Wow! Thank you so much 😊It's my pleasure! As an IP lawyer, patent agent & creator, I greatly appreciate that you've made a tool that lets me work with sources without submitting them for training. I've been using NotebookLM every day and can't wait to share more videos ✨Looking forward to seeing what other features you add!
Wow, thank you for the introduction to Notebook LM. I had taken a look at it but stopped short of doing anything with it. Your comprehensive and concise run-through of how to use this tool has given me all the knowledge I need to start using it. And for that, you get a subscription! Thank you 😊🙏.
My pleasure! I'm so glad you found it helpful 😊 Was there anything in particular that resonated & inspired you to use it more? Appreciate the subscription a lot 🫡
This NEEDS more exposure. I have found your channel recently and have given Obsidian a chance again but THIS?! Great insight and thank you for the info
My pleasure! Thanks so much for letting me know how helpful you find it 😊 So glad to hear that you're giving Obsidian a shot! As you can see in this video, I think it's going to be such a great pairing for AI, especially NotebookLM. Let me know if you have any questions! Feel free to share with anyone you think might find it helpful! That's the best way to get more people to see it 🫡
Obsidian is pretty great, but like any other notetaking-'2nd Brain' tool, not for everyone. But Notebook LM, this is just a whole other 'next level' of where you can take Obsidain.. or any of the better/similar tools. As you can see, even if you just dump all of your digital flotsam in Google Drive.. just as good. My head was spinning and sleepless night the first time I used Notebook LM, dreaming up possible use cases.
Sure! but it does depend. Notion, for example, uses proprietary markup language, which might make it impossible to add notes from Notion into AI like NotebookLM in the future. Future-proofing in software is underrated imo. For sure, Google Drive is going to be a key element for a lot of people. I think it's cool that it's an add-on to products people are already using. And I'm so glad to hear it! That sounds like my experience when I came across it too haha, there are so many possibilities ✨ Let me know if you have any questions about use-cases, and maybe I'll be able to work them into future videos 🫡
The podcast feature is quickly becoming one of my favorite features when doing research for personal learning projects. Especially since they added the ability to customize it. Pull all my sources together, throw them in, run the podcast generation, and then that starts giving me ideas for what to talk to the bot about, adding more notes, then I can regenerate the podcast for different angles and depth.
Yes! Exactly ✨ this is one of my favourite features as well. I spent all week exploring the podcast generations actually. Excited to make it a regular part of my creation process 🎙️
I go through dozens of videos like this every week and rarely do they achieve what this video has so clearly and compellingly. Well done, Callum! I've subscribed.
Wow! Thanks Manley, that means a lot 😊 what made this one so clear/compelling for you? I hope to bring those traits to other videos so would love to hear your thoughts! Thanks for subscribing 🫡
@@Wanderloots If your goal was to add value for your viewers, you hit the mark. Thorough, hype-free, enough detail but not too much detail. I also appreciated that you didn't focus solely on the flash (i.e., the podcast feature) but shared some really helpful and novel use cases for the tool.
@@ManleyWalker Thanks for sharing! Really appreciate the feedback 😊I think there are so many great use cases for AI, but people often get caught up in the flash, so I'm trying to go deeper where I can! Let me know if there's every anything you're looking for a video about :)
Callum, I appreciate your thorough, thoughtful, possibly overly logical 😆, methodical and analytical.. overview of Notebook LM, you did the single best explanation of this new tool I've found so far, been using it, it's game-changing. I hope you'll do more with this as I think your subscribers will find this fascinating as well as your own ideas on how to use for different purposes. Good one, Bro 🙏👍
Thank you 😊 appreciate you saying that a lot! Was there something that made this video stand out? I’m so glad to hear you’ve been finding it helpful! How have you been using it in a game-changing way? Also, I’m working on my ideas for the next part in the series so I am definitely doing more! If there’s something in particular you’d like to see, let me know 🫡
Thank you! Glad you found it helpful 😊 was there anything in particular you resonated with? Also, I hadn’t heard of heptabase! Thanks for sharing. Does it operate in markdown as well?
@Wanderloots especially analysis of longer RUclips videos and brainstorming. Also as another source of AI with access to current internet sources (like perplexity).
