On Thursday morning, I attended a Q&A panel with 4 prime Samsung smartphone executives. Till 2025, Samsung was the world’s largest smartphone producer, and by affiliation, the world’s largest maker of cameras. It’s nonetheless the second largest after Apple.
Samsung handed me the microphone first. I requested:
We see a divide in society between individuals who need AI to do spectacular issues with their photographs and movies, and those that don’t need AI to do something with photographs and movies as a result of it’s eroding our capacity to consider that what we have now seen is actual, destroying the idea of photographic proof.
Metadata instruments like C2PA have completely didn’t stem the tide. Does Samsung have any new and completely different concepts on how one can forestall AI photographs from taking up the world?
Samsung’s 4 executives didn’t have any new and completely different concepts to share.
I’ll credit score Gained-Joon Choi, the cell division’s COO and R&D chief, for not dodging the query. He advised the room that the erosion of actuality is an issue and he needs to repair it.
However he, and different Samsung executives, instructed that the corporate must stability the will for photographic actuality with letting smartphone consumers be “extra artistic.” They handed the buck by suggesting it’s an industry-wide drawback, one which requires a broader dialog, and instructed that Samsung has already partially solved just by including a watermark to AI-generated photographs. A watermark that may simply be eliminated.
Given a gap later, an exec instructed our emotions towards AI-generated content material would possibly develop into extra favorable sooner or later.
Listed here are a few of these solutions.
“We acknowledge the problem, as a result of a number of content material is generated by AI,” Choi started. “On the one hand, folks wish to be extra artistic, so we consider we have now to supply an answer so folks could be extra artistic,” he mentioned. “Then again, it’s actually exhausting these days to differentiate the true photographs and movies from the faux ones. I feel this can be a drawback, we acknowledge that, and I feel it’s an issue we have now to unravel at an {industry} degree.”
“The C2PA, you could view it as a failure, nevertheless it’s nonetheless sufficient to supply a mechanism if folks wish to validate that these footage and movies are made by AI. I feel we have now to supply a mechanism so folks can use it. I feel it’s an ongoing effort all through the {industry} to unravel this drawback,” he added.
“I feel if we make joint efforts to unravel this drawback, I consider that we should always be capable of do it,” he concluded.
As my colleague Jess Weatherbed wrote earlier this week, we’re beginning to fear that this form of “the {industry} will remedy this collectively” rhetoric, and the C2PA itself, are substitutes for precise significant motion.
Maybe responding to our Samsung AI slop watch reporting, Samsung America exec Dave Das additionally chimed in to say that the corporate’s nonetheless studying about how a lot AI is suitable when creating its personal adverts. He admitted that the corporate has “simply began using some AI content material in our artistic, and the suggestions has been fairly clear.”
“We’re attempting to discern what’s the proper place to make use of it, and completely how one can be very clear about once we are utilizing AI generated content material vs naturally generated content material,” Das mentioned.
However Das nonetheless frames it as a balancing act between enterprise priorities, not a social accountability. “It actually boils all the way down to giving the creator selection,” he mentioned, suggesting that Samsung’s job is to seek out “the fitting stability.”
Later within the morning, KTLA-TV tech reporter Wealthy DeMuro felt out Samsung’s stance with a really completely different query. He requested whether or not Samsung would possibly wish to make it simpler for patrons to take away its AI watermark from AI-generated photographs.
“If I’m making a Christmas card with all of your enjoyable instruments, do I really need that to say ‘AI-generated content material?’” requested DeMuro.
Drew Blackard, Samsung America’s SVP of cell product administration, took that one. He concluded by saying that “If we really feel like shoppers really need that capacity to take away watermarks, and there are different methods to handle the primary level round authenticity on the similar time, we’ll do each issues and ship the most effective of what individuals are wanting out of the experiences.”
“At the least proper now, there’s sufficient concern from shoppers round authenticity that that’s the first factor we’ve solved for, when it comes to the watermarking each within the metadata and on the photograph itself. Not all companies try this,” he mentioned.
Blackard additionally instructed we would look again in the future and understand that AI-generated content material isn’t such a giant deal, that our perceptions would possibly change over time, the identical approach perceptions modified towards user-generated content material to start with. Individuals have been involved then with “the quantity of content material now flowing into the system” and the way it wasn’t the skilled video product they have been used to, he mentioned.
I’m wondering if Samsung and fellow smartphone makers have thought of the other: that perceptions of AI-generated imagery would possibly develop into even much less favorable after folks lose their jobs in favor of cheap-to-produce AI slop, and after it turns into far simpler to lie and cheat and steal within the absence of dependable recorded proof.
I’m wondering in the event that they’ve thought of fixing this drawback ”at an {industry} degree” earlier than they break the dam, and whether or not notion would possibly flip in opposition to Samsung for serving to create the flood that comes after.
