Goldman Sachs Chairman on AI and the Future of Finance | The a16z Show
In risk management, most of what we do isn't predicting the future—it's contingency planning. Go around the table and ask: What could happen? Don't focus on probabilities; focus on what you'll do if it happens. This mental preparation makes you incredibly alert. When things get triggered, you're off
1h 13mKey Takeaway
In risk management, most of what we do isn't predicting the future—it's contingency planning. Go around the table and ask: What could happen? Don't focus on probabilities; focus on what you'll do if it happens. This mental preparation makes you incredibly alert. When things get triggered, you're off the mark so quickly that people think you predicted it. What you really did was hear the gun go off before anyone else.
Episode Overview
Former Goldman Sachs co-president shares insights on risk management, crisis leadership, and organizational culture. He discusses the importance of contingency planning over prediction, treating mistakes with nuance, and maintaining dual roles as both risk-taker and risk-manager in high-stakes environments.
Key Insights
The Dual Role of Risk-Taking and Risk Management
Anyone investing or trading must balance two contradictory roles: aggressively taking risks to generate returns, while simultaneously managing and limiting those risks. The biggest challenge isn't getting people to manage risk—it's getting them to take more risk when they've been burned. Management must shame or encourage people into taking calculated risks while also preventing overexposure.
Distinguish Between Being Wrong and Being Stupid
Smart people are frequently wrong, but they rarely do stupid things. The critical management mistake is treating someone who made a wrong call as if they were stupid. Managers must avoid letting after-acquired information seep into their judgment of decisions made in the fog of uncertainty. Evaluate people based on what they knew at the time, not what became clear later.
The False Start Philosophy of Crisis Response
In track and field, leaving within a tenth of a second after the gun is a false start—your reaction time can't be that fast. The goal in crisis management is to be 'called for a false start' because you hear the gun so much quicker than everyone else. This comes from extensive contingency planning that makes you hyper-alert to trigger events.
Technology Adoption Requires Redundancy
In regulated industries like finance, you can't afford mistakes. You must run new systems simultaneously with proven systems, operating both until you have complete confidence in the new technology. This doubles costs initially but prevents catastrophic failures. Technology first augments costs, only later reducing them as you transition from one proven system to another.
Create Information Flow Through Approachability
Never tell someone 'I already know' when they bring you information, even if you've heard it multiple times. This prevents people from self-censoring in the future. When listening, you learn both the content and insights about the messenger. Make everyone comfortable sharing with you, even junior employees, to maintain constant information flow from all levels.
Notable Quotes
"Anybody who's investing, you're doing two things. You're trying to make money for yourselves and your clients. And so you're trying to get out there and take risk. And you're also trying to be a risk manager. And you have to do both."
"Once the present turns into the past, everybody's a genius. Most of what we do with respect to risk is not so much predicting. It's a lot of contingency planning."
"Before this technological age, not just AI, but in general, could you have a mistake that could cost billions of dollars? Um, not really. But now you can leave a a piece of software could go out and do 70,000 transactions."
"Smart people are wrong. Totally. Smart people tend not to do stupid things, but they tend to be wrong. You know, the old you know, saw about, you know, the best hitters in baseball you know, make out two-thirds of the time."
"I want everybody to be called for a false start because they hear the gun so much quicker than anybody else. They get off the mark."
Action Items
-
1
Conduct Regular Contingency Planning Sessions
Schedule meetings where you go around the table asking 'What could happen?' Ignore probabilities and focus on: What will you do if it happens? What can you do today to mitigate consequences at low cost? This mental rehearsal creates alertness that enables rapid response.
-
2
Implement the 'Never Say I Know' Rule
When people bring you information, never say 'I already know that'—even if you've heard it before. Listen fully to prevent people from self-censoring in the future. You'll learn about both the content and the messenger, while maintaining open information channels.
-
3
Run Dual Systems During Technology Transitions
When adopting new technology, continue running your proven system while simultaneously testing the new one. Only switch when you have complete confidence in the new system's reliability. Accept the doubled costs as insurance against catastrophic failure.
-
4
Separate 'Wrong' from 'Stupid' in Performance Reviews
When evaluating decisions that went badly, consciously exclude after-acquired information. Ask: Based on what they knew at the time, was this a reasonable decision? Smart people are frequently wrong—that's different from being stupid. Judge the decision-making process, not just outcomes.