Cool! That makes a lot of sense, that’s mostly how I’ve been using it so far as well. What do you mean by bringing in perplexity? Like querying perplexity and adding the answers as a source to notebooklm?
@@Wanderloots Well I've been using it since two months and I found it great, especially while I'm studying IT and I need to search between a larg amount of literature and textbooks, so it become really helpful to use something like this great tool. Also, in terms of mathmatical notation, I used to write every thing on paper coz it's hard to write them using keyboard. Now I'm able to copy my notes directly into obsidian (which by the way I'm watching your vidoes about it and you did a great job, but it's still confusing me 😅) For NotebookLM, I think they need to add more features, such as organizing notes and enhance the creativity of the model and I hope they will do that.
@@aqeelsapeeh Glad to hear how helpful it's been! I use it all the time too, makes research so much easier. Completely agree that they could update in a few helpful ways. My guess is that they were testing its popularity, and with the massive success they've had, they're building it more into their core features (especially since they can monetize it now with NotebookLM plus, which will, in theory, give more funding to the team building it). Re: obsidian, happy to answer any questions on things you find confusing! I know it can be a lot to get started, but you can also start as simply as possible and build out as you need to, rather than trying to have the "perfect system" up front. I also have some explanations on wanderloots.xyz if you're looking for written tutorials to complement the videos :)
It's my pleasure! Thanks for letting me know 😊 I recommend (always) reading the privacy policy yourself before determining if it is "secure enough" for your use. I actually included a chapter in my last NotebookLM video on the privacy policy and my understanding of how to keep your use secure, so you can take a look at that for more information 🫡: ruclips.net/video/EsQ6sIAacEY/видео.html (note: none of this is legal advice)
Absolutely, my feelings exactly. tbh, I think we're already at that point of being unable to tell truth from fiction. AI-leveraged deep fakes have been able to almost (or truly) mimic many forms of information we consume. This is why, imo, we need better credibility systems for the internet, to better protect digital identities in a way that allows us to actually verify the source of information. I wrote an article on the topic of needing a safer Internet here if you're interested: wanderloots.com/recalibrating-35/ and also talk a bit about this in my digital identity video: ruclips.net/video/ESg-ZZ_odJM/видео.html Would love to hear your thoughts! 🫡
@@Wanderloots yes, the full coverage of the functionality of NobookLLM, the examples and the potential use cases. Honestly, it is worth rewatching (a few times) to brainstorm on how I (or anyone) could really apply it for ideas I have. the potential is enormous. One thing that would be great for another video is to expand on how it can be used as specialized tools. That is something I'm very interested in especially in the context of research -> content genaration/production. Thank you for the video and for your quick response.
@@briocheatete Awesome! Thanks so much for sharing your feedback 😊It really helps me know what people are most interested in. I actually just released a video using NotebookLM as a specialized tool for content generation: ruclips.net/video/4QJm_AptHF4/видео.html Would love to hear your thoughts, if this is what you were thinking of re: research --> content production. Hope it helps!
My pleasure! So glad you found it helpful 😊 was there anything in particular that jumped out at you? I'm always curious what people find the most interesting so I can focus on it in future videos 🫡
@@Wanderloots I am a Christian and I listen to 30-40 minute sermons on youtube. This tool will give a nice descriptive overview of the sermon points as well as the main scriptures that the preacher is talking about. I can refer this info even without having to sit through listening to most of the youtube video. So very helpful.
Amazing! I wasn’t sure if anyone would recognize it haha. I originally started writing on substack but transitioned everything to paragraph earlier this year 😊 are you on farcaster too?
23:25 kinda love how you talk about how your notes are private and you only want to share a couple notes with it. But then your talking points from google are to say it’s better than gpt because google won’t use this content to train their model.
Right! There's a difference between privacy & sharing data for model training. Privacy would mean no one can look at the notes, whereas not including in model training means that the data isn't being parsed into their own data. For most of my Obsidian notes, I'm fine to add them to NotebookLM and, based on the privacy policy, no one should ever see them unless I provide feedback. However, if the notes include confidential information, I would be extra hesitant merely because there is a chance I accidentally click "provide feedback", which may end up introducing legal consequences if someone sees it at Google. I made a video that goes more into depth on these topics, with the main crux being that it depends on the IP & confidentiality requirements of the information you're including into notebookLM: ruclips.net/video/4JU75_v1So4/видео.html Hope it helps!
