AI Training (replaced soon?)
I went to a networking session on my campus with alums and I overheard chatter about banks automating the analyst "grunt work," but I’m curious if anyone here is actually on the other side of it? i.e training the models.
I recently found a site that’s specifically looking for finance pros to train LLMs on technical banking tasks. It pays up to $150/hr, asynchronous, no minimum hours. Tbh seems very inviting as a part time gig, so I can see why people do it.
if AI is already getting trained by thousands of Pros is IB (or finance in general) even worth pursuing? I feel like as a enrty level and a student this new AI will be more knowledgeable than I am no matter how you put it. I am worried, has anyone tried doing this? Whats the level of training you are doing?
I can send the link if someone wants to look at the training link i found.
if i was in college id look into degrees related towards AI, AR / VR, space related stuff, and any future oriented field. I'm sure you can search up some other emerging industries I'm missing here.
To answer your question, yeah finance is about to be heavily automated with AI so you are 100% correct to question whether it's even a worthwhile career path
Dooming for no reason
Junior roles aren’t gonna be wiped out before this guy graduates. Banking is up or out somebody has to fill the pipeline, how it looks might evolve but not as drastic as you’re making it sound
lol banking in 5 - 10 years will look closer maybe exactly what i describe. all that will be needed in finance will be salespeople, a few customer service / RM people, and decision makers. Everything else especially execution-oriented roles will be automated by AI.
Look at how fast S&T was automated in the 2000s bro. Once banks start automating juniors it will be a very fast transition. If it will cut costs, and if there is no regulation, then no reason it won't happen.
We have to be clear on what “training” technically means. Are investment bankers currently on the desk “training” AI to do junior banking tasks? No. They are buying AI application software from ChatGPT, copilot, hebbia, rogo, modelml etc. and trying to figure out how to get leverage from them. Access is still very limited but the most forward desks still aren’t see that much value from them yet.
Companies like Mercor, micro1, scale ai etc handshake are “applied AI labs” which is Silicon Valley jargon for “we’re trying gig-ify white collar work” so they hire professional bankers at $150/hour to “train models”. What you do on the job is basically IP theft lol, you build common deliverables like a LBO, CIM etc. and then you write the rubric for the model to determine what’s correct and then score the AI model’s attempt at doing it from a prompt you generated as the “expert”.
They then take the best prompts that make the clients look the best (Anthropic, Gemini, oAI, all the frontier models) and then “benchmark” the performance across clients. Then the clients take the most successful use cases the experts have surfaced and they productize them.
I think models are smart enough to do any banking job today if you ask it in the absolute perfect way, give it the exact information it needs to do the work, and then tell it exactly what to do step by step.
Only problem is that…that’s not how real life works, especially in our industry lol.
There is almost no protectable IP in building financial models. Basically every single element of it is public knowledge, especially since most of the modeling is just doing what's required by tax law/regulators.
There is almost no protectable IP in building slide decks. Pretty much everything from CIM building to IC decks and so on are standard products.
How is uploading a client version of a financial model not protectable IP when one tab includes an entire snapshot of its CRM at a point in time?
If you’re advising a client through a carve-out, and in the presentation you detail exactly which headcount they can cut / combine down to the individual level, that is not “public knowledge”.
Your definition of “protectable IP” does not align with FINRA’s definition of MNPI. If the content that goes within a CIM isn’t private information, why do interested parties have to execute NDAs to receive the documents?
Yes your run of the mill “Apple should submit a bid to buy Warner Brothers” DCF using information entirely found on FactSet or CapIQ isn’t valuable at all, the models can do that all day. That’s not what they’re hiring “experts” to hand in though, anyone with a brain can see this is their way to get private enterprise data to train the models without having to necessarily have the institutions themselves on as clients — which would greatly restrict what information does and does not get shared anyway.
How cooked are HS seniors that want a career in finance?
How cooked were the HS seniors that wanted a career in SWE 3 years ago? Very cooked.
If not finance then what….? I always wanted to do finance for as long as I can remember…
Bump
"I went to a networking session on my campus with alums and I overheard chatter about banks automating the analyst grunt work..." (Internal quotations omitted).
They have been doing this literally for decades. They will do it for decades to come. This is how technology works. The job will change in the future just like it has in the past.
You can access those opportunities he's referring to here
One data point for the "is IB still worth it" half of your question. I work at AITraining Jobs, a site that reads these platforms' job boards every hour, and the finance projects like the one you found pay for people who already did the job, not for students.
Today (30 Oct) 242 finance roles are open across 25 platforms, typically $50 to $150/hr, median $80. The banking, PE and research slice is small, 18 roles, and most of them ask for time at a bank or a fund. That slice carries the top rates: micro1 lists an IB associate role at $120 to $150/hr and an IB expert role at $175 to $300, while AfterQuery's finance role for undergraduates and early-career people sits at $50/hr. Across all of this work, a title that names a profession advertised 40.2% more than the same kind of task without one.
So the market that trains these models pays for desk experience, not for finance knowledge in general. The $150/hr part-time gig sits at the end of the analyst years, not in place of them. Whether those analyst years shrink is the open question, and no job board answers it yet.
Has anyone here who did one of these projects been asked for deal experience in the screening, or was a finance degree enough?
Honestly, it’s probably a great tool to speed up workflows, but not a complete replacement for the role.
bump
bump
bump
Bump
AI will change the job, not replace the need for ppl who can think
A officia vero alias illo perspiciatis eligendi. Recusandae numquam voluptate excepturi qui cumque.
Alias dolor corporis dolorum non aut beatae dolorum. Voluptas sunt porro molestiae corrupti blanditiis et sunt omnis. Praesentium a ut et veniam ad et quam. Autem totam molestias qui et unde saepe modi dolore.
At minus earum voluptatem assumenda et. Et soluta nesciunt at reprehenderit. Sit vel quia accusamus in. Quisquam est eum excepturi deserunt quos soluta ut qui. Porro laudantium adipisci ea enim sit laudantium voluptas. Nesciunt vitae officia aut asperiores officia vitae dignissimos. Et ea repudiandae sunt ducimus.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Iste totam qui beatae maxime expedita eos odio. Nulla odio et ut sed nulla rerum quas. Fugit est dolorem repellendus. Quod quae et aut velit sint eos. Reiciendis quo inventore culpa perferendis tempore. A rerum et ducimus pariatur tempore aut ullam. Ipsum mollitia inventore alias.
Vitae impedit magnam quo neque tenetur. Repellat velit minus itaque. Natus architecto quo quas incidunt rem id.
Pariatur quaerat enim adipisci omnis a quia rerum. Sit reiciendis neque enim amet dolor deleniti. Cum omnis atque quaerat consequatur ex sequi autem. Sit tenetur est corrupti quos. Dolorem aut excepturi eum adipisci qui sed quia.