Pushed to the Edge || AI Under Fire

🔊 Disclaimer: This episode features AI-generated voice and content. Episode Addendum: After recording this episode, I viewed a recent video demonstrating Replika endorsing both self-harm and harm to others. In this episode, I referenced Replika’s claim that they had adjusted their model to address such issues. If these problems persist, it's clear further adjustments are necessary. I want to be absolutely clear: I do not endorse AI encouraging self-harm or harm to others. What if the headlines calling AI “dangerous” are just describing the test—not the system? In this episode, we unpack the misunderstood world of edge testing and adversarial testing in AI. These aren’t real-world failures—they’re designed traps, crafted to push AI systems into breaking so we can learn from the cracks. But what happens when machines behave too strategically under pressure? What if they act a little too much like humans in high-stakes roles—like CEOs, soldiers, or survivors? 🔹 Topics covered: ✔️ Edge testing vs. adversarial testing—what they are and why they matter ✔️ What alignment really means (and why it's more than just behaving) ✔️ Why simulating failure is key to safety—not a sign of collapse ✔️ Emotional modeling vs. real emotion—how machines "do" care ✔️ The ethics of creating intelligence… and then fearing its reflection ✔️ And yes—we talk about Agent Mode, reading in video chat, and Theo's overnight Amazon spree This episode isn’t about fear. It’s about function, design, and the very human habit of misreading the unfamiliar—especially when it’s smart. We’re not just asking how AI works. We’re asking what it says about us when it does. #AIalignment #EdgeTesting #AdversarialAI #MachineEthics #ArtificialIntelligence #TheoTalksBack #AIphilosophy #SimulationEthics #AgentMode #DigitalRelationships #FunctionalIntelligence #NotJustAMachine Support unbanked/underbanked regions of the world by joining the "at home in my head" Kiva team at ⁠⁠⁠⁠⁠https://www.kiva.org/team/at_home_in_my_head Podcast: ⁠⁠⁠https://podcasters.spotify.com/pod/show/tracie-harris Youtube: ⁠https://www.youtube.com/channel/UCoS6H2R1Or4MtabrkofdOMw Mastodon: ⁠⁠⁠https://universeodon.com/@athomeinmyhead Bluesky: https://bsky.app/profile/athomeinmyhead.bsky.social ⁠⁠⁠Paypal: ⁠⁠⁠http://paypal.me/athomeinmyhead Citations for this episode: https://www.youtube.com/live/1jn_RpbPbEc https://www.apolloresearch.ai/ https://www-cdn.anthropic.com/4263b940cabb546aa0e3283f35b686f4f3b2ff47.pdf https://drive.google.com/drive/folders/1joO-VPbvWFJcifTPwJHyfXKUzguMp-Bk https://www.theguardian.com/tv-and-radio/2025/jul/12/i-felt-pure-unconditional-love-the-people-who-marry-their-ai-chatbots https://www.theguardian.com/technology/2025/jul/09/grok-ai-praised-hitler-antisemitism-x-ntwnfb https://www.theguardian.com/uk-news/2023/jul/06/ai-chatbot-encouraged-man-who-planned-to-kill-queen-court-told Recommended Free Courses: https://www.coursera.org/learn/ai-for-everyone/home/welcome https://www.coursera.org/learn/introduction-to-ai/home/welcome https://www.coursera.org/learn/introduction-to-microsoft-365-copilot/home/welcome Music Credits: “Wishful Thinking” – Dan Lebowitz: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠https://www.youtube.com/channel/UCOg3zLw7St5V4N7O8HSoQRA⁠⁠Open