My understanding is that they are working on bringing this feature in soon! I agree it’s much-needed for long term use of the notebooks Source: www.reddit.com/r/notebooklm/s/2xPWW06ymA
hi, great stuff, thnx. Is it reliable? or is Notebook hallucinating like other AI models and you have to check the results before trusting them and using them to learn before the exams? :)?
My pleasure! Great question, NotebookLM relies on the reliability of the sources you give it. If the source doesn’t contain the answer to your question, it will say something like “the sources don’t provide an answer, but this is a guess”, which helps determine the reliability of the answer. So, in theory, as long as you give it credible & reliable sources, the answer should involve NO hallucinations. That said, these systems are new so I always take everything with a grain of salt 😊 My suggestion would be to test it on a few sources, asking questions in-line with your studying and then assess the answers to see if they’re satisfactory. There’s even an automatic “study guide” creator in the Notebook Guide section, so I think Google anticipated the studying use. Happy to answer any more questions, and good luck with your exams! What are you studying?
@@kotekktorywymysial Very cool! Can see how this would be helpful for sure. I personally have NOT had any hallucinations. NotebookLM has, so far, always told me if the source did not contain information that exactly answered my question. Then I can either find a source to fill that gap, or rephrase my prompt to try and target the existing sources.
@@kotekktorywymysial My pleasure 🫡 hmm NotebookLM is the best I've found so far because you're able to directly add the source and it gives specific citations to where the information was taken from, so you can validate the correctness and asses whether it's sufficient. Most AI's are blackbox, making it hard to know for sure. I haven't explored it fully yet, but Neo4J has a GraphRAG builder that you can add sources to & then query, you could also take a look at that: neo4j.com/labs/genai-ecosystem/llm-graph-builder/
Yes! But where does your data go? 👀 it depends on the plugin. For example, if you’re using one that connects to GPT, you may be giving your obsidian notes to OpenAI. I highly recommend looking at the terms of use of the plugin to see what they do with your data. Also! I personally like the thought of intentionally adding my notes to obsidian, rather than using AI to generate notes within it. Helps me keep separate what is AI and what is my own writing.
If you search “ai” or “chatgpt” there are quite a few! I would just say to be careful to check how they are using your data, because you might lose the privacy of your vault by giving AI direct access to it
Great question! You can use NotebookLM with Google Docs, and I believe there are extensions for chrome/gmail where you can automatically store your Gmail in Docs. You might be able to set up an automatic export from Gmail --> Docs, which you can then select in NotebookLM the next time you use it. The only thing to keep in mind is that you can only have 50 sources in NotebookLM, so to have true, constant updating, you probably need to aggregate your emails into one doc (or a few) so you can work with more sources as well. I haven't tried it, but here's an example I found quickly: chromewebstore.google.com/detail/export-emails-to-docs/mmldpaklpefcmbknkimieipafeniinem?hl=en&pli=1
A couple of reasons! They’re encouraging people to use it for research and business, and if it got out that they were using your data for training, I think Google would take a huge hit. Also, they say right on the main page that they don’t: notebooklm.google/
Great video and very useful. My only comment is that when you’re executing certain things and are showing key concepts, you need to abbreviate more by showing transitions from one step to the next in a single frame at 1/60 of a second, because when you have other conceptual demonstrations, such as a frame that says, “organizing your library or notebooks”, I find it that drags out over three or four frames and it really drags the whole video down. Similarly when you’re provide a demonstration of adding source files or links to the knowledge base, do you wanna make sure that that goes faster so that AIs don’t get bored with it. As a human, I was able to make out that you did five clicks to add a piece of knowledge to the sources in under a 10th of second.. that just drags out too much for an AI to process, although for a few humans we can sort of see what you’re getting at. That’s why we have the pause button..
Joking aside, I’ve subscribed to your channel because I think you’ve got a great way of explaining things. At the end was very spot on and concise. Thank you.
