Robots Make Good Junior Analysts
Also ESG originate-to-distribute, bad nickel loans, bad brokerage loans, Grayscale Filecoin Trust and Ozempic companion snacks.
One possible way to think about large language models is that they are extremely good at being mediocre at a lot of white-collar jobs. Like, if you need someone to research some question about the market for a product or the accounting treatment of a situation or the taxation of a financial instrument, you can ask ChatGPT and it will quickly, cheaply and tirelessly find you an answer that is not necessarily inspired or brilliant, but that is workmanlike and sensible and straightforward and probably right, though sometimes wrong. If you hired a first-year analyst right out of college and asked her those questions and she gave you ChatGPT’s answers, you would not necessarily think “this person is brilliant and will eventually end up running this firm,” but you would find her useful. She’d make your life easier today, even if she’s not obviously on the partner track. You can give ChatGPT a lot of work, and it will do the work for you, and that will help, though you’ll have to give it clear instructions and check its work carefully.
But what does that mean for the staffing and training of professional services firms? Like if your model is: