3. The vision presented for the future of the job application and hiring process is rosier than I could possibly ever imagine, and doesn't note the ways AI has already made this process dystopian in the present. Already, we have big data and AI finding absurd correlations, such as the fact that people who submit applications using a web browser installed by default tend to leave a job more quickly than those who install a different browser. We also have AI programs that analyze applicants' social media presence and the content of that to create reports about their predicted personality and leadership skills. This goes beyond the advice to avoid posting pictures of yourself drinking or even avoiding being on social media (that was so 2010s). Instead, it means that to have a better chance in being hired, you must have a social media presence, you must positively interact with a network of others in a publicly visible way, and your writing patterns must match some personality profile a black-boxed AI has determined will mean you're a team player (this review probably doesn't match that). But even these already existing uses of AI are not addressed.
4. Near the end, Khan argues the invention of generative AI means we're either heading to a Star Trek like utopia or a populist hell. I've been re-reading Fahrenheit 451 with my kids. I'll juxtapose two quotes.
"It [generative AI] could reformulate the news article they are reading closer to their grade level, potentially leaving out age-inappropriate details." -- Brave New Words, Salman Khan
"Classics cut to fifteen-minute radio shows, then cut again to fill a two-minute book column, winding up at last as a ten- or twelve-line dictionary resume... Many were those whose sole knowledge of Hamlet... was a one-page digest." - Fahrenheit 451, Ray Bradbury
To conclude, I do think this is a book everyone should read. I'm in awe of Khan's optimism, and it's worth hearing how he describes his vision. At the same time, it worries me that someone who spends so much time thinking about this has either skipped over or chosen not to discuss some of the serious challenges (not fluffy think pieces) that are already present in our world due to AI.
May 20, 2024
I adore Khan Academy, and I understand why Sal is so excited about the possibilities for AI in education, but this entire book is basically just a really long ad for Khanmigo. And the biggest barrier to Khanmigo being useful that I can see is that, for it to work, kids have to engage in elaborate discussions with it. Will they? Probably not. As it stands now, Khanmigo isn't going to be able to compete with the dopamine-infused allure of social media.
I would have rated the book much higher if he'd addressed that, and posted transcripts of actual kids (instead of just himself) engaging with Khanmigo.
I would have rated the book much higher if he'd addressed that, and posted transcripts of actual kids (instead of just himself) engaging with Khanmigo.
June 7, 2024
A few good ideas, but basically an ad for Khan Academy.
May 31, 2024
Li dois livros seguidos sobre a febre da IA Generativa — o primeiro “Creativity Code” (2019), escrito por Marcus du Sautoy como resposta ao impacto do GPT-1, e “Brave New Words” (2024), escrito por Sal Khan, o guru que criou a Khan Academy, como resposta ao impacto do GPT-4 — e confesso ter ficado boquiaberto, não com os feitos do GPT, mas com o deslumbramento dos autores, principalmente no segundo caso, com o livro a funcionar como uma declaração de fé na IA Generativa, suportada diretamente pela Open AI e Bill Gates. Se acalentava dúvidas quanto ao efetivo impacto da IA Generativa fora do domínio criativo, o fervor apresentado por estes livros tornou bastante claro para mim que o rei vai nu.
EN: https://medium.com/@nzagalo/ia-genera...
PT: https://narrativax.blogspot.com/2024/...
EN: https://medium.com/@nzagalo/ia-genera...
PT: https://narrativax.blogspot.com/2024/...
June 14, 2024
While I broadly agree with Kahn’s points and positivity around the potential for AI to enhance education, I agree with the reviews saying this book is basically just promo for his company’s new AI tutor. I suppose it could be helpful to generate ideas for using AI in the classroom as an educator, but none of the examples were particularly novel to me.