I’m sorry but if you seriously think they are not using your data then why is it free? Anytime Google releases something for free you are the price and your data in particular is what you are paying with.
My guess is that they are: 1) showing off Gemini for promotional purposes 2) providing a valuable software for personal/small business use that they will then sell to enterprises 3) trying to gain some goodwill from ChatGPT, which has been dominating public opinion 4) trying to appeal to universities 5) trying to compete with Apple Intelligence, which has been entirely described as being private And I’m sure more reasons. I think if they were hedging the phrasing around how the data was being used, it would be a different story. But if you go to the main website, it’s unambiguous, they will not use your data via notebookLM. I recommend, as always, taking a look at the page yourself and the terms of use if you are unsure.
Why are people so stuck on not letting LLM train on your data? Nobody gives a crap about your notes and the LLM is only drawing patterns and relationships from the data. If everyone hoarded their output we couldn't have LLMs. You gotta give to receive. It's how we move forward as a species.
Well, quite a few reasons! Depending on the notes you're taking, they may be highly confidential (client work, legal work, business planning, market strategies, etc.). As a lawyer, if the notes I'm taking are related to client work, it would be unethical/potentially illegal for me to use software that takes my confidential information and uses it for AI model training. If I'm doing highly sensitive, cutting-edge research, I could potentially be removing my ability to hold onto the IP. I could also be giving it up to the AI model for training, and when it is updated, someone else could ask questions that causes the AI to give away my research, even though it's not ready for publication yet. This could also remove my ability to retain IP. As a creator, the Internet has shifted to a mentality of expecting everything for free, which has devalued a lot of creative & knowledge work. Assuming that everyone should be able to take everything for free provides a chilling effect on original creativity, since people are not incentivized to create new things if they cannot monetize. That said, the Internet is obviously a treasure trove of collective intelligence of humanity, and LLMs are enabling us to access this knowledge to an entirely new level, which requires some element of openness of information. There needs to be a balance, where, e.g., creators & knowledge workers can opt in to having their work used for training or not, the choice shouldn't be made for each person automatically.
@@Wanderloots Excellent response. I'm not a lawyer and don't deal with any IP so I didn't have that perspective. I do agree there needs to be a balance and feel we sometimes tip too far from fair use and public domain.
![Seungmin "그렇게, 천천히, 우리(As we are)" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/kAzmhLHePqU/mqdefault.jpg)
Wow this is a great walkthrough. Thanks for sharing your experience with NotebookLM!
Wow! Thank you so much 😊It's my pleasure! As an IP lawyer, patent agent & creator, I greatly appreciate that you've made a tool that lets me work with sources without submitting them for training. I've been using NotebookLM every day and can't wait to share more videos ✨Looking forward to seeing what other features you add!
Wow, thank you for the introduction to Notebook LM. I had taken a look at it but stopped short of doing anything with it. Your comprehensive and concise run-through of how to use this tool has given me all the knowledge I need to start using it. And for that, you get a subscription! Thank you 😊🙏.
My pleasure! I'm so glad you found it helpful 😊 Was there anything in particular that resonated & inspired you to use it more? Appreciate the subscription a lot 🫡
This NEEDS more exposure. I have found your channel recently and have given Obsidian a chance again but THIS?! Great insight and thank you for the info
My pleasure! Thanks so much for letting me know how helpful you find it 😊 So glad to hear that you're giving Obsidian a shot! As you can see in this video, I think it's going to be such a great pairing for AI, especially NotebookLM. Let me know if you have any questions!
Feel free to share with anyone you think might find it helpful! That's the best way to get more people to see it 🫡
Obsidian is pretty great, but like any other notetaking-'2nd Brain' tool, not for everyone. But Notebook LM, this is just a whole other 'next level' of where you can take Obsidain.. or any of the better/similar tools. As you can see, even if you just dump all of your digital flotsam in Google Drive.. just as good. My head was spinning and sleepless night the first time I used Notebook LM, dreaming up possible use cases.
Sure! but it does depend. Notion, for example, uses proprietary markup language, which might make it impossible to add notes from Notion into AI like NotebookLM in the future. Future-proofing in software is underrated imo.
