To be fair you did ask it to get you from v4 to v6 specifically so I understand why it wants you to project more. If you asked it to just make you a better climber in general it might have included more volume climbing.
Interesting take. At first glance I am not sure I would treat the two differently i.e. climbing harder and being a better climber. But my bias here might be that my approach to coaching a V4 climber would be to make them a better climber, so that they can climb harder. The opposite might be working to achieve a single V6 with very little foundation of V5. Personally I think projecting-only for 6 months is the wrong approach for most V4 climbers. But I still wonder if ChatGPT made this kind of distinctions, and if we asked it to write a plan to "be a better climber AND climb V6" it would have written something different.
@@LatticeTraining Yes, exactly :) I think it's very much a machine AI still. It will take what you give it quite exact. At least that is what I suspect. Would be interesting to see if there was a different response with a different input.
There’s a lot of potential for this technology in the fitness and sports industry. Especially as the software gains more memory and customizability. A human will be able to teach it what it should be doing. It will make coaches more effective and probably more profitable Like an assistant you don’t have to pay.
Yes I think in the future we'll be playing a combined role of sorts. Ai doing a lot of the roles that can be systematized, allowing us as coaches to spend more time on the soft skills and fine tuning, such as communication, goal setting, chatting through tactics and detailed personalisation of a training plan.
@@LatticeTraining especially if it can remember past prompts long term and respond to feedback. You can say “write this week’s workouts for will, but make sure the weight range increase you provide is smaller than the last time you progressed it” At that point it can work like an automated spreadsheet.
Important to note that you're using GPT 3.5, not 4 which is far better :p. I think that's an important note that you missed to include. There is something to be said about prompt engineering and getting good output out of it. It's not a truth machine.
Yes great point. We are using GPT4. And the prompt engineering issue is why we think a good prior understating of training is needed, currently, if we are going to get anything good out of ChatGPT.
@@LatticeTraining Think you guys were confused around paying for gpt4 and using gpt4 as a model. You can see that the green icon is being used (gpt3.5 model) so you guys didn't use gpt4. You can also tell it isn't gpt4 as the response speed is way to fast.
I think ChatGPT can't really replace actual coaching, particularly not the kind of stuff Lattice has been putting out or Hooper's Beta. With that being said, I've actually been following a training plan ChatGPT put out for me, which is similar to the one you've outlined here but a bit faster (v4 to v6 in 2 months, 6x week). I'm already at the v5 stage (i.e. I'm projecting - I'm not really a v5 climber). Whether or not I'll actually reach v6 remains to be seen, but I'm pleased with the results so far. Although I have to say it's been rough doing 6x week, I'm having to be pretty strict with nutrition.
Wow 6x a week! Are some of these quite light training days e.g. active recovery? It sounds like a very heavy training plan. Also how many days a week were you training before you started the plan?
6x a week sounds really scary unless like the Lattice account mentioned above that's including multiple active rest days. If you're actually climbing and training 6 days a week I'd be seriously worried that you would be more likely to get injured rather than hitting your V6 goal.
@Lattice Training there are two dedicated "rest days" with some light training that involves antagonists and Fingers. I was doing 4x week before starting this new training protocol consisting of 2x climbing 2x weight training. Much prefer this new protocol though considering that it gives me more climbing days (which is what I really enjoy!)
In my limited experience, intense training that isn’t specifically climbing related is only really useful in the V9/7C or harder range. Up til that point, for almost every beginner to intermediate climber is best using their time actually climbing 90% of the time and strength training on campus/hang boards the other 10%. Standardized (moon/kilter/tension) board climbing is also a severely undervalued training tool by intermediate climbers that can rapidly build strength and technique.
*the thing about ChatGPT is that it is a standalone device. As of September 2021, the "Knowledge Cutoff" date it cannot access anything beyond its blackbox. Therefore, any training methods would have been peer reviewed and/or debunked since then. Of course it wont have all the answers.*
Yes, its not going to be current but I think the mistakes made in this plan (based on our prompts) are not based on peer-review research. These issues would have been bad in 2020.
Please don't call it open source - OpenAI is only open in name. There are real open-source efforts in this space, but OpenAI is definitively not open in any sense. Much like any country with "democratic" in the name is probably not democratic.
All you have to do is climb hard to get to V6. No hangboarding, no extra training, just climb smart safe and hard. Be on the wall as much as you can, because the less resting you do, the more you will develop the ability to climb more routes per hour. V7 and beyond probably requires more, but I got my first V8 clear by only using hangboard by feel pushing to the hardest edges I can hold for 5-10 seconds. Best to not overcomplicate the simple. Thinking is a waste of energy until you're actually in the real grades.
