Background and comments please on Sensei

  • 2
  • Question
  • Updated 12 months ago
  • (Edited)
In a blog post written by Michael Steeber dated Oct. 24th 2017 https://9to5mac.com/2017/10/24/interview-adobe-tom-hogarty-future-of-mac-iphone-x-camera-photography...

Michael write: Sensei is described by Adobe as a creative assistant, technology that
learns from each user and is trained by hundreds of millions of assets
in the Creative Cloud library.


I am wondering how Sensei has access to hundreds of millions of assets. Does it have access to my assets that may be stored on the cloud? Are those assets also available to all humans, or only selected humans? Who gave Sensei access to all those assets? Do we have the option to opt out of the license that Sensei seems to have to see and use our assets?

And who is really learning from Sensei? Will this result in the homogenization of all photographers with the net outcome being the dilution of our ability to recognize and celebrate true human creativity?
Photo of John Hansen

John Hansen

  • 31 Posts
  • 8 Reply Likes

Posted 12 months ago

  • 2
Photo of Rob Rippengale

Rob Rippengale

  • 83 Posts
  • 56 Reply Likes
Good questions. This is why I don't store my images in anybody's cloud. I don't wany Adobe deciding which of my pictures are the best or helping me become more like the photographer they want me to be, for whatever reason they might have. Individual growth and unique perspective is better than group homogeny. I also don't want Netflix telling me what movies I should watch. Don't let robots vote.

Maybe Adobe will assure us that no person or algorithm looks at the photos so many people are giving them. I doubt they will or can say that.
Photo of Victoria Bampton - Lightroom Queen

Victoria Bampton - Lightroom Queen, Champion

  • 4436 Posts
  • 1644 Reply Likes
> Maybe Adobe will assure us that no person or algorithm looks at the photos

Machines do look at the photos, because that's what's driving a lot of the tech. You can turn it off in your account if you don't want to allow machine learning though.
Photo of Rob Rippengale

Rob Rippengale

  • 83 Posts
  • 56 Reply Likes
Yeah, I noticed that clumsy phrase too late. But is this machine going to respect privacy? If person A tags a friend in their collection and the machine learns how to auto-tag them, it will then be able to find that person in someone else's collection. Maybe the person being tagged doesn't want to be found here, there and everywhere. These huge picture collections must be pretty tempting for a number of people.
Photo of Victoria Bampton - Lightroom Queen

Victoria Bampton - Lightroom Queen, Champion

  • 4436 Posts
  • 1644 Reply Likes
Facebook's been capable of doing that kind of thing for a long time. The difference is for a free product like Facebook, the customer is the product, so they have more of a vested interest in doing sneaky stuff like that. We're paying Adobe to spend their time using that tech for photography-related things instead, and that's where they'll make their money. They've already used it for the beginnings of tagging, there's new "best photos" tech in technology previews, and there's more to come.

With all that said, I think the privacy policy should be written in plain English instead of legal jargon, so we all know exactly where we stand.
Photo of Warren Heaton

Warren Heaton

  • 148 Posts
  • 54 Reply Likes
The moment I saw the Sensei demonstrations all I could think is, “Well, there go those billable hours.”
Photo of John Hansen

John Hansen

  • 31 Posts
  • 8 Reply Likes
This has me thinking more about the creative aspect of our work, and the licenses that are being monetized in this digital age. As I peer into the future, I see a vision of some large company claiming that our work was not ours alone and that we had the help of an algorithm that they wrote. In this brave new world, I see the first court filing where the author of the algorithm is asking the judge to make us share our royalties for our creative efforts because their algorithm did more than 50% of the creative work. Never mind the fact that we were the one who set up the camera to a precise location to capture just one of an infinite number of perspectives on the subject, and tripped the shutter at the decisive moment.
Photo of Rob Rippengale

Rob Rippengale

  • 83 Posts
  • 56 Reply Likes
It's a new-age problem. It would be somewhat true for the software owner to say they "made" the picture if they made all the filter and lighting decisions. Consider programs like Painnt (yes, double-N) which transform a photo into artwork. When I run the program on one of my images and a new image emerges, who made that? I pushed a button. Maybe I set some switches a little off the default, but still it is the software that generates the output. As we increase automatic photographic processing, who will own the images, especially if the original "snapshot" was given to a corporate server that did all the adjustments? And who owns the data that knows how to find Aunt Mildred in your photo collection -- can it be used to find her in someone else's collection and log her activities by date and location? The cloud seems to be a very sticky situation.
Photo of john beardsworth

john beardsworth

  • 1099 Posts
  • 258 Reply Likes
See Adobe's demonstration of how Photoshop's Content-Aware Fill may develop with machine learning - here.

But also see their Machine Leaning FAQ which includes "Can I turn off (opt out of) machine learning?"