We’re entering an era where AI systems don’t just follow prompts—they act independently. But what happens when we hand over too much control?
In this episode of Agents of Tech, our hosts Stephen Horn, Autria Godfrey and Laila Rizvi sit down with Margaret Mitchell, Chief Ethics Scientist at Hugging Face and one of the world’s leading voices on AI ethics. Together, they explore what autonomy really means in artificial intelligence—and why we can’t afford to be passive observers.
#AIEthics #ResponsibleAI #TechForGood #EthicalAI #HumanInTheLoop #AIRegulation
Topics include:
🔹 Why AI autonomy is fundamentally different from traditional automation
🔹 The real risks of removing human oversight
🔹 How trust and anthropomorphism distort our judgment
🔹 What “human in the loop” must mean going forward
This is a must-watch conversation for anyone working with AI—or impacted by it (which means all of us).
https://huggingface.co/papers/2502.02649
🎧 Listen now on your favorite podcast platform and subscribe for more deep tech insight.
0:00:00 - Intro: Are we giving AI too much control?
0:00:17 - Meet DeepSeek and Monica – the next-gen AI agents
0:00:58 - The ethics of autonomous AI
0:02:13 - Guest intro: Margaret Mitchell on AI autonomy
0:04:12 - Human control vs machine independence
0:06:25 - AI, society, and the illusion of moral reasoning
0:07:49 - Security risks from autonomous coding agents
0:11:27 - The BBC analogy & user-generated chaos
0:13:42 - Deepfakes, consent & harmful content
0:15:45 - Why AI doesn’t think like us
0:17:22 - Sensitive data, agents & social media nightmares
0:19:30 - Ease vs privacy: Why people give up control
0:21:00 - Good uses of agents: Accessibility & productivity
0:22:30 - AI and the future of creative jobs
0:24:20 - Capitalism, AGI & the wealth imbalance
0:26:10 - Margaret's message to governments
0:27:50 - Rights-based regulation vs restriction
0:29:35 - Looking 10 years ahead: Margaret’s fears & hopes
0:31:30 - Final reflections: Are we too late?
0:35:30 - Outro: Like, share & subscribe!