I don’t think this is true for everyone, especially depending on weight and body type. Some people also might want the most efficient, high yield way to accomplish this goal as well, which may not always be “just climb hard”
All you have to do is go to the gym with the easiest grading you can get to, climb V6 there, and viola. Never try at a harder gym and *definitely* don't go outside. There we go, you are now a V6 climber.
Interesting ideas but honestly what is a coach going to tell you? Don't get injured? Wait for your tendons to condition? Use your feet mindfully? I think UA-cam itself has made coaching redundant until you're an advanced climber. It's not like you can pay a coach to make your finger tendons recover faster.
Now if lattice were to share their information to the open source community maybe we could have better information for climber's... Instead lattice hoard's all the information they obtain for themselves. At least the power company release's their data. AI might be a good resource if you would realse your data.
I think the community will catch up on this eventually. And GDPR means that dissatisfied athletes can always call lattice up and ask to have their data removed.
We've routinely shared our data and research helping our community understand training, observations and furthering sports science. We don't share our database of individual data points, that we have collected from years of assessments because this acts as one of the foundations of our business. The assessment service we offer supports our company and the people we employ. We want, and do, share as much information as possible including all the free content on this UA-cam channel, our blogs, research papers and Instagram, but we also need to protect our company and its employees to keep doing what we do, which I am sure you understand.
@@LatticeTraining You could always charge a fee for access to that database of individual data points. I imagine that could be a nice additional revenue stream for Lattice. Might have to wrap it in some type of non-disclosure agreement to keep it private. I think a lot of people would pay for that access regardless. Just a thought.
I did like the way you didn’t just slag it off and kept very constructive about it being of some value. Good effort guys. Well produced.
is the video out of sync for anyone else?
To be fair you did ask it to get you from v4 to v6 specifically so I understand why it wants you to project more.
If you asked it to just make you a better climber in general it might have included more volume climbing.
Interesting take. At first glance I am not sure I would treat the two differently i.e. climbing harder and being a better climber. But my bias here might be that my approach to coaching a V4 climber would be to make them a better climber, so that they can climb harder. The opposite might be working to achieve a single V6 with very little foundation of V5. Personally I think projecting-only for 6 months is the wrong approach for most V4 climbers. But I still wonder if ChatGPT made this kind of distinctions, and if we asked it to write a plan to "be a better climber AND climb V6" it would have written something different.
@@LatticeTraining Yes, exactly :) I think it's very much a machine AI still. It will take what you give it quite exact. At least that is what I suspect. Would be interesting to see if there was a different response with a different input.
Nice work with the chat GPT content- I think it’s super useful to have conversations around how we can use it as a training tool. 🎉
I got really psyched using it for all sorts of things the last few months but am realising it can also lead you astray 😂
That's a lot of training hours for a V4 climber. Honestly there are way better plans available for free.
There’s a lot of potential for this technology in the fitness and sports industry. Especially as the software gains more memory and customizability. A human will be able to teach it what it should be doing. It will make coaches more effective and probably more profitable
Like an assistant you don’t have to pay.
Yes I think in the future we'll be playing a combined role of sorts. Ai doing a lot of the roles that can be systematized, allowing us as coaches to spend more time on the soft skills and fine tuning, such as communication, goal setting, chatting through tactics and detailed personalisation of a training plan.
@@LatticeTraining especially if it can remember past prompts long term and respond to feedback. You can say “write this week’s workouts for will, but make sure the weight range increase you provide is smaller than the last time you progressed it” At that point it can work like an automated spreadsheet.
I know who I'd prefer to give me a Plan - Lattice over ChatGPT any time ☺
Important to note that you're using GPT 3.5, not 4 which is far better :p. I think that's an important note that you missed to include. There is something to be said about prompt engineering and getting good output out of it. It's not a truth machine.
Yes great point. We are using GPT4. And the prompt engineering issue is why we think a good prior understating of training is needed, currently, if we are going to get anything good out of ChatGPT.
@@LatticeTraining Ah I was under the impression it was 3.5 due to the green icon of chatGPT ~10:00
@@LatticeTraining Think you guys were confused around paying for gpt4 and using gpt4 as a model. You can see that the green icon is being used (gpt3.5 model) so you guys didn't use gpt4. You can also tell it isn't gpt4 as the response speed is way to fast.
@@Bros10 Oh right, how do you access gpt4?
@@LatticeTraining for non-technical usage (api / programming stuff), go with the plus (paid version) will enable gpt-4
I think ChatGPT can't really replace actual coaching, particularly not the kind of stuff Lattice has been putting out or Hooper's Beta. With that being said, I've actually been following a training plan ChatGPT put out for me, which is similar to the one you've outlined here but a bit faster (v4 to v6 in 2 months, 6x week). I'm already at the v5 stage (i.e. I'm projecting - I'm not really a v5 climber). Whether or not I'll actually reach v6 remains to be seen, but I'm pleased with the results so far. Although I have to say it's been rough doing 6x week, I'm having to be pretty strict with nutrition.
