Everything Goldman Sachs Taught Me About AI (In 10 minutes)

Everything Goldman Sachs Taught Me About AI (In 10 minutes)

by Nate Herk

Applied Sciences & Engineering Economics & Finance KEconomics, Finance, Business and ManagementKJBusiness and Management
🎙 Nate Herk 👥 964K 📅 August 24, 2026 ⏱ 10 min 👁 11K 📄 expert opinion 🧭 2026-08-28
Available in: English (current) Français

Keywords

AI principlesverify outputhuman in the looptransparencyaugment not replace

Summary

The video presents five AI principles derived from the author’s experience at Goldman Sachs, summarized by the acronym VAULT: Verify, Augment, Understand the why, Loop humans in, and Transparency. The author emphasizes the importance of verifying AI outputs, especially in high-stakes environments, and building verification into processes. He advises augmenting existing systems with AI rather than replacing them, using deterministic automation for rule-based tasks and AI for tasks requiring judgment. Understanding the purpose of an AI system is crucial, as is starting with the problem rather than the technology. Human oversight is essential, with approval steps scaled to the consequence of actions, to prevent AI agents from causing significant errors. Finally, transparency in AI systems, including logging and documentation, is necessary for trust, maintenance, and handoff. The video includes a personal anecdote about an AI agent sending a discount code to 200,000 people, illustrating the risks of autonomous agents. The author also promotes his community and resources for learning AI automation.

161 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides practical, actionable advice on AI implementation, drawing from the author’s experience at a major financial institution. The VAULT framework is clearly explained and illustrated with examples, making it accessible to a broad audience. The argumentation is coherent, emphasizing risk management and human oversight, which are critical in AI deployment. However, the value is limited by the lack of empirical evidence or case studies beyond the author’s own experience. The advice is general and aligns with common best practices, but it does not offer novel insights or deep technical analysis.

Scientific Rigor, Source Quality, Title Accuracy

The video does not cite any external sources or references, relying solely on the author’s personal experience and anecdotes. The title accurately reflects the content, which is a summary of AI principles from a Goldman Sachs perspective. The lack of citations reduces the scientific rigor, though the advice is consistent with industry standards. The video includes promotional segments for the author’s community and tools, which are clearly marked but do not detract from the core content.

183 words

Title / Content Match

The title accurately reflects the content, which distills AI principles learned from working at Goldman Sachs into a 10-minute video.

Quality & Reliability

6/10

The video presents practical AI principles based on the author's experience at Goldman Sachs, but relies heavily on anecdotal evidence and lacks verifiable sources or citations. The advice is sound and aligns with industry best practices, but the lack of external references and the promotional nature of the content lower its scientific reliability.

Chapters

Cited Sources

Concurring Sources

  • Marco Argenti on AI and human oversight — The video references Marco Argenti, Goldman's CIO, but no direct source is provided.

Contribution & Novelties

The video offers a practical framework (VAULT) for AI implementation, based on the author’s experience at Goldman Sachs. While the principles are not new, the packaging and emphasis on risk management in high-stakes environments provide a useful perspective for businesses. The anecdote about an AI agent error illustrates the importance of human oversight.

Pour aller plus loin :

  • Human-in-the-loop — Relevant to the principle of looping humans in for oversight.
  • AI alignment — Discusses ensuring AI systems act in accordance with human intentions, related to transparency and verification.
  • Explainable AI — Relevant to the transparency principle, focusing on making AI decisions understandable.

102 words

Radar Profile

The radar profile shows moderate scores across all dimensions, indicating a balanced but not exceptional video. The content is practical and relevant, but lacks depth and scientific rigor, resulting in a moderate overall quality.

Reliability 5/10