I actually really like this, as I always read better with my ears, and the human quality of the current AI voices makes it so much easier to listen to than previous TTS.
Yep! ElevenLabs. I've honestly kind of spoiled myself and struggle with other audio solutions now. I would love if various bloggers embraced this tech to create this quality of audio versions of all their posts. I would love to work with any who want any assistance on making this a reality.
chatGPT is the best general tool but perplexity AI is a better online search tool. Show some love to the other language models please. I am sure Google will be able to challenge chatGPT once they figure out to leverage all that data they have on us already.
I just pre-ordered your book 📕. Thanks for all you great insights.
Yesterday, I wrote a post on Linkedin. I spoke about the use of AI in class after a discussion with a professor. According to him, Most of the students use AI to do the work and not just to assist them. This could led to a cognitive disaster. I wrote the post to share my discussions. I finished with : « It is urgent that the education system draws up a list of timeless knowledge that will allow a young human to be an enlightened citizen in the age of AI. »
It’s a big question and i’m expert enough to answer. No one knows, but i think you know better than other. So I would be very interested to have your point of view..
happy new year ! And again, thanks for your great work.
I would add that the technology for using AI is going to improve. It is like a smart phone with hardly any apps available. I may be wrong, but I expect that the GPT store is going to accelerate adoption. If not, then something else will.
Word on the street is that Chris Hayes might be interested in talking about how AI might impact education. He says "there's a bunch of interesting questions here that I don't feel like I have super clear answers to".
I think the education aspects are even wilder. There's a lot of creativity at work.
At the same time, academia is wrongfooted. Most campuses don't have AI policies now, and punt decisions to departments and individual faculty. We only have hints about just how many students are using AI.
Ethan, you're so good at condensing what I'm thinking into a really great guide, translating it so that it's easy enough for laypeople to follow. I appreciate all the effort that goes into this!
I always love your posts, Ethan. They're the only ones I read in full.
Just one question:
You mention "... most companies have decided to treat it as some sort of standard knowledge management tool, a task that LLMs are not actually that good at."
What makes you say that? LLMs seem great at aggregating large areas of knowledge and making them accessible to NLQs.
Anyone worried about the cost of energy for the required compute units in the coming AI nirvana can rest assured that, any day now, electricity will be too cheap to meter.
I'm curious how you respond to those who are far less optimistic about the transformative impacts of generative AI, especially with respect to education. I watched the recorded presentations from HILT on YouTube and Justin Reich in particular singled out your remarks as very similar to predictions that have been made in the past regarding transformative edtech developments which fell far short of the anticipated expectations. I think you and Justin are talking about some different issues, but he is much, much less bold in his analysis of the impact of generative AI in K-12 classrooms based on the history of similar hype around previous technologies. I'm interested where you see the differences in your views and why generative AI may be different. He was fairly dismissive of the notion that AI was going to democratize education in any meaningful way.
Ooh I think I know this one! I believe the response is simply that the edtech developments of the past were not LLMs. If I understand correctly - which I probably don’t, but felt compelled to reply anyway - we are in uncharted territory here.
These previous edtech products referred to couldn’t, for instance, pass not only the Turing test in a way never seen before, not to mention gpt-4 specifically is able to pass the bar, mcats, etc in the top 90 percentile.
Gpt-4 is available to everyone in the world through Bing, which looks like unprecedented access to such powerful software.
But, and it’s a big one, I am also always skeptical of my own predictions when it comes to the future, so I get where reich is coming from (I have not heard his comments myself, just responding to what you’ve described to me so I may be off base as well with that)
Phew, that was more than I planned on writing. Steve, I want to apologize to you and anyone else if you’ve made it this far. Sorry for wasting your time lol
I tend to agree that there may be something about the new LLM's and AI in general that may be different than past edtech innovations, but Reich has a lot of compelling arguments that schools, especially K-12, are simply not equipped nor especially good at integrating new technologies into their curriculums and nothing like the kind of transformative predictions of the "techno-optimists" are likely to take place. His view is definitely the opposite of Ethan's - in fact, in the video he specifically says that he would bet Ethan on likely outcomes - and he is much more of a techno-pessimist. His view throws a lot of cold water on the extended hype you see surrounding the generative AI predictions, at least with respect to K-12 education. He does not see ChatBots as doing much with respect to changing day to day work by teachers nor does he think it will be of much appeal to students. Check out his remarks at the HILT conference which you can find here: https://www.youtube.com/embed/W3Hghx54_js?si=tnpHM2HoxeWOrrEn
Thank you Steve for the thoughtful reply! I will have to check out the video. So before I say too much more without watching and making my usual irrelevant or inaccurate remarks, I will completely agree k-12, and schools in general are *awful* at adopting and integrating new technology in any timely fashion - especially when not compelled to do so by a major cause that would justify the expense out of their already stretched budgets.