Wow 6x a week! Are some of these quite light training days e.g. active recovery? It sounds like a very heavy training plan. Also how many days a week were you training before you started the plan?
6x a week sounds really scary unless like the Lattice account mentioned above that's including multiple active rest days. If you're actually climbing and training 6 days a week I'd be seriously worried that you would be more likely to get injured rather than hitting your V6 goal.
@Lattice Training there are two dedicated "rest days" with some light training that involves antagonists and Fingers. I was doing 4x week before starting this new training protocol consisting of 2x climbing 2x weight training. Much prefer this new protocol though considering that it gives me more climbing days (which is what I really enjoy!)
In my limited experience, intense training that isn’t specifically climbing related is only really useful in the V9/7C or harder range. Up til that point, for almost every beginner to intermediate climber is best using their time actually climbing 90% of the time and strength training on campus/hang boards the other 10%. Standardized (moon/kilter/tension) board climbing is also a severely undervalued training tool by intermediate climbers that can rapidly build strength and technique.
I believe i was that climber who posted that training plan on reddit.
Have you been following the plan since?
*the thing about ChatGPT is that it is a standalone device. As of September 2021, the "Knowledge Cutoff" date it cannot access anything beyond its blackbox. Therefore, any training methods would have been peer reviewed and/or debunked since then. Of course it wont have all the answers.*
Yes, its not going to be current but I think the mistakes made in this plan (based on our prompts) are not based on peer-review research. These issues would have been bad in 2020.
@Lattice Training
1) True, we're just getting over Finger Antagonist Training these days.
2) Damn, you replied before the video ended, cheers!
Please don't call it open source - OpenAI is only open in name. There are real open-source efforts in this space, but OpenAI is definitively not open in any sense. Much like any country with "democratic" in the name is probably not democratic.
Thanks for flagging. We've removed this from the video but UA-cam is taking its time to render the new edit.
Good observation on "democratic"!
Would've been nice to have done it together with someone who is avtual a v4 ish climber
#NotAnAd
Just kidding. Love the content. Fun idea.
What a surprise the coach doesn’t like AI taking his job.
All you have to do is climb hard to get to V6. No hangboarding, no extra training, just climb smart safe and hard. Be on the wall as much as you can, because the less resting you do, the more you will develop the ability to climb more routes per hour. V7 and beyond probably requires more, but I got my first V8 clear by only using hangboard by feel pushing to the hardest edges I can hold for 5-10 seconds. Best to not overcomplicate the simple. Thinking is a waste of energy until you're actually in the real grades.
I don’t think this is true for everyone, especially depending on weight and body type. Some people also might want the most efficient, high yield way to accomplish this goal as well, which may not always be “just climb hard”
All you have to do is go to the gym with the easiest grading you can get to, climb V6 there, and viola. Never try at a harder gym and *definitely* don't go outside. There we go, you are now a V6 climber.
Interesting ideas but honestly what is a coach going to tell you? Don't get injured? Wait for your tendons to condition? Use your feet mindfully? I think UA-cam itself has made coaching redundant until you're an advanced climber. It's not like you can pay a coach to make your finger tendons recover faster.
Terrible advice
@@arkadaurum298 Found the guy who can't climb V6 at the super soft gym. Have you tried shirt off, beanie on, and power screaming?
Now if lattice were to share their information to the open source community maybe we could have better information for climber's... Instead lattice hoard's all the information they obtain for themselves. At least the power company release's their data. AI might be a good resource if you would realse your data.
I think the community will catch up on this eventually. And GDPR means that dissatisfied athletes can always call lattice up and ask to have their data removed.
We've routinely shared our data and research helping our community understand training, observations and furthering sports science. We don't share our database of individual data points, that we have collected from years of assessments because this acts as one of the foundations of our business. The assessment service we offer supports our company and the people we employ. We want, and do, share as much information as possible including all the free content on this UA-cam channel, our blogs, research papers and Instagram, but we also need to protect our company and its employees to keep doing what we do, which I am sure you understand.
They are still have to keep their business alive. The informations that they released to the public are already a gold mine.
Open source is great but I guess no sane company would give away their biggest competitive advantage.
@@LatticeTraining You could always charge a fee for access to that database of individual data points. I imagine that could be a nice additional revenue stream for Lattice. Might have to wrap it in some type of non-disclosure agreement to keep it private. I think a lot of people would pay for that access regardless. Just a thought.