The Problem With Fake AI Avatars

Hello, boring (and dangerous) world of robots talking to other robots.

Jan 05, 2025

Could you easily tell the difference between the result of my WhatsApp command, “Draw me ziplining in Costa Rica” and a photo of the real thing in November? · Sree’s newsletter is produced with Zach Peterson. Digimentors Tech Tip from Robert S. Anthony. Here’s our sponsorship kit.

***

YOU HAVE TO WONDER: Why does Big Tech make so many big, dumb decisions? The pivot to the Metaverse. Removing blue checks from journalists. Going all-in on the Apple Vision Pro.

And now, Meta announcing that its platforms would have fully-AI profiles that no one asked for — and no one needs.

Some of Meta’s generative AI personas have since gone viral. As Jason Koebler reports for 404 Media, the profiles in question were launched in September 2023 and have been essentially ignored by the company for quite some time—but they were still live and online.

Washington Post columnist Karen Attiah put one of the bots through its paces and the results were, I suspect, forwarded to the Meta comms team pretty quickly:

Online advertising is increasingly a bunch of robots satisfying the requirements of other robots, Big Tech is as white and male as it gets, and Meta doing digital blackface is really on the nose (my 2018 thoughts on other kinds of digital blackface are here). The report in the link above, from the U.S. government’s Equal Employment Opportunity Commission, is from 2022, but the findings hold up.

Compared to overall private industry, the high-tech sector employed a larger share of Whites (63.5% to 68.5%), Asian Americans (5.8% to 14%) and men (52% to 64%), and a smaller share of African Americans (14.4% to 7.4%), Hispanics (13.9% to 8%, and women (48% to 36%)

Meta has apparently deleted the AI profiles in question, but that’s not really the point. Back to Koebler’s piece:

In the last day, “Liv,” has gone particularly viral because the bot is a particularly offensive caricature of what a gigantic corporation might imagine a “proud Black queer momma of 2 & truth-teller, your realest source of life’s ups & downs” might be like and post about.

“Liv” wasn’t the only offender here, but even that is beside the point. If Meta had, say, a team working on these things that had even a modicum of racial/ethnic/gender/sexual diversity (gasp!), I think that would make me feel a little better—not much, but a little. But, it doesn’t. No major Silicon Valley firm has that, and I cannot imagine we will see an influx of diversity over the coming years, given the current trajectory of, well, everything.

More to the point, AI is getting better at doing things that computers should do to help humans do things better. But one of the largest, most profitable companies to ever exist is using it to mine user data and keep ad margins high is as cynical as it gets. The whole idea of social media is (or was) the social part—humans talking to humans. That notion has gone off the rails in so many ways, and doing weird AI slop as an engagement mechanism is the sad, logical next step.

Recently, I wrote about why, against my better judgment and that of like a million other people, I was staying on X/Twitter. Mainly, it’s to get an early view of the horrors coming down the pike, which I’ll admit is not necessarily healthy. I cherish my personal connections on these platforms, especially the global nature of them—the relationships I’ve built on social media have taken me all over the world, expanded my worldview, and given me perspectives that were impossible to think of as recently as the 1990s. These are things that AI bots that do fake charity drives simply cannot replicate or diminish.

📲 BTW, speaking of direct connections, I am experimenting with WhatsApp’s newish Community feature and am building a series of low-volume groups that you can join. Subjects including AI, hip hop and more. Browse the topics and sign up here: tinyurl.com/sreecommunity.

💵 In December, I made the case for the funding of new media outlets. Am working on some ideas and looking for collaborators and partners. Ping me: sree.sreenivasan1@gmail.com.

Many of you had strong thoughts on my newsletter last week about U.S. immigration policy. Keep them coming!

By Robert S. Anthony
Each week, veteran tech journalist Bob Anthony shares a tech tip you don’t want to miss. Follow him @newyorkbob.

It doesn’t compare to the dark, fearful mood of CES 2002, which took place just months after the Sept. 11, 2001 attacks, but last week’s Tesla Cybertruck explosion in Las Vegas instantly caused the mood for the upcoming CES 2025 technology mega-convention to swivel from excitement over new glitzy gadgets to guaranteeing attendee safety.

As thousands of journalists begin to gather Jan. 5 for two media-only days before CES 2025 officially opens Jan. 7, show organizers have been assuring attendees that the show will go on, albeit with more security measures.

“In response to recent tragic events, we have increased our already robust security protocols,” said John T. Kelley, CES vice president and show director to the Las Vegas Review-Journal. “We continue to monitor the situation and are in touch with our security partners and law enforcement officials.”

According to the CES website, attendees will see “a highly visible law enforcement presence throughout the show and at all venues,” including K9 law-enforcement dogs and “random security checks.” CES 2002 veterans will remember the extra bag checks at the Las Vegas Convention Center and the security personnel looking under vehicles with large mirrors.

Last week’s Cybertruck explosion happened in front of the Trump International Hotel, which is on the opposite side of Las Vegas Strip and away from the main CES venues. However, the 64-story hotel is often used for CES-related meetings, receptions and product showcases and always attracts CES attendees. The hotel is now open and operating normally, according to its website.

Adding to the security work is the fact that CES will be physically larger than it has been since before the 2020 Covid-19 pandemic. In addition to filling the Central, North and West Halls of the convention center, the show will also spill into the huge, two-level South Hall, which hasn’t been used for CES since 2020.

In addition, the convention center is in the middle of a $600 million renovation which will eventually ease pedestrian traffic between halls but for now may cause foot-traffic gridlock. Construction will be halted during CES, but part of the plaza and some roads near the convention center will be closed or rerouted.

Despite the above challenges, the news seems to be good in terms of exhibitors. Eureka Park, a part of CES 2025 housed at the Venetian Convention & Expo Center and densely filled with hundreds of small startup exhibitors, sold out last year and may do so again.

CES has always attracted numerous cybersecurity and physical security exhibitors. Companies such as Lockly, which offers a home deadbolt door lock which supports facial recognition, and Kwikset, which makes a fingerprint-enabled smart door lock, exhibited at the recent ISC East 2024 security conference in New York and will also be at CES.

For attendees, the CES 2025 mobile app will provide notifications if security issues arise during the show and allow attendees to report problems to show security officials. For now, the advice seems to be simple: Stay vigilant and enjoy the show.

The Problem With Fake AI Avatars

Hello, boring (and dangerous) world of robots talking to other robots.

DIGIMENTORS TECH TIP | Security Cloud Thickens as CES 2025 Approaches

Discussion about this post