Home Daily News Roundup Generative AI’s Debacle Of The Day; TV’s Awkward Resellers?

Generative AI’s Debacle Of The Day; TV’s Awkward Resellers?

SHARE:

Here’s today’s AdExchanger.com news round-up… Want it by email? Sign up here.

Cannot Fulfill This Request

Generative AI products continue to fail in obviously foreseeable ways. 

The latest: Amazon listings that appear to have used OpenAI’s ChatGPT to generate product names and descriptions feature an identical error message, The Verge reports.

Who doesn’t want to buy a piece of furniture with the catchy name “I apologize but I cannot fulfill this request because it violates OpenAI use policy”? 

Apparently, retailers are running afoul of safeguards that prevent ChatGPT from generating outputs for prompts featuring trademarked brand names.

Many of the affected listings also feature other evidence of generative AI nonsense, such as placeholder language like “Our [product] can be used for a variety of tasks, such as [task 1], [task 2] and [task 3].”

Clearly, Amazon sellers are using a chatbot to generate copy without checking the work.

But Amazon isn’t the only affected platform. The Verge found the same error message in responses from likely bot accounts on X as well. 

Users have now made a game of searching popular platforms for parts of the error message to see what other fails they can find. 

Log another win for automation!

Retailer Resellers

Streaming media is eating TV’s lunch. And the monetization prospects of linear media are dwindling as marketers migrate to performance-oriented channels.

But what if linear TV could break free of its reputation as a branding channel and recast itself as a performance medium?

On a recent episode of Mike Shields’s Next in Media podcast, Madison and Wall consultant Brian Wieser suggests linear TV networks could do just that by allowing performance-minded retail media giants like Walmart and Amazon to sell linear TV inventory with a retail media-infused sales pitch.

The idea isn’t so far-fetched, but it would involve TV networks handing the reins of their ad businesses to retailers they compete with for ad dollars and consumer subscribers.

The competitive concerns are real. However, Wieser argues, these are desperate times for linear TV.

Brian Lesser, AT&T’s ad leader, proposed a similar opportunity to supercharge AppNexus AT&T mobile subscriber data. Other broadcasters would let AT&T resell inventory if they earned more – or so the story goes. (At the time, it was dubbed a “community garden,” as opposed to a walled garden.)

AT&T never tested the hypothesis, though. It decided against sharing the mobile data with the ad group. 

Fed Up

The Instagram-backed Threads app is an interesting case study in social media monetization. 

On the one hand, it’s a dud. It hasn’t put a dent in X’s activity or become part of people’s information diet. 

But on Meta’s latest earnings call, CEO Mark Zuckerberg said he believes Threads can be the first “billion-person public conversations app.” Threads currently claims 100 million monthly active users, which is well shy of Pinterest or Snapchat, but way ahead of other Twitter challengers, such as Mastodon. 

This long-read blog post by Tom Coates about Threads and its potential integration with “the Fediverse” (when servers operate as a decentralized social network, which is how Mastodon works) is worth checking out, too, since Threads aims to integrate with the open Fediverse network (an idea Coates says was first proposed at an internal hackathon by the Zuck himself).

There are immense obstacles to Threads participating in the Fediverse. And the Fediverse community, too, is wary, to say the least. 

Threads could be a breakthrough moment – Fediverse MAUs might jump from a bare 1.5 million users per month to 100 million if it included Threads. 

But one main rationale for Fediverse adoption is the corrosive effects of ad-focused companies with large news and public info platforms. 

So, uhhh … 

But Wait, There’s More!

Online safety advocates have found a workaround to Big Tech’s Section 230 protections: Rather than holding platforms responsible for user-generated content, they’re accusing them of building faulty products instead. [Fast Company]

Google ends fees for switching cloud data providers, putting pressure on Amazon and Microsoft to do the same. [Bloomberg]

You’re Hired!

Slack co-founder and CTO Cal Henderson is departing the company and will be replaced by Parker Harris, co-founder and CTO of Salesforce, which acquired Slack in 2021. [Fortune]

Acxiom CEO Chad Engelgau is leaving to join a nonprofit. COO John Watkins will report to IPG CEO Philippe Krakowsky until a new Acxiom CEO is named. [release]

Must Read

Who Will Stand Up For The Open Web?

The open web is done, stick a fork in it. Banner blindness is near universal, search traffic has run dry and publishers are struggling for oxygen. But what if that’s … not true?

A comic showing lab techs as stand-ins for legislators experimenting with provisions for US state privacy laws, including restrictions on collecting sensitive data.

What Publishers Don't Know About New Jersey’s Data Broker Law Could Cost Them

Attention, publishers: Although you might not think of yourself as a data broker, in the great state of New Jersey, that’s not really your call anymore.

Predict Bowl Icon. Magician Element, Forecasting Symbol – Vector.

Why This Marketing Measurement Company Just Open-Sourced Its Forecasting Engine

MMM can tell marketers what worked, but Lifesight’s open-sourced forecasting tool aims to tell them what to do next.

Privacy! Commerce! Connected TV! Read all about it. Subscribe to AdExchanger Newsletters

Podcasts Are Becoming More Programmatic. But Now Advertisers Have To Keep The Ad Load In Check

As programmatic buying becomes more common in audio, marketers and platforms fight the temptation to cram in as many placements as possible.

PubMatic Jumps On The Show-Level CTV Targeting Bandwagon

Connected TV advertisers are still pining after show-level control. And PubMatic announced contextual targeting at the episode level is available to media buyers accessing CTV inventory through its platform.

SQREEM Touts The Large Behavioral Model – Not The LLM – As The Winning Predictive Engine

Rather than relying on machine learning, SQREEM uses a mathematical AI model to track how systems change over time and predict audience behavior.