Reflective journal entries using GPT-4 and Obsidian that demand less willpower.
post by Solenoid_Entity · 2023-04-15T12:45:25.982Z · LW · GW · 24 commentsContents
The solution: My continuation: Response: My continuation: Response: And voilà: None 24 comments
Something I've wanted to do for years, but can't make myself do consistently because I'm bad at forming new habits and seem to have a limited reservoir of willpower that is generally depleted by the end of the day:
- Write a short journal entry at the end of every day, summarizing the day and reflecting on it.
- Put this in Obsidian, and identify key concepts and make them back-links.
- Enjoy as, over several years of consistent writing, the map of my life unfolds through the graph in Obsidian. Maybe this generates helpful insights, shows me patterns. Mostly it's just useful as a resource.
It's a lot of effort to be detailed and reflective, and it's a lot of effort to add the back-links.
The solution:
Rather than having to create a detailed, reflective entry from whole cloth, I can just write some short, abbreviated, incoherent mumbled paragraph, like a surly teenager at family dinner, and then ask ChatGPT to ask more questions to flesh out the story.
This works shockingly well.
It's surprisingly easy to write detailed answers to specific QUESTIONS about your day. It's like a good listener, it can draw you out with its insightful questions.
Then you can ask it to mark all the key words, names, etc, with double square brackets.
The end result isn't perfect but it's pretty good, and much, much better than I would ever do all by myself. It takes less than half the time and virtually no willpower.
Here's what I prompted ChatGPT today (That diary entry is roughly the level of writing I'm capable of at the end of the day.)[1]
Ask me 10 questions about the following diary entry, aiming to 'flesh it out' with extra details. If there are inconsistencies or missing information/context, you can ask for clarification or further information. Your aim is to ask questions that help make this a more complete and coherent account of the day. Some of the questions should relate to the theme of 'gratitude', and help provide more context and identify things to be grateful about in the text. Here's the text:[2]
Woke up early for no reason, lay in bed for 20 minutes thinking, then got up, said goodbye to Xena, headed off at around 7am for the train up to Waitara. Ended up missing the stupid thing because I was waiting on the wrong platform like an idiot. Still got to coffee in time with Adele, which was a relief since I've been an unreliable friend recently. Good coffee with her, turns out the guy who owns the cafe knows her as a regular and she doesn't even have to order, he just brings her french toast as soon as he sees her sit down. Anyway we talked about her new startup, then I left and caught up with my friend who's building a paperclip-optimising machine, which sounds stressful for him tbh, but his new girlfriend is nice, they seem like a good match. Got the late train over to Emu Plains and finally got to see my parents. Hopefully cook something nice with Mum tomorrow before she goes to line-dancing club.
Response:[3]
1. Can you provide more details about Xena? What is your relationship with her and how long have you known her?
2. How did you feel when you missed the train, and were there any positive aspects to this experience that you are grateful for?
3. How do you and Adele usually stay in touch, and how long have you been friends?
4. What specifically makes you feel that you've been an unreliable friend recently? Are there any instances you can recall?
5. Can you tell me more about the cafe where you met Adele? What is it about the atmosphere or service that you appreciate?
6. What are some aspects of your conversation with Adele that you are grateful for or that made you feel more connected to her?
7. Can you provide more information about your friend's paperclip-optimizing machine? What is its purpose, and how do you feel about his project?
8. How did you first meet your friend's new girlfriend, and what about her makes you think that they are a good match?
9. When was the last time you visited your parents in Emu Plains, and what are some things you are looking forward to doing with them during this visit?
10. What are some of your favorite dishes to cook with your Mum, and how do you feel when you share this activity together?
My continuation:
1. She's my friend from a long time ago, but she's been super busy with her job as a warrior princess and doesn't have much time for friends. I was there the night before to have dinner and drinks because she finally got a week off work, ended up having more drinks than planned. Her couch was not very comfortable which is probably why I woke up early.
