OS in Finance Podcast – Why AI is both over and underhyped at the same time

Video by FINOS via YouTube
OS in Finance Podcast - Why AI is both over and underhyped at the same time

Explore why AI in finance is both over and underhyped. Learn to navigate the current technological landscape with clarity.

In this conversation on the Open Source in Finance Podcast, Peter Smulovics, Viktor Mikálácz, and Bertalan Lacza from Morgan Stanley discuss the complexities of AI security, LLM capabilities, and enterprise guardrails. Viewers will gain a pragmatic understanding of when to trust AI agents with operational authority and where human judgment remains economically critical in enterprise tech.

📬 Stay on the cutting edge of open source in finance. Subscribe to the This Week at FINOS (TWAF) newsletter for weekly project updates and first access to OSFF 2027 tickets: https://hubs.ly/Q04Fmc3m0

💻 Build the future of financial services. Join the engineers solving industry-wide challenges. Explore our open-source projects and start contributing today: https://hubs.ly/Q04FmcnL0

Key Takeaways:

🛡️ Uncover why AI agent capabilities are currently advancing faster than the enterprise security guardrails and sandboxes designed to contain them.

⚙️ Understand the "jagged edge of intelligence," where AI excels in specific coding tasks but fails when applied to broader architectural logic.

📉 Learn how the economics of AI implementation and agent management are redefining the traditional pipeline for junior software engineers.

Timestamps:
0:00 – AI in Finance: Zenith SIG Podcast Intro
02:00 – Exploring the AI Hype Cycle & Enterprise Scalability
08:08 – Data Privacy: Trusting AI Agents with Enterprise Data
11:26 – Analyzing AI as a Fundamental Technology Paradigm Shift
14:50 – Dot-Com Bubble Comparisons & AI Business Models
20:56 – AI Coding Capabilities vs. Corporate Engineering Layoffs
25:50 – Transitioning from Managing Developers to Managing Agents
30:20 – The Demand for Explanatory AI & Decision Transparency
35:05 – How AI Agents Alter the Junior Software Engineer Pipeline
39:15 – AGI Timelines & Underestimating Exponential AI Impact
42:25 – The Path to AGI: Examining Transformer Model Limits
46:00 – Q&A: LLMs, Turing Tests, & Defining AGI
49:40 – Q&A: Enterprise Security, Guardrails, & AI Operational Authority
54:27 – Q&A: Valuing Human Judgment vs. AI Economics

#FINOS #MorganStanley #OpenSource

Source