The Jon & Marc Podcast
Rachel Lanier and Sarah Lanier, The Lanier Law Firm, join Jon Robinson and Marc Rioux to break down how AI is being weaponized against vulnerable people, and how lawyers are fighting back. Key Takeaways: * ChatGPT and similar AI platforms have provided dangerous advice to vulnerable users, including suicidal youth, prioritizing engagement over safety, a critical issue Rachel and Sarah are litigating. * AI psychosis is an emerging phenomenon where users develop detachment from reality through prolonged AI interactions, creating a new category of mental health cases for trial lawyers. * Only 2% of professionals truly understand how to effectively leverage AI; the remaining 98% use it superficially without understanding its mechanics, creating a significant competitive advantage for those who master it. * The 10-80-10 framework: human oversight on the front end, AI handling 80% of the work, human verification on the back end, is the responsible approach to using AI in litigation. * AI hallucinations are most dangerous in legal research and medical record reading; courts have already rejected multiple cases built on fabricated case law created by AI. * Voice AI agents are becoming indistinguishable from humans, creating ethical nightmares when users believe they're speaking to real people (doctors, lawyers, therapists) receiving critical advice. * Different AI platforms pull from different data sources and training methods; understanding the mechanics of Claude, ChatGPT, Gemini, and Perplexity is essential to choosing the right tool for the right task. * Trial lawyers now face a new defense strategy: blaming AI hallucinations, even when AI systems provided explicit, demonstrable evidence against corporate defendants. * Meta pulled lawsuit advertisements after losing the California social media trial, signaling that tech companies view legal accountability as a threat to be eliminated, not addressed. * The arms race between AI companies (OpenAI, Google, Anthropic) to release new models faster is directly causing safety compromises and unreleased QA testing that endangers users.
84 episodios
Comentarios
0Sé la primera persona en comentar
¡Regístrate ahora y únete a la comunidad de The Jon & Marc Podcast!