I’d be curious what the transition/adoption rate was for k-12/higher during the personal pc era of the early 90s when computers first became household staples akin to televisions replacing radios. Because this seems more like we are talking about the scale of how disruptive those were, than we are talking about a new version of PowerSchool or ParentSquare.
Then again I also have no idea what I’m talking about and I’ve gone and said a bunch without watching the video. I’m enjoying discussing it with you though, so thank you! It’s rare for me to be able to have a nice conversation about this stuff :)
But yeah it feels more like we’re going to be moving very quickly on the “up” part of the s-curve of technological advancement with this one!
On the topic of "Hearing is no longer believing", here is you narrating this entire post:
https://soundcloud.com/thomas-askwho-askew/signs-and-portents
I actually really like this, as I always read better with my ears, and the human quality of the current AI voices makes it so much easier to listen to than previous TTS.
What are you using to create these, if you don't mind my asking - ElevenLabs? The quality is superb!
Yep! ElevenLabs. I've honestly kind of spoiled myself and struggle with other audio solutions now. I would love if various bloggers embraced this tech to create this quality of audio versions of all their posts. I would love to work with any who want any assistance on making this a reality.
I am looking forward to this being applied to famous authors (whose voices were recorded) reading their own books. Like Mark Twain reading Tom Sawyer.
Another brilliant and insightful post Ethan. I look forward to each one of these. Keep em' coming in 2024. You're an inspiration.
chatGPT is the best general tool but perplexity AI is a better online search tool. Show some love to the other language models please. I am sure Google will be able to challenge chatGPT once they figure out to leverage all that data they have on us already.
Dear Ethan,
I just pre-ordered your book 📕. Thanks for all you great insights.
Yesterday, I wrote a post on Linkedin. I spoke about the use of AI in class after a discussion with a professor. According to him, Most of the students use AI to do the work and not just to assist them. This could led to a cognitive disaster. I wrote the post to share my discussions. I finished with : « It is urgent that the education system draws up a list of timeless knowledge that will allow a young human to be an enlightened citizen in the age of AI. »
It’s a big question and i’m expert enough to answer. No one knows, but i think you know better than other. So I would be very interested to have your point of view..
happy new year ! And again, thanks for your great work.
Amazing. Need to read it another three time to make sure I haven't missed anything. Should be compulsory reading for anybody interested in AI
I would add that the technology for using AI is going to improve. It is like a smart phone with hardly any apps available. I may be wrong, but I expect that the GPT store is going to accelerate adoption. If not, then something else will.
Often the change due to a new tech (here AI) comes in unexpected areas.
Let us consider the Mobile Aloha project at Stanford.
https://mobile-aloha.github.io/
It is a cheap hardware and open source software project
where the learning is by observing and imitating.
*This* is the future.
But it is a future imperfect.
https://www.youtube.com/watch?v=xGNNW6smDPQ
Fantastic Article, thanks Ethan! 🫶
„As have studies on writing, programming, and innovation.“ Could you cite these studies as well? 🙏
I have posted about them a lot here - https://www.oneusefulthing.org/p/automating-creativity
Excellent paper, thanks Ethan
Word on the street is that Chris Hayes might be interested in talking about how AI might impact education. He says "there's a bunch of interesting questions here that I don't feel like I have super clear answers to".
https://www.msnbc.com/msnbc-podcast/why-is-this-happening/answering-your-questions-our-withpod-holiday-mailbag-podcast-transcript-n1308086
Professor Mollick, congratulations on the new book!
Would you like to appear on my video program in April or May, to discuss and promote it?
