Is it really a matter of when with AI?
First of all, I'll admit that much of this post is motivated by fear, because I want to working in banking/PE/HF & I think it's a cool career, but it's also a conversation that I've been hearing from various sources for about a year or so now: will artificial intelligence tools remove jobs from the IB industry? Lately, this has been presented as more of a "when" than an "if".
My basic understanding of a first-year analyst IB role (particularly in M&A) is that of a information gatherer/analyzer - get the CIM or public statements, build the models, value the target, build a presentation, help on roadshow, etc. In theory, the first 3-4 steps of this wouldn't be hard for something like ChatGPT to do if you feed it the proper info. It would most likely fail in several areas/build things badly in some cases, but it removes a lot of the human work, and lets an analyst take more of an oversight role.
As this idea evolves, it isn't difficult for me to imagine someone building and training an LLM specifically for banking, making it better at churning out valuations. The most extreme version of this that I've heard is an imagined firm where analysts are removed entirely from the hierarchy and replaced by AI that is supervised by associates etc. that verify and fix its work. The idea is that jobs will be removed, and/or income will be lowered, and/or undergrads will start hiring with associate roles rather than analyst roles.
I'm not as informed as I'd like on AI, tech, etc. (nor am I a banking expert by any means), so I wanted to put this forward to see if anyone more experienced/in the industry already has any thoughts.
I'd say AI may streamline tasks like data gathering and basic modeling, but it won’t replace human bankers. It just frees up juniors to concentrate on strategy, client relationships, and nuanced judgment.
That sounds to me like you need less juniors.
I don't see why AI-driven efficiency gains would accrue to the benefit of the worker, rather than to the bottom line.
Same in PE and any other research- / process-driven job.
Interesting outlook. I suppose it also depends on if juniors will be allowed to take on more responsibility like that. An evolution of the industry either way
If juniors are freed up from certain tasks people can find a whole lot of other things for them to do.
AI will only increase the need for junior bankers
the same way excel and ppt did
Curious about this take - I can see AI reducing the amount of grunt work for juniors but isn't that inherently the bulk of junior work? What's left for them to do or are we foreseeing trickling work downward from the top so that juniors are going to be handling higher seniority responsibilities that associates/VP's were previously tasked with? (And associates/VP's getting trickle down of MD/D responsibilities.)
The alternative would be that junior hours get significantly better but I'd be even more surprised to see firm leadership opt to keep the same number of juniors at lower hours required per junior vs. just hiring fewer juniors.
Agreed. AI will make research more efficient, but it would do so for every company. Just like when PPT was released, it was not like they needed less people now that these didn't have to draw slides. Work evolved. PPT just made one part of the process efficient -- and as it became more commonplace, any competitive advantage because of this efficiency was removed. Work evolved.
I think this is where we are. In short-term, jobs might reduce (if at all). But as the marketplace evolves and the profit opportunities from AI get exhausted, jobs will bounce back.
When any new innovation comes up (like industrial revolution), profits jump a lot, way more than the normal. But the profits decline and margins get lower with time until a new innovation comes and disrupts the market again.
This is completely false.
Yep
Yes, but it’s not going to happen when you’re competing for junior roles. AI has unbelievable potential, but right now is a bit of a parlor game, and people, or AI itself, hasn’t figured out how to not be moronic. The amount of poorly written slop and blatantly incorrect answers doesn’t have me worried for job loss anytime soon.
It’ll happen, but it isn’t happening tomorrow.
Bankers are the LAST people on Earth to ask about this topic.
If the labor-capital mix in an industry shifts dramatically toward being more capital-intensive and less labor-intensive, the only way you could keep the raw amount of labor input constant is if production ramps up dramatically.
Is that technically possible? Of course. But in practice, it would probably turn out more like coal mining did. Some of the guys who had been swinging pickaxes will get trained on operating the mechanized machinery. The older ones will probably just retire early. But the younger and middle aged ones will have to find something else to do.
If nothing else, I think a bunch of 40-year-old laid-off banking managers having to take their severance packages and go sign up for nursing or physical therapy classes to start a new career from scratch might do a number on the way we talk about displaced workers and work in general. People love to mock laid off factory workers and underemployed humanities majors and claim they should have made better choices in life and majored in something like business or STEM. Well, who's laughing now?