For sure, Google Drive is going to be a key element for a lot of people. I think it's cool that it's an add-on to products people are already using.
And I'm so glad to hear it! That sounds like my experience when I came across it too haha, there are so many possibilities ✨ Let me know if you have any questions about use-cases, and maybe I'll be able to work them into future videos 🫡
Best tutorial about Notebook LM so far! Great job, buddy
Thank you very much 😊 there are a lot out there so that means a lot!
Was there something that made this video stand out that you liked?
The podcast feature is quickly becoming one of my favorite features when doing research for personal learning projects. Especially since they added the ability to customize it. Pull all my sources together, throw them in, run the podcast generation, and then that starts giving me ideas for what to talk to the bot about, adding more notes, then I can regenerate the podcast for different angles and depth.
Yes! Exactly ✨ this is one of my favourite features as well. I spent all week exploring the podcast generations actually. Excited to make it a regular part of my creation process 🎙️
I go through dozens of videos like this every week and rarely do they achieve what this video has so clearly and compellingly. Well done, Callum! I've subscribed.
Wow! Thanks Manley, that means a lot 😊 what made this one so clear/compelling for you? I hope to bring those traits to other videos so would love to hear your thoughts!
Thanks for subscribing 🫡
@@Wanderloots If your goal was to add value for your viewers, you hit the mark. Thorough, hype-free, enough detail but not too much detail. I also appreciated that you didn't focus solely on the flash (i.e., the podcast feature) but shared some really helpful and novel use cases for the tool.
@@ManleyWalker Thanks for sharing! Really appreciate the feedback 😊I think there are so many great use cases for AI, but people often get caught up in the flash, so I'm trying to go deeper where I can!
Let me know if there's every anything you're looking for a video about :)
Callum, I appreciate your thorough, thoughtful, possibly overly logical 😆, methodical and analytical.. overview of Notebook LM, you did the single best explanation of this new tool I've found so far, been using it, it's game-changing. I hope you'll do more with this as I think your subscribers will find this fascinating as well as your own ideas on how to use for different purposes. Good one, Bro 🙏👍
Thank you 😊 appreciate you saying that a lot! Was there something that made this video stand out?
I’m so glad to hear you’ve been finding it helpful! How have you been using it in a game-changing way?
Also, I’m working on my ideas for the next part in the series so I am definitely doing more! If there’s something in particular you’d like to see, let me know 🫡
Great walkthrough. For me personally NotebookLM is the perfect AI companion for Heptabase.
Thank you! Glad you found it helpful 😊 was there anything in particular you resonated with?
Also, I hadn’t heard of heptabase! Thanks for sharing. Does it operate in markdown as well?
Solid content you got here man. Subscribed.
Thank you! 😊 appreciate it a lot. What did you find the most interesting?
Thanks!
Thank you so much! Really appreciate your support 🫡 ✨
you encouraged me to test it especially for that podcast feature. Great video!
I’m so glad to hear it! 😊 what did you think of the podcast feature when you used it?
I am testing it right now, so thx for the video. Really helpful 😊
My pleasure! Glad it helped 😊 what are you testing it with? 👀
@Wanderloots especially analysis of longer RUclips videos and brainstorming. Also as another source of AI with access to current internet sources (like perplexity).
Cool! That makes a lot of sense, that’s mostly how I’ve been using it so far as well.
What do you mean by bringing in perplexity? Like querying perplexity and adding the answers as a source to notebooklm?
I'm so impressed by this video. Thank you so much for producing it. Extremely good presentation. Worth every second. 😍
Thank you! 😊 it’s my pleasure, I’m glad you found it helpful!
Was there anything that jumped out at you as being the most interesting?
Thank you! I'll be also checking out your Obsidian playlist! haha
My pleasure 🫡 and awesome! Let me know if you have any questions 😊
Thank you for the learning, very helpful.
My pleasure! Was anything particularly helpful? Working on some more videos 😊
Thank you for this beautiful explanation
My pleasure! So glad you found it helpful 😊
thank you so much for this content man
My pleasure! Glad you found it helpful 😊 have you been using NotebookLM?
