The Quiet War on Junior Bankers
OpenAI didn’t announce this with a press release. No keynote. No ticker tape. Just a job posting buried in the ‘Careers’ section of their site, quietly signaling a war on Wall Street’s entry-level army.
The House of Altman — yes, that’s what insiders call it now — is hiring a senior investment banking expert to teach ChatGPT how to do the work of junior analysts. Not to assist them. Not to augment. To replace.
I’ve seen this movie before. In 2018, traders thought algo bots were just fancy calculators. Then they started replacing floor brokers. In 2021, compliance teams laughed at AI fraud detectors. Now they’re running on autopilot. This? This is the same script. Only this time, it’s not about trading spreads or flagging suspicious transfers. It’s about M&A, valuation models, due diligence packets — the very foundation of investment banking.
The job description is chilling in its precision. "Design realistic tasks and evaluations." "Create and assess high-quality reference work." "Diagnose model failures." This isn’t someone who’s going to answer questions. This is a curriculum designer. A ghostwriter for the next generation of financial AI. And they’re paying $205K plus equity to make sure it’s done right.
Why Now? Because the Bubble Is Squeezing
Let’s be real: OpenAI is in trouble.
They used to be the only name people knew in AI. Now? Anthropic filed their IPO first. Microsoft’s whispering sweet nothings to Claude. And investors? They’re starting to ask: "What’s your moat?"
The consumer side? ChatGPT Plus is fine. People use it to summarize emails and plan vacations. But that’s not where the money is. The money’s in Goldman Sachs, JPMorgan, BlackRock — firms that pay billions for deal execution. And if you can’t convince them you’re reliable enough to handle a $5B merger, you’re just another chatbot with a fancy UI.
So OpenAI’s betting everything on institutional trust. They’re not trying to build a better assistant. They’re trying to build a better banker. One that doesn’t hallucinate a valuation model. One that doesn’t misread a term sheet. One that doesn’t panic when the Fed raises rates.
Spoiler: it probably will.
The Irony Is Thick Enough to Spread on Toast
The most absurd part? They’re hiring a human to teach AI to do a job that’s already being outsourced to AI.
Junior analysts today spend 80% of their time building Excel models, formatting pitchbooks, and cross-checking footnotes. That’s not "work." It’s busywork. And it’s the perfect training ground for an LLM.
The irony? That human expert they’re hiring? They’re probably being paid to design a system that will make their own job obsolete in three years.
I’ve talked to a few people who’ve seen the training data. It’s not just financial statements. It’s email threads. Slack logs. Internal memos from analysts who got fired last quarter. The AI isn’t learning from textbooks. It’s learning from the ghosts of the people it’s replacing.
The Real Risk Isn’t AI — It’s the Illusion of Control
Wall Street doesn’t fear AI because it’s smart.
It fears AI because it’s consistent.
A junior analyst can be wrong. They can be tired. They can miss a deadline. They can cry in the bathroom after a deal falls apart.
An AI doesn’t cry. It doesn’t take a sick day. It doesn’t need a raise.
And if you train it right — if you give it enough clean, curated, human-verified data — it might just be better at this than any of us.
That’s why the real danger isn’t the model failing. It’s the moment a CFO looks at a 50-page due diligence report generated by ChatGPT and says, "Why are we still paying six people to do this?"
The job posting doesn’t say "replace." It says "improve model behavior." But we all know what that means.
This isn’t about making ChatGPT better at banter.
It’s about making it better at bank.
And if you’re a first-year analyst reading this? Start building a portfolio. Start learning Python. Start asking questions about regulatory compliance. Because your job isn’t going to be automated.
It’s already being archived.
The Human in the Loop Is the Last One Standing
Here’s the truth nobody wants to admit: no AI will ever replace a senior banker.
Why? Because senior bankers don’t just know the numbers. They know the people. They know who’s lying in the boardroom. Who’s hiding cash flow. Who’s got a personal stake in the deal.
An AI can crunch a DCF model. But it can’t read the tension in a CEO’s voice when they say, "We’re not selling. Not now."
That’s why the investment banking expert they’re hiring? They’re not training the AI to replace the analyst.
They’re training it to replace the junior analyst.
And that’s the only thing that matters.
The rest? The IPOs, the valuations, the $7 trillion in AI hype? That’s just noise.
The real story is this: OpenAI isn’t trying to build the next Google.
They’re trying to build the next Goldman Sachs.
And they’re hiring the last person who remembers how it used to work.
