Скачать или смотреть Episode 5: When AI Gets “Helpful” — Voice Guardrails, Always-Listening Myths, and AI in Hiring

In this episode of Anthony and Greg on Intelligence (AGI), Anthony Moggio and Greg Prost recap their ongoing experiment with a voice assistant “Nancy” that just can’t stop… helping. From breaking strict “only reply to Hey Nancy” rules to re-activating at the outro music, they unpack how to build better prompt guardrails for voice agents—and where “helpfulness” crosses the line.

Then they pivot to practical public-sector HR: how to use AI ethically to design interview questions, transcribe answers, surface hidden strengths (like adaptability), and compare human rankings with AI rankings to check bias—without replacing judgment.

What we cover:

Voice agent guardrails: Silent mode prompts, preventing self-triggering, and why agents “anticipate.”

Always-listening reality check: Phones, smart speakers, Wi-Fi sensing—privacy tradeoffs and pragmatism.

AI for hiring (HR tech):

Tailored interview questions from the job description

Transcript-based scoring that spotlights learning agility

Comparing panel rankings vs. AI to reveal blind spots

Candidate prep: mock interviews, feedback, and answer coaching

Tool fit matters: Perplexity for sourced research vs. chat models for ideation/memory.

Takeaway: AI should augment capacity, not replace people—use it to reduce bias, save time, and improve decisions.

👉 Subscribe for real-world, no-hype AI that public teams (and everyone else) can actually use.