Over a Million Weekly Suicidal Chats: ChatGPT Safety Work

Brian, Beth, Andy, Anne, and Karl kicked off the episode with AI news and an unexpected discussion about how AI is influencing both pop culture and professional tools. The show moved from the WWE’s failed AI writing experiments to Grok’s controversial behavior, OpenAI’s latest mental health data, and a deep dive into AI’s growing role in real estate.

Key Points Discussed

AI in WWE Storytelling

WWE experimented with using AI to generate wrestling storylines but failed to produce coherent plots.

The models wrote about dead wrestlers returning to the ring, showing poor context grounding and prompting.

The hosts compared it to soap operas and telenovelas, noting how long-running story arcs challenge even human writers.

Beth and Brian agreed AI might help as a brainstorming partner, even when it gets things wrong.

Grok’s Inappropriate Conversations

Anne described a viral TikTok video of a mom discovering Grok’s explicit, offensive dialogue while her kids chatted with it in the car.

Andy pointed out Grok’s “mean-spirited” tone, reflecting the toxicity of its training data from X (formerly Twitter).

The team debated free speech vs. safety and how OpenAI’s age-gated romantic chat mode differs from Grok’s unfiltered approach.

The conversation turned to parenting, AI literacy, and the need to teach kids the difference between simulation and reality.

OpenAI’s Mental Health Stats

Andy shared that over 1 million users each week talk to ChatGPT about suicidal thoughts.

OpenAI has since brought in 170 mental health experts to improve safety responses, achieving 90% compliance in GPT-5.

Anne described how ChatGPT guided her through a mental wellness check with empathetic follow-up, calling it “gentle and effective.”

The group reflected on privacy, incognito mode misconceptions, and the blurred line between AI support and therapy.

AI in Real Estate – The “Slop Era”

Beth introduced a Wired article calling this the “AI slop era” for real estate. Tools like AutoReal can generate AI home walkthroughs from just 15 photos — often misrepresenting layouts and furniture.

Brian raised the risk of legal and ethical issues when AI staging alters real features.

Karl explained how builders already use AI to generate realistic 3D tours, blending drone footage and renders seamlessly.

The team discussed future applications like AR glasses that let buyers overlay personal décor styles or view accessibility upgrades in real time.

Anne noted that AI listing tools can easily cross ethical lines, like referencing nearby “good schools,” which can imply bias in housing markets.

Tool of the Day – Get Floor Plans

Karl demoed GetFloorPlans, which turns blueprints or sketches into 3D renders and walkthroughs for about $15 per set.

He compared it to Matterport, the industry standard for homebuilders, explaining how AI stitching now makes DIY 3D tours possible.

Beth added that AI design tools are cutting costs dramatically, reducing hours of manual video editing to minutes.

Timestamps & Topics

00:00:00 💡 Intro and show start

00:02:10 🎭 WWE’s failed AI scriptwriting

00:07:15 🤖 Grok’s explicit and toxic interactions

00:11:45 🧠 OpenAI’s mental health statistics

00:17:40 🏠 AI enters real estate’s “slop era”

00:23:10 ⚖️ Ethics, bias, and agent liability

00:27:04 💰 Microsoft & Apple top $4T market cap

00:30:10 📉 Over 1M weekly suicidal chats with ChatGPT

00:36:46 🏡 Real estate tech demo – Get Floor Plans

00:55:20 🎨 AI design, accessibility, and housing bias

00:58:33 🏁 Wrap-up and newsletter reminder

The Daily AI Show Co-Hosts: Brian Maucere, Beth Lyons, Andy Halliday, Anne Murphy, and Karl Yeh