GPT 5 out, quite underwhelming
I’ve seen a lot of posts recently about people worried that AI will take entry level jobs in finance especially IB where you’re a model monkey, and while we absolutely do not know what the future will hold, it’s developments like these that give me some faith that society won’t collapse due to full automation anytime too soon.
GPT 5 seems like an improved GPT 4 but nothing really new honestly so while maybe finance in the next 3-5 years will change, I’m not holding my breath that suddenly analyst jobs will get wiped off the board
Curious to hear your thoughts though because this field is constantly evolving of course
GROK MASTER RACE
Based on the most helpful WSO content, the sentiment around AI's impact on entry-level finance roles, particularly in investment banking, is mixed but leans toward cautious optimism. While advancements like GPT-5 and beyond are impressive, they don't yet signal the immediate collapse of entry-level roles. Here's why:
Incremental Improvements, Not Revolutionary Changes: Many users have noted that while each new iteration of AI (e.g., GPT-5) brings enhancements, these are often evolutionary rather than revolutionary. This means that while AI tools can assist with tasks like modeling, research, and data analysis, they aren't yet capable of fully replacing the nuanced, relationship-driven, and strategic aspects of finance roles.
AI as a Tool, Not a Replacement (Yet): AI is increasingly seen as a productivity enhancer rather than a direct replacement for human workers in the short term. For example, tools like GPT-4 and GPT-5 can streamline repetitive tasks, but they still require human oversight, especially in high-stakes environments like investment banking.
Relationship-Driven Work Remains Safe: Senior roles that rely heavily on relationship-building, such as MDs in investment banking, are considered less vulnerable to automation. Even at the junior level, tasks that involve client interaction or strategic thinking are harder to automate.
Cultural and Operational Lag: As highlighted in WSO discussions, the adoption of AI tools often lags due to cultural and operational inertia. Many professionals in finance are slow to adapt to new technologies, which could delay the widespread impact of AI on the job market.
Long-Term Uncertainty: While the immediate future may not see a complete overhaul of entry-level roles, the long-term trajectory of AI remains uncertain. Some users predict that as AI continues to improve, it could eventually replace a significant portion of junior-level work, especially as tools become more sophisticated and capable of mimicking human decision-making.
In summary, while GPT-5 and similar advancements are impressive, they don't yet pose an existential threat to entry-level finance roles. However, the industry is evolving, and professionals should stay adaptable, learning to leverage AI tools to enhance their productivity and value.
Sources: Thoughts on AI and the future of the job market?, GPT-4 & Microsoft Co-Pilot, Work at top AM -- AI will virtually kill of entry-level roles in AM over the next 5yrs, Google partners with Goldman Sachs in automating Investment Banking, https://www.wallstreetoasis.com/forum/investment-banking/goldmans-new-policy-for-investment-banking-analysts-3-reasons-why-it-wont?customgpt=1
IMO the gpt5 hate is overblown, yes they had a godawful presentation and hyped it way too hard but it's pretty clear that was in service of the 500bn valuation that just dropped, in retrospect
but I think GPT-5's improvements in hallucination are more valuable than improvements in intelligence to the average consumer. I bet 90%+ of office jobs don't require an advanced knowledge of competition level math or emerging law and are instead a series of little tasks that can easily be done by AI already - and once AI can do it 100% accurate or even in the ballpark of how accurate humans do it we're all fucked
That’s a good point. Once hallucinations are removed then a ton of jobs will be in danger. I hope we don’t all get screwed that way
you can't remove "hallucinations" everything LLMs output is a "hallucination". it's just probabilistic strings of text being produced, which will never be fully accurate as they are probabilistic and not deterministic
I studied Math in college and part of that was taking a few ML courses and getting involved in research. While I can say the field progresses extremely fast, a lot of the current papers I'm reading on ML has been suprisingly underwhelming in the field of generative Ai. This is primarily because everyone is trying to push transformer technology(the key part of chatbots like gpt) to their absolute limit, and while sometimes this can make big leaps, for the most part at this point we've reached the "natural" limit of transformers. While many continue to make good improvements(reasoning models and SSMs are both promising) right now we need to see a breakthrough in something like reinforcement learning to really push things forwards.
Ai and LLMs are still impressive, but they're also incredibly expensive due to the quadratic nature of input data(meaning every input you give an LLM scales it by n^4 computational needs), meaning we are going to need an INCREDIBLE amount of compute or(more realistically) a change in models over to SSMs which scale linearly before we actually see profitable use of commercial AI.
Every LLM has been underwhelming. When was the last time a new technology truly changed your life? It’s been a minute.
