EP4-6 | An Ex-Anthropic Dev on AI Control and Why He’s Building Transparent AI in Taiwan cover image

EP4-6 | An Ex-Anthropic Dev on AI Control and Why He’s Building Transparent AI in Taiwan

EP4-6 | An Ex-Anthropic Dev on AI Control and Why He’s Building Transparent AI in Taiwan

Listen for free

View show details

Season's Savings | 0.99 €/mo for 3 months

6.99 €/month thereafter - terms apply. Cancel monthly.
In this episode, we sit down with Ninad Jagdish, founder of Dwnrt Absrd and creator of SupaSim, an AI platform that models and simulates how organizations work. Ninad argues that while AI systems like ChatGPT can know a lot, that doesn’t necessarily mean they understand how a business or organization actually works.So what’s missing? Ninad believes AI needs explicit models that people can inspect, test, and challenge. We discuss what SupaSim can do differently from ChatGPT, spreadsheets, and consultants, how we know when its models are wrong, and why he argues that memory and skills alone aren’t enough for AI.We also tackle the bigger question of AI control. As researchers warn about increasingly capable AI systems, could making AI’s models more transparent help keep humans in control? And with Taiwan pushing for transparency, explainability, and human oversight in AI, could Taiwan become a place to build and test a different approach?In this episode, you'll learn:Why more data doesn’t necessarily mean AI understands a businessWhat SupaSim does differently from ChatGPT or a spreadsheetWhy Ninad believes AI needs more than memory and skillsHow transparent models could help humans question AI decisionsWhat AI safety concerns mean for SupaSim’s approachHow Taiwan fits into the push for more transparent AIWhat happens if AI eventually moves from testing decisions to making them前 Anthropic 開發者提出 AI 控制警訊：這家台灣新創如何以「透明模型」守護人類主導權？ ft. Dwnrt Absrd 創辦人 Ninad Jagdish當以大型語言模型為核心的 AI 展現驚人記憶力與產出能力時，這是否意味著系統真的理解企業運作的底層邏輯？曾任職於 Anthropic 等頂尖團隊的 Ninad Jagdish 指出，海量資料與技能庫並不能等同於具備組織運行脈絡的認知模型。如果決策過程依然深藏於黑盒子中，人類將難以驗證其真偽。為此，Ninad 在台灣創立 Dwnrt Absrd 並推出 SupaSim，打造可受檢驗、推演與挑戰的透明組織模擬架構。本集節目深入探討 AI 安全與可解釋性的關鍵命題，以及台灣如何憑藉對信任與人本監管的重視，成為新世代負責任 AI 的淬鍊基地。本集重點：資料量不等於理解力：剖析語言模型的認知盲點，為何單純的記憶與技能無法替代真實商業邏輯突破試算表與顧問框架：SupaSim 如何透過動態模擬組織運作，提供比傳統工具更精確的驗證途徑打開黑盒子的透明模型：唯有讓人類能夠檢視並質疑演算依據，才能落實人機協作的安全底線防範失控的關鍵防線：直擊 AI 安全研究者的核心焦慮，探討模型透明化如何確保人類握有最終控制權台灣的制度與信任優勢：在注重可解釋性與人為監督的政策環境下，台灣如何成為透明 AI 技術的理想試驗場📍 About Host: Paolo Joseph LisingSenior Global Community Builder, Business Next Media Corp. 巨思媒體集團#GlobalFounders #TaiwanStartup #StartupEcosystem🌏 Helping foreign founders navigate Taiwan’s startup ecosystem by bringing together founders, investors, and innovators to share practical insights and support more global startups in Taiwan.🔗 LinkedIn: https://www.linkedin.com/in/paolojoseph/📚 Learn more about Taiwan’s startup ecosystem 👉🏼 https://meetstartup.pse.is/4jhd5u🚀 Stay connected with Taiwan–Silicon Valley innovation via the Startup Island TAIWAN Silicon Valley HubWebsite: https://sv.startupisland.tw/LinkedIn: https://www.linkedin.com/company/startup-island-taiwan-sv/Facebook: https://www.facebook.com/people/Startup-Island-Taiwan-Silicon-Valley-Hub/61586221303202/ Powered by Firstory Hosting
adbl_web_anon_alc_button_suppression_t1
No reviews yet