I think this is an excellent take. I believe this will cause trades to gain popularity again after having reached peak college education. At the end of the day, everything is cyclic
For everyone saying it will streamline simpler tasks and free people up to do more thoughtful, valuable tasks: how many more thinkers does the world need? Isn't the number of thinkers basically the number of people who do it today? If the world needed one more thinker (i.e. a senior banker/lawyer/consultant etc) to do the intellectually heavy work, that position would already exist today right?
I suppose we could say that the world could use more thinkers, but there's still too much grunt work tying them up. I lean against that theory because all of these white collar jobs are competitive to get into. There are many qualified folks who aren't landing those jobs, and if we needed more thinkers we'd just have more of them and hire more juniors underneath them to do the grunt work.
Hate to be bleak and would love to be convinced that I'm wrong. Haven't quite saved enough to be automated away yet.
I don't think it's that you're wrong. I think you are probably not considering that there are other systems beyond the status quo though. A system where many high paying jobs (e.g., banking, PE, legal, etc.) get cut will close the window to wealth for a lot of people. The whole monetary system is built on trust and if you remove the window to move up, it breaks down. Also, cutting a lot of high paying jobs would be massively deflationary and for many reasons the US cannot exist in a deflationary environment.
I hear that. I think the opposite may be even more unimaginable though . . imagine keeping demanding jobs needlessly in place just so they can exist as a path to upward mobility. I'm not sure they could even remain paths to upward mobility in a world where they're not needed.
What seems most likely to me is UBI or something similar under a different name.
On deflation I'd like your further thoughts or the thoughts of anyone else. Tough topic IMHO but I always thought we could balance out the deflationary force with more money printing so you get price stability. Not saying it works the other way . . it can be difficult to offset an inflationary force with monetary policy because you're taking away people's jobs when you do that. But if there's a deflationary force I think that's more feasible to offset. I think that's what we've done on a smaller scale with the deflationary impacts of internet technology.
Someone has forgotten how retarded chatgpt is – wrong thing to ever worry about
A lot of white collar jobs are just moving things around the way Chat GPT does. Even if you take the most bearish view of how useful it will become, I think a big chunk of today's post-grad jobs are replaced by it.
In my role we have a guy who's job it is to generate industry research on a specific industry we follow. He digests published reports, makes some decisions on which insights are useful and not, and generates them into his own reports that are more customized to the developments we want to know about.
At a minimum, ChatGPT today can take over his summarizing function. We could have ChatGPT read those reports, look for XYZ topical areas, and spit out his report.
More likely, ChatGPT can generate the reports we're subscribing to for him to summarize. Knock out not just his job, but he jobs of the people upstream from him.
weirdly enough today I asked a major AI for an example of companies doing [specific thing] on their earnings call- the AI came back with a lengthy answer about how Google did that in their Q4 2019 earnings call and it caused lots of confusion and speculation and such.
I go and pull the transcript- no such thing happened. I pulled the transcripts for the 3 calls before and after that one- nothing. Totally made up, even about it causing confusion and speculation. Like talking to a pathological liar.
Yeah when I first started using it, I joked with everyone that it reminds me of the dumb rich legacy kid in prep school . . just giving super confident, totally wrong answers.
Your example gives me hope.
I fear that, unfortunately for those of us who hoped to make our bones doing superior research, that it's still way too comparatively strong to not kill those jobs. Using your example: sure it just made up total nonsense, but you only lost a few minutes tracking down that mistake. On the other hand, if it correctly found an example of a company doing ___ thing on an earnings call, how much time was saved? I'd think maybe tens of hours or more, all highly compensated hours.
And then of course there's the improvements in the AI.
IMHO this is all really bad career-wise for a certain subset of people. My biggest hope at the moment is that it won't improve as much as we think. It's possible to me that in terms of the tendency to just make shit up, there will always be a reliability issue because that's just fundamental to how this tech was built. Wishful thinking probably.
Technology has already wiped out thousands of jobs in S&T. Now, AI is coming for IB. Harsh but inevitable.