June 8, 2024
At first, I thought this was a sales pitch for Khanmigo, but as I read this, I learned new educational uses for AI and a great argument of how AI as a tool can change the landscape of education by engaging students and teachers in a different way of learning. All educators and parents could benefit from reading this book. Khanmigo and AI is a way to close educational gaps. The book was thought-provoking and well worth the time invested. Great jon Salman Khan.
March 8, 2024
In Brave New Words, Khan Academy founder Salman Khan presents an optimistic and practical vision for how artificial intelligence, particularly large language models like GPT, will revolutionize education in the coming years. As a respected pioneer in educational technology, Khan draws on his deep expertise to persuasively argue that rather than fearing AI, educators should embrace it as a powerful tool to personalize learning, support teachers, and unlock the potential of every student.
Khan's book serves as an invaluable guide for anyone seeking to understand the implications of AI for learning - from parents and teachers to administrators and policymakers. He accessibly explains the core technologies underpinning this AI revolution and lays out specific ways they can be leveraged to provide each learner with adaptive instruction and feedback tailored to their unique needs, interests and pace.
While acknowledging that AI is still an imperfect technology, Khan compellingly makes the case that its thoughtful adoption in classrooms will enhance, rather than replace, human interaction and creativity. He envisions AI as a tool to empower teachers, inspire students, expand access to quality education, and equip learners with the skills needed to thrive in an increasingly digital future.
Beyond the classroom, Khan explores the broader societal implications of educational AI - from its potential to make hiring more meritocratic to the ethical considerations in its development and deployment. His insights provide a valuable roadmap for how we can proactively shape these technologies as an overwhelming force for good.
At its core, Brave New Words is a profoundly hopeful book, reflecting Khan's lifelong passion to harness technology to improve lives through learning. While educational AI undoubtedly brings new risks and challenges that require ongoing vigilance, Khan convincingly argues the opportunities for expanding educational access and efficacy are simply too great to ignore. For anyone who cares about the future of education, this book is essential reading.
Khan's book serves as an invaluable guide for anyone seeking to understand the implications of AI for learning - from parents and teachers to administrators and policymakers. He accessibly explains the core technologies underpinning this AI revolution and lays out specific ways they can be leveraged to provide each learner with adaptive instruction and feedback tailored to their unique needs, interests and pace.
While acknowledging that AI is still an imperfect technology, Khan compellingly makes the case that its thoughtful adoption in classrooms will enhance, rather than replace, human interaction and creativity. He envisions AI as a tool to empower teachers, inspire students, expand access to quality education, and equip learners with the skills needed to thrive in an increasingly digital future.
Beyond the classroom, Khan explores the broader societal implications of educational AI - from its potential to make hiring more meritocratic to the ethical considerations in its development and deployment. His insights provide a valuable roadmap for how we can proactively shape these technologies as an overwhelming force for good.
At its core, Brave New Words is a profoundly hopeful book, reflecting Khan's lifelong passion to harness technology to improve lives through learning. While educational AI undoubtedly brings new risks and challenges that require ongoing vigilance, Khan convincingly argues the opportunities for expanding educational access and efficacy are simply too great to ignore. For anyone who cares about the future of education, this book is essential reading.
June 21, 2024
Pretty much a long form advertisement for Khan Academy—which I’m cool with. Sal Khan is easily a top 5 answer for the “who would you have dinner with in history?” question. Forget the tech billionaires of the world. Khan is where it’s at.
And donate to Khan Academy!
And donate to Khan Academy!
May 24, 2024
The book is about all of the ways generative AI can help us improve education and how it is already doing it with Khanmigo.
I liked it, but let's see how well it ages. These things change quickly and what's true today may not be true in some weeks.
I liked it, but let's see how well it ages. These things change quickly and what's true today may not be true in some weeks.
May 26, 2024
This book is the epitome of why when it comes to learning about AI and quickly moving topics, articles are infinitely more relevant and insightful than books that take over a year to get published.
As a big fan of Khan Academy and the edtech space, I really wanted to like this. Unfortunately, the book employs a consistently humdrum formula: a topic, some prompts Sal finds mind-blowing in 2022 (but isn't to us, reading it in the gpt-4o era), some scary aspects, why those aspects are overblown, and very now-obvious future use cases of AI.
As a big fan of Khan Academy and the edtech space, I really wanted to like this. Unfortunately, the book employs a consistently humdrum formula: a topic, some prompts Sal finds mind-blowing in 2022 (but isn't to us, reading it in the gpt-4o era), some scary aspects, why those aspects are overblown, and very now-obvious future use cases of AI.
July 30, 2024
Khan’s thesis can be boiled down to: AI offers the potential for a personalized tutor for every student. He has an ambitious (and utopian) ideal of ‘every student having personalized support through their education using AI assistants.
Overall, I liked the book. Despite my criticisms, I would still recommend it to anyone working in/around education who wants to take a journey through of all the different influences and implications of AI on learners with a few whimsical ideations to keep us hopeful.
I admire where Khan is coming from and his ‘tech for good’ mentality that has made Khan Academy such an incredible resource for learners. He identifies through that with enough philanthropic and government investment, AI tools like Khanmigo (his new product, mentioned heavily throughout) could be a completely free resource for students. He outlines (with actual examples using generative AI different use cases for teachers and learners. The chapter on assessment and admissions gives us hope of better, more informative testing experiences for students – without plugging Khanmigo 🤣 and the chapter on work provides balanced perspectives on how jobs will change and tangible recommendations on how workers can adapt to remain relevant: be a centaur. Be a cyborg. Learn how to leverage these tools early, often, and effectively to advance your skills, but not replace them.
HOWEVER, read this with the massive caveat that, at times, the book just feels like a gigantic sales pitch for Khan Academy and Khanmigo. While I am generally an AI optimist, Khan’s takes in this book are simply too positive to take in whole. Other than at the end, where he heeds a warning of the “AI for evil” narrative we could fall into, this book lacks enough concern or critique. He glosses over significant fears and risks, almost to check a box that he’s mentioned them, and forges ahead with all of the potential AI (often his product specifically) has to offer.
Overall, I liked the book. Despite my criticisms, I would still recommend it to anyone working in/around education who wants to take a journey through of all the different influences and implications of AI on learners with a few whimsical ideations to keep us hopeful.
I admire where Khan is coming from and his ‘tech for good’ mentality that has made Khan Academy such an incredible resource for learners. He identifies through that with enough philanthropic and government investment, AI tools like Khanmigo (his new product, mentioned heavily throughout) could be a completely free resource for students. He outlines (with actual examples using generative AI different use cases for teachers and learners. The chapter on assessment and admissions gives us hope of better, more informative testing experiences for students – without plugging Khanmigo 🤣 and the chapter on work provides balanced perspectives on how jobs will change and tangible recommendations on how workers can adapt to remain relevant: be a centaur. Be a cyborg. Learn how to leverage these tools early, often, and effectively to advance your skills, but not replace them.
HOWEVER, read this with the massive caveat that, at times, the book just feels like a gigantic sales pitch for Khan Academy and Khanmigo. While I am generally an AI optimist, Khan’s takes in this book are simply too positive to take in whole. Other than at the end, where he heeds a warning of the “AI for evil” narrative we could fall into, this book lacks enough concern or critique. He glosses over significant fears and risks, almost to check a box that he’s mentioned them, and forges ahead with all of the potential AI (often his product specifically) has to offer.
July 27, 2024
"Brave New Words" explores the potential of Large Language Models (LLMs) to provide answers and frameworks, using Khanmigo as a key example. While the book offers practical insights, it may not present new information for those already familiar with the field and the limitations of LLMs, such as hallucinations and the probabilistic nature of their outputs. The frequent references to Khanmigo make the book feel somewhat like an advertisement, and the content could have been effectively conveyed in a one-hour talk and demo rather than an entire book.
The rapid pace of technological advancement means that parts of the book are already somewhat outdated, particularly given the emergence of new AI models with multi-modal capabilities, including voice, video, and image interactions. Despite this, the book presents several valuable ideas for improving pedagogy, tutoring, parent-child and child-teacher dynamics, and addressing mental health issues.
"Brave New Words" maintains a balanced perspective, considering both the advantages and disadvantages of technology. It effectively explains the complexities of measuring LLMs in contexts like hiring and discusses how these systems can be audited. The book is particularly valuable for its examples of how parents, students, and teachers can use technology effectively while preserving the human element. It also prompts readers to question existing biases and consider how AI can become more equitable and auditable, which is a positive takeaway.
The rapid pace of technological advancement means that parts of the book are already somewhat outdated, particularly given the emergence of new AI models with multi-modal capabilities, including voice, video, and image interactions. Despite this, the book presents several valuable ideas for improving pedagogy, tutoring, parent-child and child-teacher dynamics, and addressing mental health issues.
"Brave New Words" maintains a balanced perspective, considering both the advantages and disadvantages of technology. It effectively explains the complexities of measuring LLMs in contexts like hiring and discusses how these systems can be audited. The book is particularly valuable for its examples of how parents, students, and teachers can use technology effectively while preserving the human element. It also prompts readers to question existing biases and consider how AI can become more equitable and auditable, which is a positive takeaway.
August 17, 2024
With all due respect to Salman Khan for his achievements in education and EdTech, this is book is less about what it promises in the title, and more about the history and capabilities of Khanmigo, an AI assistant by Khan Academy.
June 19, 2024
I like it that Salman Khan is thinking about AI for education, because if I was to trust anyone’s words on this topic it would be his.
The book details some very interesting experiments Khan Academy did with ChatGPT, and by exploring which I have gone from fearful and confused to cautiously optimistic.
The book details some very interesting experiments Khan Academy did with ChatGPT, and by exploring which I have gone from fearful and confused to cautiously optimistic.
June 19, 2024
It's a good 101-level look at what's happening at the intersection of generative AI and education. I don't give it a 5 because it looks at this intersection solely through the lens of what Khan Academy and OpenAI are building individually and together instead of the industry more broadly. Still, since they are leaders in this space, the book's contents help inform the industry's current state and where it's heading.
Learning more about Salman Khan's background and journey to founding Khan Academy was great. Some education concepts were new to me and will help me be a better educator and learner.
I wish more time were spent on the risks of the technology and possible solutions for solving the challenges ahead.
Learning more about Salman Khan's background and journey to founding Khan Academy was great. Some education concepts were new to me and will help me be a better educator and learner.
I wish more time were spent on the risks of the technology and possible solutions for solving the challenges ahead.
June 14, 2024
Must read education book. I think Sal Khan is incredibly optimistic about the future of AI in education, but so much of the nuances are just not acknowledged tbh. His take on AI and “therapy” threw me off…. There’s some hot takes in here for sure, but overall loved learning about the ways in which the education system needs to change to adapt to ai.
April 25, 2024
I find this book to be insightful and beneficial. I do believe Khan is hyper optimistic but I also do not believe that's necessarily a negative thing as long as readers go into reading it with an open and critical mindset. This book is a great way to enter the discussion of AI and how it can work alongside educators and in the. I would definitely recommend reading it.
August 1, 2024
3.5? I have mixed feeling about this book. On one hand, it has valuable insight on the uses of AI in education and in general, digestible and implementable tips for many audiences, and discusses solid, well-conveyed ideas. I wish all educators took the time to read this or look into these ideas themselves— and parents and students, for that matter.
On the other hand, there were some points I am critical of, since I think the issues of AI in the classroom (and outside of it) are a bit more nuanced than the book gives them credit for, and I think it would have been better if it tackled AI from more perspectives. For example, while it considers key principles of learning and how AI can implement them to improve education, I think there are parts of the psychology of learning (for both child and adult learning) that AI has limitations in which would be valuable to consider(1). I think there are also possible issues it doesn’t delve into deeply enough(2), both informational and cultural(3).
Note: the numbers are footnotes, for examples.
Some people (understandably) complain that it’s one long ad for Khanmigo. While that is a reasonable complaint, I don’t think it takes much critical thinking to take valuable information from it if you look at the descriptions and processes of creating Khanmigo as some of the best-researched methods for using generative AI in GENERAL for education, and easily adapt them for your own pursuits, prompt engineering, or for creating an AI tutor of your own.
My main complaint is that there are parts that seem a bit TOO optimistic, especially since the author has an obvious incentive (promoting Khanmigo) to not consider the full nuance or possible issues of AI.(4) I am skeptical at just how effective some of the suggestions would be in practice(5).
In conclusion, I wasn’t sure how many stars to give this one; I feel like reading it as-it-is would be a different experience than reading it critically while actively looking for ways to make it useful, so it could be a 3 or 4 star book depending on your experience level, perspective/role going in, or the way read it.
I would still recommend it for anyone looking to improve their own AI utilization— especially educators. But approach it with your brain turned on.
—————————————————
EXAMPLES, since I want to be specific but my review is already too long.
(1) AI is not a person, and people will treat it as such. I am skeptical that (especially) kids will have just as much motivation and meaning attached to interactions with a computer as they would a teacher. We are social creatures, and I bet much of our learning as kids comes from the expectations and opinions of adults and peers, something that AI will not do.
Also, all the examples given were sample convos from the author himself, not actual kids. I want to see how THEY would actually use it.
(2) For example, one solution to “cheating” that Khanmigo uses is to not directly answer questions but to use leading questions to get the student to answer it themselves. While this is great in general, just reading the sample convos was already getting in my nerves because sometimes you just want to ask what a p-value is without entire five-page back-and-forth of metaphors and related topics to get you to answer it yourself. Personally, this would discourage me from consulting Kahnmigo with my questions much of the time. It just seems like AI takes it to an unhelpful degree that a human tutor/teacher would know not to do.
(3) For example, it explained how we can use AI to create “history simulations” or have conversations roleplaying as historical figures, like Williamsburg actors but better. This is cool, but I would be more careful about this; AI has the tendency to make things up and make assumptions rather than say “I don’t know,” and unless the historical figure has substantial writings, letters, and published opinions, I get the feeling a lot of what the AI says will not be direct thoughts from that person. And since it won’t necessarily cite its sources and point out what the figure did and did not say, people (especially kids) will likely take everything it says as fact and assume that that is indeed what the person would say, when in reality that information might be assumed or hallucinated to fill in gaps.
(4) in “Part II: Giving Voice to the Social Sciences,” it mentioned GPT’s tendency to hallucinate citations or web links. Khan claims to have fixed the issue in Khanmigo by not allowing it to give citations that haven’t been pre-approved.
However, I’m a bit critical about how that solves the issue.
Personally, when asking GPT to answer questions, I ALWAYS tell it to cite its sources so I can fact check at every step of the way and provide myself with more in-depth resources on the subject; AI providing a faulty source for something is a helpful indicator that I have to check out that information more thoroughly myself before talking it as fact. Removing sources would be harmful to me.
Congrats if you actually read this whole thing. Have a cookie 🍪
On the other hand, there were some points I am critical of, since I think the issues of AI in the classroom (and outside of it) are a bit more nuanced than the book gives them credit for, and I think it would have been better if it tackled AI from more perspectives. For example, while it considers key principles of learning and how AI can implement them to improve education, I think there are parts of the psychology of learning (for both child and adult learning) that AI has limitations in which would be valuable to consider(1). I think there are also possible issues it doesn’t delve into deeply enough(2), both informational and cultural(3).
Note: the numbers are footnotes, for examples.
Some people (understandably) complain that it’s one long ad for Khanmigo. While that is a reasonable complaint, I don’t think it takes much critical thinking to take valuable information from it if you look at the descriptions and processes of creating Khanmigo as some of the best-researched methods for using generative AI in GENERAL for education, and easily adapt them for your own pursuits, prompt engineering, or for creating an AI tutor of your own.
My main complaint is that there are parts that seem a bit TOO optimistic, especially since the author has an obvious incentive (promoting Khanmigo) to not consider the full nuance or possible issues of AI.(4) I am skeptical at just how effective some of the suggestions would be in practice(5).
In conclusion, I wasn’t sure how many stars to give this one; I feel like reading it as-it-is would be a different experience than reading it critically while actively looking for ways to make it useful, so it could be a 3 or 4 star book depending on your experience level, perspective/role going in, or the way read it.
I would still recommend it for anyone looking to improve their own AI utilization— especially educators. But approach it with your brain turned on.
—————————————————
EXAMPLES, since I want to be specific but my review is already too long.
(1) AI is not a person, and people will treat it as such. I am skeptical that (especially) kids will have just as much motivation and meaning attached to interactions with a computer as they would a teacher. We are social creatures, and I bet much of our learning as kids comes from the expectations and opinions of adults and peers, something that AI will not do.
Also, all the examples given were sample convos from the author himself, not actual kids. I want to see how THEY would actually use it.
(2) For example, one solution to “cheating” that Khanmigo uses is to not directly answer questions but to use leading questions to get the student to answer it themselves. While this is great in general, just reading the sample convos was already getting in my nerves because sometimes you just want to ask what a p-value is without entire five-page back-and-forth of metaphors and related topics to get you to answer it yourself. Personally, this would discourage me from consulting Kahnmigo with my questions much of the time. It just seems like AI takes it to an unhelpful degree that a human tutor/teacher would know not to do.
(3) For example, it explained how we can use AI to create “history simulations” or have conversations roleplaying as historical figures, like Williamsburg actors but better. This is cool, but I would be more careful about this; AI has the tendency to make things up and make assumptions rather than say “I don’t know,” and unless the historical figure has substantial writings, letters, and published opinions, I get the feeling a lot of what the AI says will not be direct thoughts from that person. And since it won’t necessarily cite its sources and point out what the figure did and did not say, people (especially kids) will likely take everything it says as fact and assume that that is indeed what the person would say, when in reality that information might be assumed or hallucinated to fill in gaps.
(4) in “Part II: Giving Voice to the Social Sciences,” it mentioned GPT’s tendency to hallucinate citations or web links. Khan claims to have fixed the issue in Khanmigo by not allowing it to give citations that haven’t been pre-approved.
However, I’m a bit critical about how that solves the issue.
Personally, when asking GPT to answer questions, I ALWAYS tell it to cite its sources so I can fact check at every step of the way and provide myself with more in-depth resources on the subject; AI providing a faulty source for something is a helpful indicator that I have to check out that information more thoroughly myself before talking it as fact. Removing sources would be harmful to me.
Congrats if you actually read this whole thing. Have a cookie 🍪
June 6, 2024
I think broadly I liked this - I agree with a lot of Sal's philosophy as it applies to AI and education. I'm not an educator, I was not raised in "the education system" so obviously my opinion is colored by that.
What I did not like was Sal's handwaving of issues that generative AI compounds. His stance is that AI could be equal to or _slightly better_ than the status quo. I think that's surprisingly naive? He's got a background in software, I'd expect him to understand the fact that problems are both more difficult to solve at scale and also that there are always problems at scale that aren't apparent in a smaller environment. Using AI broadly across society is the very definition of doing things at scale.
I agree AI is the future (increasingly its the present), that its going to destroy some things but mostly just radically change things. The quote Sal uses frequently in this book is "your job won't be taken by AI, it will be taken by someone using AI." I think that's reductive though, because it dismisses the fact that many people will lose their jobs through the mass adoption of AI. Mostly I think my issue with this is that Sal seems to be dismissing valid concerns because he's so excited about the future and the possible applications of AI. Which I get, it honestly is very exciting (while also being terrifying).
This is also at times basically an ad for KhanAcademy's Khanmigo. I don't think it's intended to be an ad, but more of a "look, this is a step in the direction we ought to be going and these are the kinds of things we can do with this technology" which is valid, so I didn't mind it as much.
One thing Sal mentions as a safeguard that I think is very interesting and also very likely to be used is using AI to fight AI in terms of cheating. Essentially integrating AI into the process and making it basically a proctor for assignments and classwork. I think its interesting, I don't know if we are there though. I do agree that the education system should be integrating AI, not merely fighting it, which I think is a big task with a lot of ambiguity. I'm not an educator though, I don't know what that will look like. As a parent of a very young child though, I know that I will need to teach my kid about how to navigate AI as much as I need to teach them to navigate the internet in general.
What I did not like was Sal's handwaving of issues that generative AI compounds. His stance is that AI could be equal to or _slightly better_ than the status quo. I think that's surprisingly naive? He's got a background in software, I'd expect him to understand the fact that problems are both more difficult to solve at scale and also that there are always problems at scale that aren't apparent in a smaller environment. Using AI broadly across society is the very definition of doing things at scale.
I agree AI is the future (increasingly its the present), that its going to destroy some things but mostly just radically change things. The quote Sal uses frequently in this book is "your job won't be taken by AI, it will be taken by someone using AI." I think that's reductive though, because it dismisses the fact that many people will lose their jobs through the mass adoption of AI. Mostly I think my issue with this is that Sal seems to be dismissing valid concerns because he's so excited about the future and the possible applications of AI. Which I get, it honestly is very exciting (while also being terrifying).
This is also at times basically an ad for KhanAcademy's Khanmigo. I don't think it's intended to be an ad, but more of a "look, this is a step in the direction we ought to be going and these are the kinds of things we can do with this technology" which is valid, so I didn't mind it as much.
One thing Sal mentions as a safeguard that I think is very interesting and also very likely to be used is using AI to fight AI in terms of cheating. Essentially integrating AI into the process and making it basically a proctor for assignments and classwork. I think its interesting, I don't know if we are there though. I do agree that the education system should be integrating AI, not merely fighting it, which I think is a big task with a lot of ambiguity. I'm not an educator though, I don't know what that will look like. As a parent of a very young child though, I know that I will need to teach my kid about how to navigate AI as much as I need to teach them to navigate the internet in general.
July 21, 2024
A tough one for me to judge.
In 2011 I (and more than 100,000 others) took an online course from the elite at Stanford in AI. It was an inflection point in my thinking on education. The course was taught at the same time as the regular university course and had the same homework and tests. It was indeed one of the best educational experiences in my life. Sebastian Thrun, one of the professors was so moved by the turn out that he left his tenured position and started Udacity. I have since taken a few more courses and predicted the end of the brick and mortar college. Clearly I was wrong, as these institutions still exist.
Sal Khan has a better idea, let AI be your tutor, mentor and life coach. Make learning more interactive, adapt to the abilities of each student. An AI scales way past what a single teacher can do. It potentially raises the floor and ceiling of each student.
It all sounds great, but I don't think AI is yet ready. My crystal ball is a bit fuzzy. I see it coming, perhaps even in my lifetime, but the optimism is a bit too much for me.
In 2011 I (and more than 100,000 others) took an online course from the elite at Stanford in AI. It was an inflection point in my thinking on education. The course was taught at the same time as the regular university course and had the same homework and tests. It was indeed one of the best educational experiences in my life. Sebastian Thrun, one of the professors was so moved by the turn out that he left his tenured position and started Udacity. I have since taken a few more courses and predicted the end of the brick and mortar college. Clearly I was wrong, as these institutions still exist.
Sal Khan has a better idea, let AI be your tutor, mentor and life coach. Make learning more interactive, adapt to the abilities of each student. An AI scales way past what a single teacher can do. It potentially raises the floor and ceiling of each student.