More info here: https://forum.futureofeducation.us/
Good survey.
I think the education aspects are even wilder. There's a lot of creativity at work.
At the same time, academia is wrongfooted. Most campuses don't have AI policies now, and punt decisions to departments and individual faculty. We only have hints about just how many students are using AI.
Ethan, you're so good at condensing what I'm thinking into a really great guide, translating it so that it's easy enough for laypeople to follow. I appreciate all the effort that goes into this!
I always love your posts, Ethan. They're the only ones I read in full.
Just one question:
You mention "... most companies have decided to treat it as some sort of standard knowledge management tool, a task that LLMs are not actually that good at."
What makes you say that? LLMs seem great at aggregating large areas of knowledge and making them accessible to NLQs.
I believe it’s the fact that they’re still error prone when you try to add something to it to talk about/with, like a PDF or company knowledge base.
Anyone worried about the cost of energy for the required compute units in the coming AI nirvana can rest assured that, any day now, electricity will be too cheap to meter.
I'm curious how you respond to those who are far less optimistic about the transformative impacts of generative AI, especially with respect to education. I watched the recorded presentations from HILT on YouTube and Justin Reich in particular singled out your remarks as very similar to predictions that have been made in the past regarding transformative edtech developments which fell far short of the anticipated expectations. I think you and Justin are talking about some different issues, but he is much, much less bold in his analysis of the impact of generative AI in K-12 classrooms based on the history of similar hype around previous technologies. I'm interested where you see the differences in your views and why generative AI may be different. He was fairly dismissive of the notion that AI was going to democratize education in any meaningful way.
Ooh I think I know this one! I believe the response is simply that the edtech developments of the past were not LLMs. If I understand correctly - which I probably don’t, but felt compelled to reply anyway - we are in uncharted territory here.
These previous edtech products referred to couldn’t, for instance, pass not only the Turing test in a way never seen before, not to mention gpt-4 specifically is able to pass the bar, mcats, etc in the top 90 percentile.
Gpt-4 is available to everyone in the world through Bing, which looks like unprecedented access to such powerful software.
But, and it’s a big one, I am also always skeptical of my own predictions when it comes to the future, so I get where reich is coming from (I have not heard his comments myself, just responding to what you’ve described to me so I may be off base as well with that)
Phew, that was more than I planned on writing. Steve, I want to apologize to you and anyone else if you’ve made it this far. Sorry for wasting your time lol
I tend to agree that there may be something about the new LLM's and AI in general that may be different than past edtech innovations, but Reich has a lot of compelling arguments that schools, especially K-12, are simply not equipped nor especially good at integrating new technologies into their curriculums and nothing like the kind of transformative predictions of the "techno-optimists" are likely to take place. His view is definitely the opposite of Ethan's - in fact, in the video he specifically says that he would bet Ethan on likely outcomes - and he is much more of a techno-pessimist. His view throws a lot of cold water on the extended hype you see surrounding the generative AI predictions, at least with respect to K-12 education. He does not see ChatBots as doing much with respect to changing day to day work by teachers nor does he think it will be of much appeal to students. Check out his remarks at the HILT conference which you can find here: https://www.youtube.com/embed/W3Hghx54_js?si=tnpHM2HoxeWOrrEn
and his book, Failure to Disrupt.
Thank you Steve for the thoughtful reply! I will have to check out the video. So before I say too much more without watching and making my usual irrelevant or inaccurate remarks, I will completely agree k-12, and schools in general are *awful* at adopting and integrating new technology in any timely fashion - especially when not compelled to do so by a major cause that would justify the expense out of their already stretched budgets.
I’d be curious what the transition/adoption rate was for k-12/higher during the personal pc era of the early 90s when computers first became household staples akin to televisions replacing radios. Because this seems more like we are talking about the scale of how disruptive those were, than we are talking about a new version of PowerSchool or ParentSquare.
Then again I also have no idea what I’m talking about and I’ve gone and said a bunch without watching the video. I’m enjoying discussing it with you though, so thank you! It’s rare for me to be able to have a nice conversation about this stuff :)
But yeah it feels more like we’re going to be moving very quickly on the “up” part of the s-curve of technological advancement with this one!