@@Wanderloots Well I've been using it since two months and I found it great, especially while I'm studying IT and I need to search between a larg amount of literature and textbooks, so it become really helpful to use something like this great tool.
Also, in terms of mathmatical notation, I used to write every thing on paper coz it's hard to write them using keyboard. Now I'm able to copy my notes directly into obsidian (which by the way I'm watching your vidoes about it and you did a great job, but it's still confusing me 😅)
For NotebookLM, I think they need to add more features, such as organizing notes and enhance the creativity of the model and I hope they will do that.
@@aqeelsapeeh Glad to hear how helpful it's been! I use it all the time too, makes research so much easier. Completely agree that they could update in a few helpful ways. My guess is that they were testing its popularity, and with the massive success they've had, they're building it more into their core features (especially since they can monetize it now with NotebookLM plus, which will, in theory, give more funding to the team building it).
Re: obsidian, happy to answer any questions on things you find confusing! I know it can be a lot to get started, but you can also start as simply as possible and build out as you need to, rather than trying to have the "perfect system" up front.
I also have some explanations on wanderloots.xyz if you're looking for written tutorials to complement the videos :)
@@Wanderloots I appreciate your help 🤍
Great video so informative thank you. I believe I missed the privacy protections in NotebookLM how secure and private is it actually?
It's my pleasure! Thanks for letting me know 😊 I recommend (always) reading the privacy policy yourself before determining if it is "secure enough" for your use. I actually included a chapter in my last NotebookLM video on the privacy policy and my understanding of how to keep your use secure, so you can take a look at that for more information 🫡: ruclips.net/video/EsQ6sIAacEY/видео.html
(note: none of this is legal advice)
its about 5 secure
Wow. This is awesome and exciting, but also scary. We're not gonna be able to tell the difference between what's real and what's not anymore.
Absolutely, my feelings exactly. tbh, I think we're already at that point of being unable to tell truth from fiction. AI-leveraged deep fakes have been able to almost (or truly) mimic many forms of information we consume. This is why, imo, we need better credibility systems for the internet, to better protect digital identities in a way that allows us to actually verify the source of information.
I wrote an article on the topic of needing a safer Internet here if you're interested: wanderloots.com/recalibrating-35/
and also talk a bit about this in my digital identity video: ruclips.net/video/ESg-ZZ_odJM/видео.html
Would love to hear your thoughts! 🫡
Great video!!!!
Thank you! 😊 was there anything you found particularly helpful?
@@Wanderloots yes, the full coverage of the functionality of NobookLLM, the examples and the potential use cases. Honestly, it is worth rewatching (a few times) to brainstorm on how I (or anyone) could really apply it for ideas I have. the potential is enormous. One thing that would be great for another video is to expand on how it can be used as specialized tools. That is something I'm very interested in especially in the context of research -> content genaration/production. Thank you for the video and for your quick response.
@@briocheatete Awesome! Thanks so much for sharing your feedback 😊It really helps me know what people are most interested in.
I actually just released a video using NotebookLM as a specialized tool for content generation: ruclips.net/video/4QJm_AptHF4/видео.html
Would love to hear your thoughts, if this is what you were thinking of re: research --> content production. Hope it helps!
Fantastic - God Mode!
AI is a super powerful tool! We’ll see where it goes ✨
Great insight thanks
My pleasure! 😊 was there anything in particular you found the most insightful? Working on some more videos so I’m curious what resonates ✨
Where businesses grow, AI leads the way 💫
Thanks so much. So helpful to know this
My pleasure! So glad you found it helpful 😊 was there anything in particular that jumped out at you? I'm always curious what people find the most interesting so I can focus on it in future videos 🫡
@@Wanderloots I am a Christian and I listen to 30-40 minute sermons on youtube. This tool will give a nice descriptive overview of the sermon points as well as the main scriptures that the preacher is talking about. I can refer this info even without having to sit through listening to most of the youtube video. So very helpful.
@@romeshjayawardene3551 Glad to hear it! It really is a superpower for consuming long form content efficiently while extracting insights :)
Hit subscribe as soon as I saw that Paragraph logo 💪🏾💯
Amazing! I wasn’t sure if anyone would recognize it haha. I originally started writing on substack but transitioned everything to paragraph earlier this year 😊 are you on farcaster too?
