3 ms·

"I wonder how someone 40 years in the future would look back at 2021 iPhone pictures of themselves? " I make so many pics and videos that I really hope it will

"I wonder how someone 40 years in the future would look back at 2021 iPhone pictures of themselves? " I make so many pics and videos that I really hope it will be something like this: "Hey locally run AI that guards my data and privacy, please show me about 100 pictures and videos of holidays when the kids were in the age of 5-10, project them on the wall.", "Yes, that one in the mountains is nice, where is that? Can you show me some more of that holiday?", "How about some videos taken around that era in and around the house?" "Did we have a dog back then?" "Ah yes, there is Bello, what a nice dog that was, when did he die?" [Star Trek computer voice: "last picture of Bello was may 5 2014".] No way I'm ever going to sort this mess manually. TBH, a simple import in Shotwell already works wonders, and I absolutely love the "For You" section in "Photos" on my iPhone, I can get lost in there for a long time, but that's just a bit too much tied to Apple for my taste though, I'll be archiving all those files at some point (to 2 local hdds and then rsync them off-site (aka "my parent's house")). Edit: Maybe, at some point, you can construct an interactive narrative about your life like done in The Orville in "Lasting Impressions" [0]. Why are you all so negative about the future people, you need more Star Trek (or The Orville) in your life ;) Nobody is going to get overwhelmed, we're creating a rich source of information about ourselves and the time we live in, which we will probably never really consume as a series of Photographs, that sounds pretty boring indeed, we can do better, show some creativity! [0]: https://www.imdb.com/title/tt7826096/ https://www.imdb.com/title/tt7826096/

ekianjo 5y ago
AI can't even be bothered to do half-reliable text translations, seriously doubt that 40 years from now it will work half as well as we imagine it to be. After all, we all imagined back in in the 80s that we would have flying cars and robots around us by 2020's - we should learn that we are constantly bad predictors and completely off the mark.
- teekert 5y ago
  No way. The Photos section of the iPhone Photos app is already almost there. It makes the nicest videos from live photos (.heic), with music and it chooses the nice moments by itself. Maybe this not-super-critical-and-necessarily-100%-specific-part is where AI already shines. The "For You" content has complicated titles like "$kid'sName spring 2020" or "A day in the park, summer 2021" or "Happiness". 40 years is a lot.
- huntertwo 5y ago
  While I agree that the expectations for AI are too high, I think the use case described is almost achievable with today's tech. It's just language processing to process the request (User wants to see dog around this time period in this location) and finding relevant images (find pictures with dog with exif data that matches time period and location).
  - gwern 5y ago
    You could do most of that with some creative use of CLIP. 'Show more of the mountain' is trivial out of the box, and the 'kids aged 5-10' just needs a bit of logic: take a few photos of each kid at various ages 5-10 (perhaps already specified by the user?), embed, average along with the text prompt 'kids aged 5-10', and use those as the targets. 'Around that time' and 'in and around that house' can be done by metadata dates + GPS + image embedding of the 'house' (or just the text embedding of the word 'house'!) to filter photos by date+location, and then similarity to 'house'. And so on. The question there is how to be able to create those queries automatically in response to transcribed user voice commands. But hey, given Codex/Copilot, who can doubt that programming such queries is out of reach of a contemporary NN either?

ekianjo 5y ago

AI can't even be bothered to do half-reliable text translations, seriously doubt that 40 years from now it will work half as well as we imagine it to be. After all, we all imagined back in in the 80s that we would have flying cars and robots around us by 2020's - we should learn that we are constantly bad predictors and completely off the mark.

teekert 5y ago
No way. The Photos section of the iPhone Photos app is already almost there. It makes the nicest videos from live photos (.heic), with music and it chooses the nice moments by itself. Maybe this not-super-critical-and-necessarily-100%-specific-part is where AI already shines. The "For You" content has complicated titles like "$kid'sName spring 2020" or "A day in the park, summer 2021" or "Happiness". 40 years is a lot.
huntertwo 5y ago
While I agree that the expectations for AI are too high, I think the use case described is almost achievable with today's tech. It's just language processing to process the request (User wants to see dog around this time period in this location) and finding relevant images (find pictures with dog with exif data that matches time period and location).
- gwern 5y ago
  You could do most of that with some creative use of CLIP. 'Show more of the mountain' is trivial out of the box, and the 'kids aged 5-10' just needs a bit of logic: take a few photos of each kid at various ages 5-10 (perhaps already specified by the user?), embed, average along with the text prompt 'kids aged 5-10', and use those as the targets. 'Around that time' and 'in and around that house' can be done by metadata dates + GPS + image embedding of the 'house' (or just the text embedding of the word 'house'!) to filter photos by date+location, and then similarity to 'house'. And so on. The question there is how to be able to create those queries automatically in response to transcribed user voice commands. But hey, given Codex/Copilot, who can doubt that programming such queries is out of reach of a contemporary NN either?