What are those S&T people doing now?
doubtful. my shop has a guy who uses ChatGPT exclusively in lieu of actual reading legal doc, memoranda, and industry research. 99% of the time I cannot tie to what he saying the AI said to the actual source. He's basically synthesizing an AI summary of a complex legal document. It's never right. I don't know why people think a large LANGUAGE model is going to be able to do MATH. It's like asking Tom Brady to ice curl.
Anyways prime example. Dude had ChatGPT summarize a 100+ page energy outlook report (think EIA, ExxonMobil energy forecast) focused on combined cycle power generation. AI said the relevant pages was in pg 8, 11, and 12. I open up the actual report pg 8 was about high speed rail, pg 11 was nuclear energy, pg 12 was some other nonsense. I think genuine human comprehension and intelligence is a little better than a statistical language model that shat out words in the correct probabilistic order with no semantic association.
My concern is, even in your example how much time did the AI save by summarizing the doc? I guess your argument might be, we don't really know if it saved time because if the cites are wrong then maybe the summary is wrong too.
My experience is that it accurately summarizes books I've read and then accurately summarizes sub-topics within the book that I've asked about. I haven't asked it to find page cites but if it's accurately summarizing on a sub-topical level I'm not sure page cites will be a big thing in the future.
In my example, the time "saved" in the front-end ended up costing extra time in the back end because the person failure to actually read & comprehend the document leading to erroneous decision necessitating hours of extra work all because there was one embedded line that explained the person's question.
To your second statement. Do you plan on never confirming the veracity against the source? I literally provided an actual example of how it failed to provide accurate sourcing. The AI literally just made up information out of thin air. The naivety of source page citation being a non-issue causes me to question your decision making ability. If you trust anything AI says without primary source documentation, then you should trust a taxi driver giving you investment advice. Both are unsubstantiated and unverified sources thus should be viewed with equivalent value.
why is he still there?
Actually yes, I was playing with DeepResearch and have to admit that it's better than some interns and terrible analysts. Might even need less handholding than them. There will always be banking junior jobs, because you can bet that MDs are not going to want to even talk to Chat GPT if they can help it (hence why secretaries / "lazy girl" jobs still exist today), but the same amount of work can definitely be done with less (but smarter juniors).
That being said, AI today still seems heavily heavily context dependent. You can have a master list of prompts, but you're still going to have to hand hold it and correct it / focus it on the impt bits. All of that is context.
I think most people build that context through being forced to do said desktop research and other annoying junior legwork, plus guidance. Some people have enough common sense to get there on their own very fast, some people in fact never get there. So an incoming analyst in the future is going to have less runway to pick up those skillsets, and the bar is going to be set higher.
I rmb that when I started, alot of the less egoistic MDs would look at CVs and comment that if they'd applied to the same jobs today, they would have never gotten in. So the bar is going to be higher in terms of critical thinking skills. But the bonus for those who get in is not having to do stupid shit for 1,000 repetitions.
FWIW, I think the assertion that MSFT office increased demand for junior bankers is completely wrong. The outsized growth of finance is what increased the demand for junior bankers. We're always shortstaffed because MDs only hire when they absolutely need to.
And income is always just a matter of different industries fighting for the best talent. Doesn't take more than 3-4 years (bout the length of time for an undergrad degree) to correct itself.
Et molestias labore harum. Consequatur ipsum totam quasi dolores dolore molestias. Illo vel a rerum et.
Quasi fuga dolor quisquam voluptas porro. Rerum impedit labore sunt dolor dolorem. Neque facilis iusto voluptatum possimus voluptatem velit.
Ipsum assumenda non fugiat blanditiis. Magnam molestias nobis incidunt.
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...
Ipsum fugit aliquam eum dolorem. Esse aut quisquam est saepe corporis facilis dolore. Et nihil quas alias aspernatur est molestiae eos dolore. Veniam iusto a et facere. Quia odio nulla rerum qui. Amet rem corrupti corrupti sunt.
Quos autem pariatur officiis quam earum et quo. Ratione ab at vel nihil minima adipisci sit molestiae. Aut quam harum vel et. Vitae ab debitis esse neque. Provident autem eos quam animi quisquam est voluptatem.