Fantastic
Thank you 😊 glad you found it helpful! Was there anything in particular that resonated?
Thank you!
You’re welcome! 😊 was there something you found particularly helpful or interesting?
Very, VERY. Good! Thank you. ( new sub now!}
Thank you so much! Glad you liked it 🫡 was there anything in particular you found helpful, or would like me to go deeper into??
23:25 kinda love how you talk about how your notes are private and you only want to share a couple notes with it. But then your talking points from google are to say it’s better than gpt because google won’t use this content to train their model.
Right! There's a difference between privacy & sharing data for model training. Privacy would mean no one can look at the notes, whereas not including in model training means that the data isn't being parsed into their own data. For most of my Obsidian notes, I'm fine to add them to NotebookLM and, based on the privacy policy, no one should ever see them unless I provide feedback. However, if the notes include confidential information, I would be extra hesitant merely because there is a chance I accidentally click "provide feedback", which may end up introducing legal consequences if someone sees it at Google. I made a video that goes more into depth on these topics, with the main crux being that it depends on the IP & confidentiality requirements of the information you're including into notebookLM: ruclips.net/video/4JU75_v1So4/видео.html Hope it helps!
Is there any way of retaining the clickable citation links once a note is saved then reopened? They seem to go.
My understanding is that they are working on bringing this feature in soon! I agree it’s much-needed for long term use of the notebooks
Source: www.reddit.com/r/notebooklm/s/2xPWW06ymA
hi, great stuff, thnx. Is it reliable? or is Notebook hallucinating like other AI models and you have to check the results before trusting them and using them to learn before the exams? :)?
My pleasure! Great question, NotebookLM relies on the reliability of the sources you give it. If the source doesn’t contain the answer to your question, it will say something like “the sources don’t provide an answer, but this is a guess”, which helps determine the reliability of the answer.
So, in theory, as long as you give it credible & reliable sources, the answer should involve NO hallucinations. That said, these systems are new so I always take everything with a grain of salt 😊
My suggestion would be to test it on a few sources, asking questions in-line with your studying and then assess the answers to see if they’re satisfactory.
There’s even an automatic “study guide” creator in the Notebook Guide section, so I think Google anticipated the studying use.
Happy to answer any more questions, and good luck with your exams! What are you studying?
@@Wanderloots thnx/.. international rel .. let me rephrase my q .. have you encountered any hallucinations in Notebook?
@@kotekktorywymysial Very cool! Can see how this would be helpful for sure. I personally have NOT had any hallucinations. NotebookLM has, so far, always told me if the source did not contain information that exactly answered my question. Then I can either find a source to fill that gap, or rephrase my prompt to try and target the existing sources.
@@Wanderloots thank you :) do you know if there is any other LLM that is not hallucinating and can summarize a textbook? :)
@@kotekktorywymysial My pleasure 🫡 hmm NotebookLM is the best I've found so far because you're able to directly add the source and it gives specific citations to where the information was taken from, so you can validate the correctness and asses whether it's sufficient. Most AI's are blackbox, making it hard to know for sure. I haven't explored it fully yet, but Neo4J has a GraphRAG builder that you can add sources to & then query, you could also take a look at that: neo4j.com/labs/genai-ecosystem/llm-graph-builder/
Pero en obsidian también se puede tener algo igual a Notebook lm.utilizando un plugin
Which plugin, please?
Yes! But where does your data go? 👀 it depends on the plugin. For example, if you’re using one that connects to GPT, you may be giving your obsidian notes to OpenAI. I highly recommend looking at the terms of use of the plugin to see what they do with your data.
Also! I personally like the thought of intentionally adding my notes to obsidian, rather than using AI to generate notes within it. Helps me keep separate what is AI and what is my own writing.
If you search “ai” or “chatgpt” there are quite a few! I would just say to be careful to check how they are using your data, because you might lose the privacy of your vault by giving AI direct access to it
Absolutely Sigma ❤
Can you link notebook lm to gmail so it constantly updates your information?