2. Really annoyed with myself, it was such an avoidable mistake. Honestly, it was pretty positive because I was only 10 minutes late, and Adele was running late anyway. I was able to find a good alternative train route using the app, so it worked out ok.
3. We try to have a phone call every couple of weeks, but it's hard because her life is so unpredictable what with the record deals and the touring and the paparazzi and all that. Still, I was grateful for her making the time.
4. Just several times we've arranged to meet up and then I've had to cancel because of chaos in my life getting in the way. It's one of those things that just happens twice in a row and suddenly feels like the new norm.
5. I forget what it was called. Honestly the food was just great, they accommodated my weird order with no fuss, it wasn't to loud or busy, good energy and vibe, attentive service but not annoying. No complaints, 5 stars.
6. It was nice hearing all her stories from the road, crazy things that happened backstage at her shows, all those sorts of details. I like that she feels she can share that sort of stuff with me, I guess she really trusts me not to sell the info to the magazines, and that feels special!
7. I don't want to give you any more information about that. I don't like his project.
8. Oh, he brought her along to coffee so I could meet her. Her name's Ivana. She's super smart like him, plus she's beautiful (must be the russian in her) she's previously worked in office supplies, mainly with an x-risk focus, but now she works in some weird role, I didn't really understand it but the company she works for was called the FSD or the FBS or something. She seems to really like him and she's a great listener, I felt like I could tell her anything.
9. I try to see my parents every couple of months. They always feed me well.
10. She makes great pies. Like seriously great, I don't know if I'll ever be able to make them like that. It's nice, it's a good bonding exercise, I used to cook with her as a kid before she sent me off to that remote violin academy.
Thanks for the questions. Now, rewrite the original diary entry, in regular, relaxed plain conversational English, but implement all the new information from my answers to make it longer and more detailed.
Response:
Woke up early for no reason at my friend Xena's place after having dinner and drinks with her the night before. She's a warrior princess who's super busy with her job, so I was really grateful for her making time to catch up. Her couch wasn't very comfortable, which is probably why I woke up early. I spent 20 minutes just lying in bed thinking before I got up, said goodbye to Xena, and headed off around 7am for the train up to Waitara.
I ended up missing the stupid train because I was waiting on the wrong platform like an idiot. I was really annoyed with myself, but luckily, it turned out to be a positive experience. I was only 10 minutes late and found a good alternative train route using the app. Adele, who I was meeting for coffee, was running late anyway.
Adele and I try to stay in touch with phone calls every couple of weeks, but it's hard because of her unpredictable life with record deals, touring, and dealing with paparazzi. I've been an unreliable friend recently, having to cancel a few meetups because of chaos in my life. It was great to finally catch up with her at the cafe, which had fantastic food, a great vibe, and attentive service. The guy who owns the cafe knows her as a regular and just brings her french toast as soon as she sits down.
We had a nice chat about her new startup and all her stories from the road, crazy things that happened backstage at her shows, and other details. I really appreciated her trust in sharing those stories with me. Afterward, I left to catch up with my friend who's building a paperclip-optimizing machine. He introduced me to his new girlfriend, Ivana, who's super smart, beautiful, and works for a company called FSD or FBS. She's a great listener, and I felt like I could tell her anything.
Finally, I took the late train over to Emu Plains to see my parents, who I try to visit every couple of months. They always feed me well. I'm looking forward to cooking something nice with Mum tomorrow before she goes to her line-dancing club. She makes seriously amazing pies, and cooking with her is a great bonding exercise. It reminds me of when I used to cook with her as a kid before she sent me off to that remote violin academy.
My continuation:
Surround (with double square brackets) every proper noun, family member, mode of transport, company or workplace name or type, hobby, or any other words or phrases that are very important to this story.
Response:
Woke up early for no reason at my friend [[Xena]]'s place after having dinner and drinks with her the night before. She's a [[warrior princess]] who's super busy with her job, so I was really grateful for her making time to catch up. Her couch wasn't very comfortable, which is probably why I woke up early. I spent 20 minutes just lying in bed thinking before I got up, said goodbye to [[Xena]], and headed off around 7am for the [[train]] up to [[Waitara]].
