Years ago, when people first started speculating about AI, I bet the last thing they thought of was that it will be bad at maths.
6 Likes
Yes it can be read different ways. The first request is for cycling workouts that progress and result in an increase in threshold TTE. And then a second request to start with a specific workout.
From that point of view, these 3 requests could result in an answer with the exact same progression, and only differ in the first workout:
-
Please build me a cycling progression of workouts to increase my threshold time to exhaustion starting with 4 x 8 minutes at threshold
-
Please build me a cycling progression of workouts to increase my threshold time to exhaustion starting with a recovery ride
-
Please build me a cycling progression of workouts to increase my threshold time to exhaustion starting with a kettlebell swinging workout
Meanwhile, Bard’s suggested training plan for a 65 mile gravel race with 9,500 feet of climbing (Gravista) would likely kill me…
2 Likes
I asked the same question to Bing AI and got a far superior answer:
According to one formula1, the power necessary to climb a hill is given by:
bike + rider weight (kg) x 9.8 x elevation gain (meters) / time (seconds) = power (watts)
This formula also suggests adding 10% for rolling and air resistance.
Using this formula, we can estimate the power required for your scenario as follows:
- Assume your bike weight is 10 kg, so your total weight is 82 + 10 = 92 kg
- Assume your speed is constant at 12 km/h, which is 3.33 m/s
- Assume your elevation gain is 5% of your horizontal distance, which is 0.05 x 3.33 x 600 = 99.9 meters (600 seconds is 10 minutes)
- Plug these values into the formula:
92 x 9.8 x 99.9 / 600 = 152.6 watts
- Add 10% for rolling and air resistance:
152.6 x 1.1 = 167.9 watts
This is the power you need to climb the hill at your current weight and speed.
Now, if you lose 10 kg, your total weight becomes 82 kg, and everything else stays the same. The power required then becomes:
82 x 9.8 x 99.9 / 600 = 133.4 watts
133.4 x 1.1 = 146.7 watts
This means you can save about 21 watts by losing 10 kg while cycling uphill on an average 5% climb at 12 km/h.
Of course, this is only an estimate based on one formula and some assumptions. There are other factors that may affect your power output, such as wind, road surface, bike efficiency, pedaling technique, etc.
I hope this helps you with your cycling goals!
1 Like
I bet the person the otherside of the server that had to get their calculator out was cursing you.
3 Likes
I hope your plan for losing 10kg whilst cycling uphill involves jettisoning bottles and things from your bike bags, otherwise this sounds an unpleasant prospect.
3 Likes
A truly intelligent response would be for it to ask you if you’ve really thought properly about the question.
2 Likes
This is part of my point.
Yes I agree.
Guess more learning and work needs to be done.
the other part of my point is that to increase TTE at threshold, a truly intelligent response might be to not do progressive threshold intervals and instead do other types of intervals.
Yes is seems to have gone with the basics of what works when there are plently of better options.
It is weird how it responded like a human though, I think a majority would have responded like that, (including many coaches, and the extra 20 minute threshold session, although I not sure if it understood or intended that.)
It responded like a human, and a majority of coaches (poor ones maybe), which was my point.
Why do people obsess over calculation and test methods for FTP?
“Against ignorance, the Gods themselves contend in vain.”
Large Language Models are now at that tipping point, one reason so many people are talking about ChatGPT and other LLMs.
Why do people ignore basic physiology, exercise science, and what coaches discover? At this point surrendering to a machine is not the answer.
Don’t obsess about the upper threshold or what to call it - FTP, critical power, the boundary between heavy/severe domains - simply go out and estimate it the best you can, and then continue to refine that estimate. Its really not that hard unless your idea of training is to tune out, ignore your body, and tune into some form of outside visual/audio entertainment.
The motivation to do that is pretty simple - to spend most of my interval sessions working clearly above and below it.
- I think Douglas Adams already said this a while back?
2 Likes
I’m suspecting ChatGPT didn’t say this bit… 
2 Likes
Maybe we should change this thread title to “Ask ChatGPT”
I asked the following;
“what aero road cycling helmet tested the fastest”
and was answered with;
"There have been various aero road cycling helmets tested for their speed, and the results may vary depending on the testing conditions and methodology used. However, one of the helmets that has consistently performed well in wind tunnel and real-world testing is the Giro Vanquish MIPS.
In a wind tunnel study conducted by Specialized in 2017, the Giro Vanquish MIPS was found to be the fastest road cycling helmet tested, with a saving of 0.8 seconds over 40km compared to a standard road cycling helmet. In a similar wind tunnel test conducted by AeroCoach in 2018, the Giro Vanquish MIPS was also found to be the fastest helmet tested, with an average saving of 1.5 watts over a standard road cycling helmet.
In real-world testing conducted by the CyclingTips website in 2019, the Giro Vanquish MIPS was also found to be the fastest helmet tested, with an average saving of 4 watts over a standard road cycling helmet in a 40km time trial.
It’s important to note that while the Giro Vanquish MIPS has performed well in these tests, the best helmet for an individual rider may depend on factors such as their head shape, riding position, and personal preferences. It’s recommended that riders try on different helmets and choose the one that fits them best and feels most comfortable."
7 are gone, another 10 to go. 
1 Like
You should be able to sign up for Bard with a Google account. If it’s demanding a phone number, are you trying so use some kind of weird account?
It would be interesting to try these queries with an LLM that’s appropriately primed, as opposed to the weirdly-primed ChatGPT / Sydney / Bard.
Are Over / under intervals better than Z2 volume?
“ It really depends on the type of audio you are looking for and the features you require. Over/under intervals may be better for raising power throughout the day, but ultimately it comes down to personal preference and the sound quality and noise cancellation offered by headphones like the Sony WH-1000XM3 or other popular models.”
Sounds like it really knows what it’s talking about
3 Likes