faangguyindia
Joined 4 years ago
Joined 4 years ago
A 2020 systematic review by Fuller et al. found that consumer wearables often perform poorly at estimating energy expenditure, even when they measure steps and heart rate reasonably well. Across brands like Apple, Garmin, Polar, and Withings, calorie burn estimates frequently showed substantial error, often in the 15–40% range and sometimes exceeding 50%. Wearables can still be useful for tracking trends, but not for precise calorie calculations.
protip: if you actually want to do better, look into "adaptive TDEE".
They'll use flawed research to claim diet and exercise don't work for obesity, then the general public uses that research and claims, "Diet doesn't work for fat people."
Soon, everyone is convinced every guy with a trivial amount of weight to lose needs to be on GLP-1s.
This is one of the biggest quality of life boosts provided by AI, yet many everyday people are completely unaware of this.
The price of an AI solution is $0, and within 2-3 weeks you'll know if it works for you or not.
Compare this to any fitness coaching, or dietician, or any other method, or even GLP-1s, the cost is not zero.
Many people who do not even need GLP-1s are using GLP-1s to lose a trivial amount of weight, even though their appetite regulation is not a problem!
The thing is, they are refusing to believe that it's even possible.
There are many people using steroids and peptides, if you put a starving person in famine affected sub Saharan Africa on steroids, can you have them return to normal weight without providing them additional food? But many "foolishly" believe this is possible with steroids and peptides without giving them more food.
Users spend tokens on DeepSeek, Claude, and others -> generate variations of projects uploaded to GitHub -> Microsoft (GitHub's owner) -> trains LLMs on this -> ends up with better models.
No matter what model a user may use at any price point, microsoft can distill it.
It's distillation at a higher level
most good programmers aren't going to commit garbage solutions; they'll push it to a git remote once they see it actually solves a problem they are trying to solve, and since the git commits and their messages are also AI generated, you've good visibility into the user's intention and resulting edits to the codebase.
They want to capture it all.
I know many people who have got banned and never managed to get back on Reddit because Reddit is using some very invasive and persistent banned user detection techniques.
Thing is, they use AI to ban people; over time, many false positives accumulate and not everyone has the time and energy to appeal bans.
Slowly, more and more user accounts get banned, and even technical people I know fail to make a new account which lasts more than 3-4 days.
But the Reddit management is seeing this as a "win," not able to see the consequence of doing this.
for me it's over for python