I ended up missing the stupid [[train]] because I was waiting on the wrong platform like an idiot. I was really annoyed with myself, but luckily, it turned out to be a positive experience. I was only 10 minutes late and found a good alternative [[train]] route using the [[app]]. [[Adele]], who I was meeting for coffee, was running late anyway.
[[Adele]] and I try to stay in touch with phone calls every couple of weeks, but it's hard because of her unpredictable life with [[record deals]], touring, and dealing with [[paparazzi]]. I've been an unreliable friend recently, having to cancel a few meetups because of chaos in my life. It was great to finally catch up with her at the [[cafe]], which had fantastic food, a great vibe, and attentive service. The guy who owns the [[cafe]] knows her as a regular and just brings her french toast as soon as she sits down.
We had a nice chat about her new [[startup]] and all her stories from the road, crazy things that happened backstage at her shows, and other details. I really appreciated her trust in sharing those stories with me. Afterward, I left to catch up with my friend who's building a [[paperclip-optimizing machine]]. He introduced me to his new girlfriend, [[Ivana]], who's super smart, beautiful, and works for a company called [[FSD]] or [[FBS]]. She's a great listener, and I felt like I could tell her anything.
Finally, I took the late [[train]] over to [[Emu Plains]] to see my [[parents]], who I try to visit every couple of months. They always feed me well. I'm looking forward to cooking something nice with [[Mum]] tomorrow before she goes to her [[line-dancing club]]. She makes seriously amazing pies, and cooking with her is a great bonding exercise. It reminds me of when I used to cook with her as a kid before she sent me off to that remote [[violin academy]].
And voilà:
Next improvement will probably be to use voice recognition so I don't have to type my answers out, and can be even more surly.
- ^
Literally every detail and fact in the diary entry is fictional and invented, just made up for this post. Don't be thinking I take trains or eat food.
- ^
This is in my text expander now, so it's easy to reproduce.
- ^
What really impresses me about these questions is that it doesn't delve into things like 'what's Adele's startup' or 'why is your friend's job at the paperclip maximising firm stressful?'
I don't know if that's a feature or a bug. It doesn't perform great at the 'resolve inconsistencies and add context', but it seems to understand the assignment: that we're here to talk about my feelings.
24 comments
Comments sorted by top scores.
comment by lillybaeum · 2023-04-16T00:53:18.634Z · LW(p) · GW(p)
I really strongly recommend that anyone interested particularly in journaling using AI does not use anything that sends that data to a server somewhere. Use llama or another local model if you possibly can. It may be less capable but it's far more responsible to do so. Personally I find a lot of value in AI journaling, but it also made me glimpse the future possibilities of what unscrupulous companies could do with a detailed log of every day of your life for years. I mean, isn't that basically the most useful user data you could possibly have for advertising?
I'm not saying that OpenAI specifically is currently untrustworthy with this data, or that it one day will be, but it just strikes me as a really bad habit to send every aspect of your day in detail to somewhere on the internet.
On a more 'doomer' sort of note, if anyone is afraid of AI being capable of simulating their personality in a roko sort of situation, this seems like the best possible way to give some future AI as near-lossless image of your brain at present. I for one welcome Roko/Bing/Proteus and support it completely if it decides to take over the world, but remember, the most powerful aspect of LLMs is their capability to predict things given input. I can't help but feel some sort of truth behind the idea that an LLM could predict how you would live and think and act, given enough input about your life and your choices.
Replies from: Solenoid_Entity, LosPolloFowler, Simon Möller↑ comment by Solenoid_Entity · 2023-04-16T11:36:48.500Z · LW(p) · GW(p)
Yeah, I agree with a lot of this, and this privacy concern was actually my main reason to want to switch to Obsidian in the first place, ironically.
I remember in the book In the Age of Surveillance Capitalism there's a framework for thinking about privacy where users knowingly trade away their privacy in exchange for a service which becomes more useful for them as a direct consequence of the privacy tradeoff. So for example, a maps app that remembers where you parked your car. This is contrasted with platforms where the privacy violations aren't 'paid back' to the users in terms of useful features that benefit them, they just extract value from users in exchange for providing a service at all.
