The Ambient Food Auditor
Waiting for the technology to catch up.AKA Shoulder Bastard
It started with MyFitnessPal
Food logging has one fairly obvious weakness.
It only knows what you tell it.
Which means I can reach the end of the day, look proudly at MyFitnessPal and think:
Look at that. Only 1,600 calories today.
Excellent work.
Except…
Did I log those crisps?
What about the thing I ate while standing in the kitchen?
Did I remember the handful of something I grabbed earlier?
Probably.
Surely.
Let’s go with yes.
And that’s when I started thinking that food logging would be considerably more informative if the person doing the eating wasn’t also responsible for the auditing.
What I actually need is a witness
Imagine a tiny pair of AI eyes.
Not something requiring me to photograph every meal.
Not something where I have to open an app, search for the food, find the right serving size and manually record it.
Just something aware enough to notice:
Sue appears to be eating crisps.
Then check later.
No crisps logged.
A discreet voice in my ear:
“I saw you eat those crisps. I did NOT see you log them. Shall I do that now for you?”
“Er, yes. Of course.”
Bugger.
That’s the entire concept.
Not an AI deciding what I’m allowed to eat.
Not something congratulating me for choosing a carrot.
Not an electronic diet policeman making disappointed noises because I’ve opened the biscuit cupboard.
Just an independent record of what actually happened.
Because memory is a terrible database
Manual food tracking introduces friction.
And every bit of friction creates another opportunity to think:
I’ll log that later.
Then later arrives and the snack has vanished from both the kitchen and recorded history.
An ambient system could remove most of that friction.
It sees food being eaten.
It identifies what it reasonably can.
If confidence is high enough, it offers to log it.
If it isn’t sure:
“Was that cheese?”
“Yes.”
“Approximately how much?”
“I don’t know. A bit.”
And already we’ve captured more information than the current system contains, which is:
Cheese: 0g.
Unfortunately, it requires eyes
Which brings us back to the recurring technological problem.
For this to work without becoming more annoying than manual logging, the system needs some awareness of what I’m doing.
A phone sitting on the table isn’t enough.
It needs wearable, ambient vision.
Which sounds suspiciously familiar.
Yes.
This is effectively Roiboty Cricket’s evil cousin.
Cricket sits on my shoulder helping me do things.
The Ambient Food Auditor sits there accumulating evidence.
Cricket:
“You’ve been meaning to fix that cupboard hinge.”
Food Auditor:
“Interesting. And would you like to tell MyFitnessPal about the Hobnobs?”
Bastard.
Current condition
Waiting for the technology to catch up.
Computer vision already exists.
Food recognition exists.
AI can already interpret images surprisingly well.
Wearable cameras exist.
The pieces are gradually assembling.
What I want is for them to disappear into something small, unobtrusive and effortless enough that recording what I eat becomes easier than conveniently forgetting to record it.
Until then, MyFitnessPal and I shall continue operating on the honour system.
And nobody needs to mention the crisps.