BBC News - Can this high-intensity interval bike boost your workout?
How can anyone put a £3000 price tag in this?
4 Likes
I thought someone was offering to give me £3k for doing Spanish Needle.
Dissappointed now. 
22 Likes
[…] based around the concept that two 20-second sprints within an eight-minute gentle cycle are just as good for you as a 45-minute run.
Gentle cycle they say 
Clearly they are not talking about Spanish Needle, and have yet to meet Coach Chad 
3 Likes
This, in any other guise, would be classed as a scam. The “science” they spout isn’t true, the bike looks like a regular home gym bike that you would pick up for under £200 and I fail to understand how AI enters the equation; seems like they’re simply using it as a buzz word. Shame on the BBC for effectively advertising this product.
5 Likes
I agree with you . Hope people are not buying them.
1 Like
It will be amazing next to my AbbRoller, my AbbKing, my Stair Master and my collection of 2 minute workout DVDs. BTW I love spending hours watching infomercials telling me how to get in shape without having to actually workout in just 2 minutes a day!
#couchpotatotoheroinaday
2 Likes
Not quite 2 mins…
2 Likes
Please delete this before Coach Chad learns about ‘Supra Maximal’ workouts 
3 Likes
And I am sitting here looking at different bikes, having problem what to choose…and here is such a treat! And it is within my budget! Thank you TR!
1 Like
Yeah, well, technically (or rather statiscally), supervised learning includes regression, which is still called AI even though it is a rather easy and relatively straightforward statistical analysis. If they’d just collected data on a representative sample (let’s say demographics and peak power) and they tune in those parameters and using a formula as simple as addition and multiplication they can tell you which watts you should be able to hit for 20 seconds, you can statistically call it AI.
But other than that, the science does not check out. I mean it’s in the “ad” itself: acute glycogen deprivation = training effect? And it’s equivalent to an “endurance” ride? So basically (exaggerating), it’s equivalent to doing a Pettit? That’s it? And that costs 3000 pounds? Doesn’t make sense to me.
I think it’s nothing more than appealing to “lazy” people or “time pressed people” and comfort / easy in living. “Look I am a professional athlete, science says so, and it’s just couple of 20 seconds a week”.
EDIT: btw… like your nick name, dude
2 Likes
I must admit my knowledge on AI is extremely limited. I thought AI is some sort of machine learning or problem solving. You feed it data and it works something out on it’s own without having a pre-programmed formula. It sounds like what they are doing could be done using a spreadsheet! Perhaps I missed something, but I have no intention of rewatching the ad.
The dude abides!
1 Like
Oh what a surprise. The participants in the study were “physically inactive”.
Also, what the BBC thing doesn’t tell you is that the exercise program in their scientific study didn’t just have the 2x20sec sprints with 3min recovery (which is all that the marketing talks about), but the following workouts as well.
So yeah. Marketing scam.
1 Like
Sixty 8s on / 12s off sounds like a pretty decent workout! Also not convinced anyone ‘physically inactive’ could do that without breaks in between.
She makes reference to that at the end of the vid.
IMO this is marketing at it’s best and I’m sure some people without sports background will buy this.
What I really find interesting is all the AI power and top of the edge tech but… you don’t have a “live” condition update. Is all based in past exercises. So if you are feeling tired or not it doesn’t matter. What’s the point?
Ah. Buried at the end!
I’d love to see a study of people using this bike, and people being just told to do the same exercise on a dumb exercise bike.
TBH the whole thing makes me look a lot more kindly on Peloton. At least with them, your money is going on live lessons and the coaches’ time. With this, it’s going on a workout library of 3 workouts.
Artificial intelligence is the answer to how we’re going to solve any unsolved hard problem in computing. Once it’s solved, it turns out the solution has a boring name, like k-means clustering.
It’s a little silly to call many of the machine learning algorithms “artificial intelligence”, especially something as simple as regression.
1 Like
Hard problems are indeed part of artificial intelligence. But soft problems (which typically can be tackled using data training sets and supervised learning algorithms) are as much as part of AI as are the hard problems. Read Life 3.0 by Max Telmarck, and also see the machine learning books by Trevor Hastie and colleagues (both published by Springer).
Though I agree that regression is simple and calling it “we used this super cool new AI” where it is just regression that has existed since forever basically is stretching it really far.
Most AI problems (the soft problems) are still being solved by supervised machine learning algorithms (oftentimes classification). The next advanced thing are neural nets. But neural nets have been oversold as the default answer, AI is much more than neural nets. They just tend to fit many problems we are currently trying to solve rather well, similar to how the normal distribution in statistics tend to capture a lot of distributions quite well in nature.
EDIT: by “many problems” I mean monstly in engineering (Tesla, Apple, Google, etc.). I know nothing about medicine and biology etc (though neural nets are based on how the think our brain works, obviously).
1 Like
… And that’s how you learn something new!
Cheers to all