It all sounds great, but I don't think AI is yet ready. My crystal ball is a bit fuzzy. I see it coming, perhaps even in my lifetime, but the optimism is a bit too much for me.
August 15, 2024
Very good. It makes me want to look more into Khan, who created Khan academy so that education can be more accessible to all, due to the amount of benefit his work has had across the globe. This makes me want to do more research on global education and disparities in both America and abroad.
Absolutely crazy to think of how far AI has come in the past few years and it will continue to progress even quicker. AI tutors and collaborative teaching assistants are what Khan sees as the future of AI in education and in the near future at that.
I will say I think that while Khan points out the negatives that AI could have on society, overall he takes an extremely optimistic approach for individuals and society as a whole. He concludes the book recognizing that it can be seen as a race between good and evil when it comes to AI which further emphasizes the need for good people to utilize AI well.
A lot factors in of course.
Absolutely crazy to think of how far AI has come in the past few years and it will continue to progress even quicker. AI tutors and collaborative teaching assistants are what Khan sees as the future of AI in education and in the near future at that.
I will say I think that while Khan points out the negatives that AI could have on society, overall he takes an extremely optimistic approach for individuals and society as a whole. He concludes the book recognizing that it can be seen as a race between good and evil when it comes to AI which further emphasizes the need for good people to utilize AI well.
A lot factors in of course.
October 11, 2024
“Hey ChatGPT, can you write me a book of around 300 pages that could essentially be a pamphlet of unbridled optimism around the concept of AI and how it’s inevitable so we should all just get on board? If you come across any ethically murky or possibly negative areas to think about, acknowledge them really briefly and then move on without any in depth analysis of anyone’s concerns. Oh, and please please be sure to include the origin story for Khan Academy at least 3 times just so people remember how benevolent I really am.”
July 30, 2024
Not that this book is bad, it’s just not great. While I appreciate his futurism and excitement surrounding AI, I think that a more detailed examination of how the technology is being used would be more useful. This book however is still worth the read as these are issues that our children will be using.
July 1, 2024
This book is an ad for the authors own AI platform Khanmigo. Additionally it avoids thorny issues by saying we should we careful, but offers no practical solutions. All in all the only compelling reason the author gives to use AI is that everyone else will be and you shouldn’t be left behind.
July 12, 2024
This book felt like a giant advert for Salman Khan's foundation. Virtually every page mentions him, his company or his company's product. It would be nice to have more insight, especially as the book has such a generic title.
October 9, 2024
Pues este es el primer libro que habla sobre la IA Generativa de forma divulgativa y con pleno conocimiento de causa Poder leer a Khan y ver su perspectiva desde la educación, pero tocando las dudas de la mayoría de la población es muy gratificante.
June 1, 2024
Overall, Brave New Words is a timely and compelling read for anyone interested in the future of education and the role of AI in shaping it. - written by an AI
August 15, 2024
It is along advertisement for Kahnmigo but he sold me on it.
August 11, 2024
Not the most enthralling book but actually a really important look at how AI will affect education. The concept of AI as a personalized tutor becomes much more clear with the examples provided.
May 24, 2024
2nd half of the book was great
The beginning was meh
I’ll still give the book 5 stars because of the message of the second half of the book and the positivity for using ai in the future.
The first half was not bad, but he spent a lot of time plugging his khanmigo language model and how it can be used for each specific subject in school.
Overall, anyone that is using ChatGPT or similar models should read this book to get some ideas of how to better use it in school, in general, and in the future.
The beginning was meh
I’ll still give the book 5 stars because of the message of the second half of the book and the positivity for using ai in the future.
The first half was not bad, but he spent a lot of time plugging his khanmigo language model and how it can be used for each specific subject in school.
Overall, anyone that is using ChatGPT or similar models should read this book to get some ideas of how to better use it in school, in general, and in the future.
Displaying 1 - 30 of 213 reviews