ChatGPT has already changed my life
Are you one of those people who think that you’re best friends with a chatbot?
Good. If improvements become more incremental, then it’s beneficial vs cannibalistic to our jobs
I don't get the hate. Even before GPT-5, ChatGPT was already a magical, borderline alien-like technology. If you’re underwhelmed by GPT-5, what more do you want at this point?
This thing does a majority of my job now, plus a ton outside of my 9-to-5. Imagine if ChatGPT and all other LLM models were switched off, how much harder and more inconvenient would your life be?
Humans tend to adapt to new technologies so quickly that even the most remarkable inventions become the "new normal". You could probably bring an automobile to 3000 BC Mesopotamia, and within a week, you'd see people complaining about flat tires. Probably a great reason we are so innovative, it's never enough.
I still don't use any ChatGPT or LLM in my daily life so it wouldn't get any more inconvenient at all.
Is this a bit? My life would be exactly the same. Arguably better with a lack of the idiocy flooding social media. Maybe Google results would be reliable again and smartphone companies would actually try and innovate instead of letting us create custom emojis or give inaccurate summaries of emails and telling us this is the future. What a crock of shit.
LLMs are generative chatbots which are routinely incorrect and are designed to gas you up. What more do I want at this point? Maybe actual artificial intelligence. Maybe instead of untalented morons thinking that AI writing and art and music, all of which are based on stolen IP, are the next step in creative expression we can actually get artificial intelligence and robotics to do the tasks we don’t want to do and let us all have better lives as a result. Why the fuck would I want a computer to do the things I actually do want to do instead of the things that I don’t?
If you not only need ChatGPT to write an email but are excited about that as a possibility, you are the type to get excited about shadow puppets dancing along the cave wall. Technology was so promising once. Now it is so utterly stupid.
It seems how useful someone finds ChatGPT is highly dependent on the nature of their work. My job is somewhat technical and I've found it amazing for writing complex SQL queries in seconds which previously took 30+ mins.
Outside of work, I've found it great for condensed tutorials on somewhat new or unfamiliar tasks so I don't have to spend hours scrolling through YouTube videos and online guides. It's been a huge time saver.
Genuinely curious for those in the know - how exactly is AI going to generate new ideas, creative forms of expression, etc.? The only thing I've used AI for are the following:
Those are definitely helpful in my workflow and a moderate productivity boost, but I'm really struggling to see where people are automating their entire day through ChatGPT or some of the other claims that are made. From everything I've seen and read, all these LLMs are good for is synthesizing and understanding data that has already been provided. That said, there is (or at least should be) a lot more critical thinking involved in evaluating what info it gives you and evaluating whether it's accurate, the best path to pursue, etc.
Additionally, a lot of my, and I'd assume most people on here's, job is moreso about telling a story through a financial model and creating assumptions that would pass multiple stakeholders' smell test. Maybe AI can help create those assumptions, stories, etc., but is it really going to be capable of understanding different peoples' incentives and working out a solution based on that? It seems like all these models are trained on the same data. So if the data is crap or the assumptions used in training the model are crap, then the AI output will be crap too, no?
Again, I could be completely wrong about all this, but so far I see a lot more hype around AI by tech bros trying to pump up their startups' valuations than I see true commercial application. If I am wrong, please do explain to me how/why so I can bet the farm on $QQQ and make sure my miniscule stake in all these public tech companies will supplement the meager UBI the government supplies us when we all truly do lose all our jobs in 5 years.
Of course it is. Generative AI (and the tech sector in general recently) relies on massive hype to justify truly absurd funding, otherwise it's a total dead end.
AI isn't going to replace jobs at scale any time soon. It will be super useful in aiding humans in doing their job, but at the end of the day AI isn't actually intelligent - as someone else said, everything that comes out of AI is a hallucination. The problem is that as social and especially legal awareness catches up, companies are going to start being held responsible for that output, and that'll stifle growth.
Instead of taking the Elon Musks and Sam Altmans of the world at their word, we should look at what AI actually does. And all it really does is summarize and find patterns. Those are valuable outputs to be sure, but not revolutionary. AI may have an impact on the level of the internet or the smartphone - it'll change the way we interface with the world and each other, but it isn't going to upend the human condition. Both of those innovations were culturally massive shifts, and did lead to the obsolesence of some jobs... but also created new ones. Travel agents and print journalists may be shit out of luck because of the internet, but bloggers and website designers have more opportunity.
Non fugit doloribus quidem deserunt unde ex. Perspiciatis est magni dolorem dolorem numquam. Ex earum autem voluptas quam cupiditate error quis.
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...