Tag Archives: video

Surplus

De-Sora

I first mentioned OpenAI’s Sora last October in connection with the privacy concerns of its Cameo feature. Does Cameo create an innocuous likeness, or an evil deepfake?

As it turns out, neither. Sora is going to the tech graveyard.

From https://x.com/soraofficialapp/status/2036546752535470382

We’re saying goodbye to the Sora app. To everyone who created with Sora, shared it, and built community around it: thank you. What you made with Sora mattered, and we know this news is disappointing.

We’ll share more soon, including timelines for the app and API and details on preserving your work. – The Sora Team

Data Labelers Gonna Label, and Class Action Lawyers Gonna Lawyer

On Wednesday, I described how Meta’s Kenyan data labelers ended up watching explicit videos from people who presumably didn’t know that smart glasses were recording their activity.

To no one’s surprise, class action lawyers are now involved.

“In the newly filed complaint, plaintiffs Gina Bartone of New Jersey and Mateo Canu of California, represented by the public interest-focused Clarkson Law Firm, allege that Meta violated privacy laws and engaged in false advertising.

“The complaint alleges that the Meta AI smart glasses are advertised using promises like “designed for privacy, controlled by you,” and “built for your privacy,” which might not lead customers to assume their glasses’ footage, including intimate moments, was being watched by overseas workers. The plaintiffs believed Meta’s marketing and said they saw no disclaimer or information that contradicted the advertised privacy protections.”

So what does Meta say?

Meta. From https://www.meta.com/ai-glasses/privacy/.

“Clear, easy device and app settings help you manage your information, giving you control over what content you choose to share with others, and when.”

Except that according to Clarkson, people can’t opt out of the data labeling process.

This could get very revealing.

Many Ways To Conceptualize Stories, and Tell Them

Depending upon your talents and resources, your company may choose different ways to tell critically important stories to your prospects. But how do you get there?

Early this week, one of Bredemarket’s clients expressed an urgent need for a story. My job was to figure out the concept and pass it on to a talented person inside the company who would use my concept to create the final version.

Now I had no idea what format the final story would take. An infographic? A video? Something else?

Google Gemini.

But my concept didn’t need to be in the final format. It just had to contain the concept.

For all the client cared, I could have sketched the concept out in Microsoft Excel. Which works great for storyboards, especially when the story is fluid and needs to be re-sorted.

In the end, I used a different Microsoft product—PowerPoint. Not that it mattered.

Google Gemini.

When working with creative talent, you have to give them enough of your intent without constraining them. And I definitely did not constrain.

  • My PowerPoint used unformatted slides and default fonts.
  • The graphic concept that was central to the entire story consisted, in my concept, of three boxes with words in them. Later it became four boxes.
  • I used Google Gemini to create two subordinate concept images, but added indicators to show they should NOT be used. Even if the images were spectacular (they weren’t), we all know that my client couldn’t copyright them.
  • After I was supposedly done, I took one last pass through the slides and removed every unnecessary word.

I can’t share what happened after I completed the concept, but the creative talent had enough information to move forward.

And I saved my client a lot of time by performing the initial conceptual work so the client could execute immediately.

And that’s what matters.

Data Labelers Gonna Label

Before diving in, I should note that this is not just a Meta Ray Ban AI glasses issue.

This is an issue with ANY video feed that requires AI processing.

Because AI can’t do its job on its own.

To ensure that the AI is trained properly, an army of humans looks at the same data and uses data labeling to classify it.

We allow this when we sign those Terms of Service. And I personally believe it’s a good thing, since it helps correct errors from uncontrolled AI.

But Futurism notes the types of video feeds that the human data labelers have to label.

“I saw a video where a man puts the glasses on the bedside table and leaves the room,” one data annotator told the newspapers. “Shortly afterwards his wife comes in and changes her clothes.”

Grok.

Basically we record more than we should. One example: a bank card.

But regardless of whether data labelers are present or not, assume that any recording device will record anything, and potentially distribute it.