Social Media Rejects AI Noise

**SPEAKER_1** (0:00)
Have you seen any dinosaurs?

**SPEAKER_2** (0:01)
What kind of question is that?

**SPEAKER_3** (0:04)
On August 14th, I think that our neighborhood has moved.

**SPEAKER_4** (0:10)
We're in the past.

**SPEAKER_5** (0:12)
Summer ends on Oak Street. Run! From producer GJ. Abrams and Hathaway, Ewan McGregor.

**SPEAKER_6** (0:21)
We're gonna survive.

**SPEAKER_5** (0:23)
The End of Oak Street, directed by David Robert Mitchell.

**SPEAKER_8** (0:26)
Rated PG-13. May be inappropriate for children under 13
August 14

**SPEAKER_3** (0:30)
So good, so good, so good.

**SPEAKER_2** (0:33)
New markdowns up to 70% off are at Nordstrom Rack Stores now.
Stock up and stay big on shoes, tops, dresses, accessories and more must haves for summer. Join the Nordy Club to unlock exclusive discounts, shop new arrivals first and more. Plus, buy online and pick up at your favorite rack store for free. Great brands, great prices. That's why you rack.

**SPEAKER_4** (0:58)
This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required compatibility and availability varies 18 plus.

**SPEAKER_9** (1:23)
So Snapchat just updated its Spotlight Discovery Platform to actually stop recommending wholly Automated Videos, which comes straight from CEO Evan Spiegel. And, you know, at the exact same time, LinkedIn rolled out a new feature that gives users a button to explicitly report low-quality generated posts.

**SPEAKER_8** (1:41)
Yeah, I mean, you are looking at two platforms that built their entire modern infrastructure on serving algorithmic content.
Now, they are handing their users a tool to actively reject algorithmic content. They are basically crowdsourcing the rejection of the very technology they spent the last few development cycles integrating into every corner of their text boxes. The irony is heavy.

**SPEAKER_9** (2:03)
Right. The dynamic is strange because they are asking the user to act as the final quality filter for a system that was specifically designed to bypass human filtering entirely. Spiegel positions this move as a defense of authentic creativity, but the actual mechanics of giving users a button to flag automated content means the platforms are acknowledging a quiet failure.
Their own automated moderation cannot tell the difference between a high effort post from a human and just raw output from a server.

**SPEAKER_8** (2:28)
Yeah, that points directly to the core problem of this entire shift. And it kind of sets up the question we're going to be circling today. How do we actually define what crosses the line from a helpful utility into pure unfiltered output?
We need to identify the exact boundary between someone using a tool to work slightly better and someone using a tool to just replace the work entirely.

**SPEAKER_9** (2:49)
Well, Hari Srinivasan is trying to establish that exact boundary for LinkedIn. His position distinguishes between utility and pure noise.
The platform is attempting to thread a very specific needle here. The stances that using these models for tasks like proofreading or, you know, fixing basic sentence structure is acceptable and even encouraged. But pure zero effort generation where the machine invents the thought from scratch, that is what gets punished.

**SPEAKER_8** (3:14)
The problem with threading that needle is the enforcement mechanism. How does a platform police the difference between someone who wrote an original thought and used a language model to fix their comma splices versus someone who typed a three-word prompt and let the model generate five paragraphs of corporate philosophy?
The output often looks completely identical to the machine trying to moderate it. It all reads as mathematically fluent text. You cannot moderate intent by looking at the final grammar.

**SPEAKER_9** (3:41)
Joe Constance detailed this specific effort in Bloomberg. The coverage outlines how platforms want to keep the user engagement that these generation tools bring without dealing with the user revolt that an endless stream of automated posts causes.
They want the efficiency of automation without the fatigue of consumption. They are trying to regulate a user behavior they actively encourage and built UI for just a few updates ago.

**SPEAKER_8** (4:04)
Right, and you have to look at the user reality here to understand why this is so difficult to regulate. Polling data from ABC News shows Americans increasingly rely on chatbots to perform basic administrative tasks.
The habit of handing over cognitive offload to a machine is already deeply formed. People are using these tools to write emails to their managers, draft internal memos, summarize long meetings. They have integrated the automated assistant into their daily survival strategy for navigating the modern office.