So in this case, I guess the more private information I submit to Chat-GPT, the more directly useful and relevant and insightful its responses to me get. Considering how much a life coach or career coach or therapist can cost, this is a lot of value I'm getting for it.
I understand the theoretical concern about our righteous future overlords whom I fully support and embrace, but while I think you could learn a lot about me from reading my diary, including convincingly simulating my personality, I would feel surprised if reading my diary was enough to model my brain in sufficient fidelity that it's an s-risk concern...
Replies from: christopher-king, Solenoid_Entity↑ comment by Christopher King (christopher-king) · 2023-04-16T16:05:42.838Z · LW(p) · GW(p)
So in this case, I guess the more private information I submit to Chat-GPT, the more directly useful and relevant and insightful its responses to me get.
This is even stronger for something like LLaMA because you can actually fine-tune it on your personal info or fine-tune it for document retrieval.
↑ comment by Solenoid_Entity · 2023-04-16T11:42:38.634Z · LW(p) · GW(p)
I also realise how much I sound like Chat-GPT in that comment... dammit
Replies from: awg↑ comment by Stephen Fowler (LosPolloFowler) · 2023-04-16T11:19:15.470Z · LW(p) · GW(p)
I like the way you think.
While an indepth daily journal would help simulating a person, I suspect you could achieve a reasonably high fidelity simulation without it.
I personally don't keep a regular journal, but I do send plenty of data over messenger, whatsapp etc describing my actions and thoughts.
↑ comment by Simon Möller · 2023-04-16T10:08:05.144Z · LW(p) · GW(p)
I fully agree. I tried using ChatGPT for some coaching, but tried to keep it high level and in areas where I wouldn't be too bothers if it showed up on the internet.
I think using the API, rather than ChatGPT, is better. See e.g. https://techcrunch.com/2023/03/01/addressing-criticism-openai-will-no-longer-use-customer-data-to-train-its-models-by-default/:
Starting today, OpenAI says that it won’t use any data submitted through its API for “service improvements,” including AI model training, unless a customer or organization opts in. In addition, the company is implementing a 30-day data retention policy for API users with options for stricter retention “depending on user needs,” and simplifying its terms and data ownership to make it clear that users own the input and output of the models.
I was actually thinking that having an Obsidian plugin for this sort of thing would be really neat.
Replies from: Solenoid_Entity↑ comment by Solenoid_Entity · 2023-04-16T11:44:23.729Z · LW(p) · GW(p)
There are a few Obsidian plugins that do similar stuff using LLMs, (they purport to read your notes and help you something something).
I'm thinking of mocking something up over the next week or so that does this 'diary questions' thing in a more interactive way, via the API, from inside Obsidian.
comment by Celarix · 2023-04-15T18:54:39.959Z · LW(p) · GW(p)
I might try this with GPT 3.5, as I can't yet justify spending the $20/month on GPT-4. Something like this seems quite interesting!
(oh god we're going to have to be using this near-AGI stuff for everything soon, aren't we?)
Replies from: Celarixcomment by awg · 2023-04-15T15:50:18.360Z · LW(p) · GW(p)
Love this! I use Obsidian for daily journaling, but I love your use of ChatGPT as a sort of diary-prompter for yourself.
I think probably within the next couple of years it will be possible to use an LLM to write an Obsidian plug-in that lets you use an LLM directly inside of Obsidian, connected to all of your Obsidian notes. Combining that with voice recognition like you said so that you can just have a little conversation at the end of your day. Mm, that'd be the magic!
Replies from: Simon Möller↑ comment by Simon Möller · 2023-04-16T10:03:35.566Z · LW(p) · GW(p)
Couple of years? I think we are talking about months here. I guess the biggest bottleneck would be to get all notes into the LLM context. But I doubt you really need that. I think you can probably guess a few important notes for what you are currently working on and add those as context.
comment by Ben (ben-lang) · 2023-04-17T15:03:53.214Z · LW(p) · GW(p)
This is a really nice use of the technology to make something work better for you.
