Comment by acchow
1 year ago
> The concept is simple: Take a picture of the food you are about to consume, and let the app log calories and macros for you.
> The result is an app that the creators say is 90% accurate, which appears to be good enough for many dieters.
It absolutely cannot be "90% accurate". But I'm sure it seems "90% plausible" to its millions of users.
Incredible that a product like this can exist. Do people just will the fiction into reality?
Incredibly, these are the types of app ideas you'd hear from non-tech "entrepreneurs" in 2012 looking for a co-founder. The problem being, the engineers knew it was impossible. You could fake it I guess by asking Google to search for "similar photos" and getting a plausible calorie count half the time. But the users wouldn't believe it.
We're now at the phase where any impossible idea can be fully marketable by slapping "AI" in the name. ChatGPT feels so magical that we now believe unicorns really do exist.
Truth is so overrated these days.
Exactly! The are so many alternatives, why do we have to place THE truth so highly?
With AI being weaponized by anyone who wants to make a profit, what do you expect?
[flagged]
Even the nutritional labels printed on packaging are only around 90% accurate. It’s all downhill from there.
Just read the Google Play reviews, the app still has a long way to go to be usable.
But it's still a great idea.
Also, they should calculate both the calories and power requirements for each meal analyzed. What I mean is, it should says something like: this burger has 800 kcal and the analysis consumed 1kw of datacenter power. ;)
> But it's still a great idea.
No, it isn't, because it can never work. It is, in general, impossible to determine the calorie content of food by looking at a picture of it.
kw is an instantaneous measurement. Energy usage would be kWh (kilowatt-hours). Considering it would take a tiny fraction of a server’s compute for less than a second this would be very small.
Right, speed vs distance. Still, I'd love to know the numbers.
>But it's still a great idea.
Define "great".
It has $2m revenue, so it's clearly a great idea (at this stage?) financially and 'people love it' (30% retention)
Technically it's a garbage idea, and I'd say they could get class-actioned without good T&Cs. It's literally impossible to determine the sugar and fat content of a meal.
I'd never make it for the latter reason, however you clearly need to believe in the former to make it big haha.
30% retention means that 70% of people who have paid for it didn't consider it valuable enough to pay for.
The $2m figure is unverified but, even if it was, turnover is meaningless without knowing their expenses. If they're burning through £2.1m in LLM compute and marketing then they're losing money.
Plus, the guy seems like a dick so I'm taking his story with an enormous pinch of salt.
"In CS, it can be hard to explain the difference between the easy and the virtually impossible."
It seems we're at a point where this obligatory xkcd [1] is no longer true.
1. https://xkcd.com/1425/
The bird thing was at least _possible_, though, to a reasonable level of accuracy. What they're claiming to do here is completely impossible; there is simply no way of even vaguely accurately determining the calorie content of food based on a picture of the food.
It took a bit more than 5 years and a (single) research team, though.
Anything's possible if you're willing to fake it!