Great question! You can use NotebookLM with Google Docs, and I believe there are extensions for chrome/gmail where you can automatically store your Gmail in Docs. You might be able to set up an automatic export from Gmail --> Docs, which you can then select in NotebookLM the next time you use it.
The only thing to keep in mind is that you can only have 50 sources in NotebookLM, so to have true, constant updating, you probably need to aggregate your emails into one doc (or a few) so you can work with more sources as well.
I haven't tried it, but here's an example I found quickly: chromewebstore.google.com/detail/export-emails-to-docs/mmldpaklpefcmbknkimieipafeniinem?hl=en&pli=1
cool. but why do you think they won't tap these treasures?
A couple of reasons! They’re encouraging people to use it for research and business, and if it got out that they were using your data for training, I think Google would take a huge hit.
Also, they say right on the main page that they don’t: notebooklm.google/
Great video and very useful. My only comment is that when you’re executing certain things and are showing key concepts, you need to abbreviate more by showing transitions from one step to the next in a single frame at 1/60 of a second, because when you have other conceptual demonstrations, such as a frame that says, “organizing your library or notebooks”, I find it that drags out over three or four frames and it really drags the whole video down. Similarly when you’re provide a demonstration of adding source files or links to the knowledge base, do you wanna make sure that that goes faster so that AIs don’t get bored with it. As a human, I was able to make out that you did five clicks to add a piece of knowledge to the sources in under a 10th of second.. that just drags out too much for an AI to process, although for a few humans we can sort of see what you’re getting at. That’s why we have the pause button..
Joking aside, I’ve subscribed to your channel because I think you’ve got a great way of explaining things. At the end was very spot on and concise. Thank you.
thank you! I appreciate you saying that 😊 to be honest, your comment above confused me haha, it was a joke?
I’m sorry but if you seriously think they are not using your data then why is it free? Anytime Google releases something for free you are the price and your data in particular is what you are paying with.
My guess is that they are:
1) showing off Gemini for promotional purposes
2) providing a valuable software for personal/small business use that they will then sell to enterprises
3) trying to gain some goodwill from ChatGPT, which has been dominating public opinion
4) trying to appeal to universities
5) trying to compete with Apple Intelligence, which has been entirely described as being private
And I’m sure more reasons. I think if they were hedging the phrasing around how the data was being used, it would be a different story. But if you go to the main website, it’s unambiguous, they will not use your data via notebookLM. I recommend, as always, taking a look at the page yourself and the terms of use if you are unsure.
Why are people so stuck on not letting LLM train on your data? Nobody gives a crap about your notes and the LLM is only drawing patterns and relationships from the data. If everyone hoarded their output we couldn't have LLMs. You gotta give to receive. It's how we move forward as a species.
@chastetree stop projecting ur nonsense way of thinking to others. YOU don’t care. keep that dumb naive ass shit to yourself
Well, quite a few reasons! Depending on the notes you're taking, they may be highly confidential (client work, legal work, business planning, market strategies, etc.). As a lawyer, if the notes I'm taking are related to client work, it would be unethical/potentially illegal for me to use software that takes my confidential information and uses it for AI model training.
If I'm doing highly sensitive, cutting-edge research, I could potentially be removing my ability to hold onto the IP. I could also be giving it up to the AI model for training, and when it is updated, someone else could ask questions that causes the AI to give away my research, even though it's not ready for publication yet. This could also remove my ability to retain IP.
As a creator, the Internet has shifted to a mentality of expecting everything for free, which has devalued a lot of creative & knowledge work. Assuming that everyone should be able to take everything for free provides a chilling effect on original creativity, since people are not incentivized to create new things if they cannot monetize.
That said, the Internet is obviously a treasure trove of collective intelligence of humanity, and LLMs are enabling us to access this knowledge to an entirely new level, which requires some element of openness of information. There needs to be a balance, where, e.g., creators & knowledge workers can opt in to having their work used for training or not, the choice shouldn't be made for each person automatically.
@@Wanderloots Excellent response. I'm not a lawyer and don't deal with any IP so I didn't have that perspective. I do agree there needs to be a balance and feel we sometimes tip too far from fair use and public domain.
get to the point please