And for some reason the place my mind went is some kind of Black Mirror episode where an emotionally vulnerable person has a chatbot confident/interactive journal, and then their manipulative partner has installed some dark web app on it that makes it spy on her and help him manipulate her feelings ... urggh gross.
comment by M. Y. Zuo · 2023-04-16T14:59:55.817Z · LW(p) · GW(p)
I'm a bit confused here, isn't reflecting on, integrating, and writing down these thoughts the primary benefit of keeping a daily diary in the first place?
Asking GPT-4 to do these steps doesn't seem to offer the same benefits.
Replies from: Solenoid_Entity↑ comment by Solenoid_Entity · 2023-04-16T15:58:37.604Z · LW(p) · GW(p)
I’m writing the original paragraph, and answering a bunch of questions designed to prompt me to reflect.
Replies from: M. Y. Zuo↑ comment by M. Y. Zuo · 2023-04-16T17:23:54.082Z · LW(p) · GW(p)
They might not be your reflections though?
GPT-4 has a very large degree of freedom in choosing what questions to ask, and how they're worded.
So phrasing it one way or another will prompt different thoughts from you. And GPT-4, not you, is selecting those parameters.
You would need to specify a very narrow range of possibilities, or the exact question for it to ask to avoid this, which would seem to be about as much effort as just doing it entirely yourself.
EDIT: Changed wording a bit.
Replies from: awg↑ comment by awg · 2023-04-16T17:29:08.275Z · LW(p) · GW(p)
How is this different than reflecting on your day with a close friend who asks you questions about it?
Replies from: M. Y. Zuo↑ comment by M. Y. Zuo · 2023-04-16T17:37:54.850Z · LW(p) · GW(p)
Not much of a difference, in your example the close friend would also have influenced the answers with the questions they select.
Some people are fine with that because they trust the close friend.
Others aren't and diary writing would be an exclusively solitary activity for the reasons above.
Replies from: awg↑ comment by awg · 2023-04-16T17:44:53.583Z · LW(p) · GW(p)
Totally agreed. I do morning pages which I find to be invaluable as a source of deep, personal reflection.
I think the above has its use in the sense of keeping a diary entry of each day that you could come back to to read and remember what happened. Not necessarily for deep personal reflection, but more to suss out the details on an otherwise sparser entry for a "What happened today?" type of diary.
comment by Olomana · 2023-04-16T05:56:19.427Z · LW(p) · GW(p)
Are you doing this from within Obsidian with one of the AI plugins? Or are you doing this with the ChatGPT browser interface and copy/pasting the final product over to Obsidian?
Replies from: Solenoid_Entity↑ comment by Solenoid_Entity · 2023-04-16T11:28:59.211Z · LW(p) · GW(p)
Currently just copy-pasting into GPT-4 via the web interface. I've got it working via the GPT-3 API as well today, but for now I prefer to suffer the inconvenience and get the better model. The questions it asks are MUCH more insightful.
comment by RHollerith (rhollerith_dot_com) · 2023-04-15T18:24:58.002Z · LW(p) · GW(p)
If you got used to this, you would get rusty at creating journal entries when not connected to the internet, correct? (because to use ChatGPT requires internet connectivity)
Replies from: gilch↑ comment by gilch · 2023-04-15T19:53:03.323Z · LW(p) · GW(p)
A powerful enough personal computer could run LLaMA locally. I don't think the raw model is optimized for chat, but with a suitable prompt, you might be able to get it into chat mode long enough to do this kind of thing. It also wouldn't surprise me to learn that there are more specialized derivatives now that would be suitable. I've certainly heard of people working on it.
Replies from: rhollerith_dot_com↑ comment by RHollerith (rhollerith_dot_com) · 2023-04-16T14:11:34.065Z · LW(p) · GW(p)
Thanks; have six points of upvotedness.