JPMorgan Developing ChatGPT-Like AI Service For Investment Advice (cnbc.com) 25
JPMorgan Chase has applied to trademark a product called IndexGPT, indicating its development of a ChatGPT-like software service that uses artificial intelligence (AI) to select investments for customers. CNBC reports: IndexGPT will tap "cloud computing software using artificial intelligence" for "analyzing and selecting securities tailored to customer needs," according to the filing. [...] But JPMorgan may be the first financial incumbent aiming to release a GPT-like product directly to its customers, according to Washington D.C.-based trademark attorney Josh Gerben.
"This is a real indication they might have a potential product to launch in the near future," Gerben said. "Companies like JPMorgan don't just file trademarks for the fun of it," he said. The filing includes "a sworn statement from a corporate officer essentially saying, 'Yes, we plan on using this trademark.'" JPMorgan must launch IndexGPT within about three years of approval to secure the trademark, according to the lawyer. Trademarks typically take nearly a year to be approved, thanks to backlogs at the U.S. Patent and Trademark Office, he said. The applications are typically vaguely written to give companies the broadest possible protections, Gerben said.
But JPMorgan's filing does specify that IndexGPT uses the same flavor of A.I. popularized by ChatGPT; the bank plans to use A.I. powered by "Generative Pre-trained Transformer (GPT) models." "It's an A.I. program to select financial securities," Gerben said. "This sounds to me like they're trying to put my financial advisor out of business."
"This is a real indication they might have a potential product to launch in the near future," Gerben said. "Companies like JPMorgan don't just file trademarks for the fun of it," he said. The filing includes "a sworn statement from a corporate officer essentially saying, 'Yes, we plan on using this trademark.'" JPMorgan must launch IndexGPT within about three years of approval to secure the trademark, according to the lawyer. Trademarks typically take nearly a year to be approved, thanks to backlogs at the U.S. Patent and Trademark Office, he said. The applications are typically vaguely written to give companies the broadest possible protections, Gerben said.
But JPMorgan's filing does specify that IndexGPT uses the same flavor of A.I. popularized by ChatGPT; the bank plans to use A.I. powered by "Generative Pre-trained Transformer (GPT) models." "It's an A.I. program to select financial securities," Gerben said. "This sounds to me like they're trying to put my financial advisor out of business."
You can have my monkey and dart board (Score:4, Funny)
Re: (Score:2)
You're better off with the monkey.
Some years back, to prove a point, I knocked together a model to predict the "daily number" in a state lottery. IIRC, the number was drawn twice a day. I had 20 or 30 years or so worth of past data, I honestly don't remember. Anyhow, given the previous 5 numbers, the model was able to predict the results for any day in the past with frightening accuracy.
To the greedy and stupid, this sounds amazing. If you're laughing right now, you've probably already realized that JPM
Re: (Score:2)
Hahahaha, excellent. Well, those that think that of course money must make profit are easy prey because they have already demonstrated they have no clue about the real world.
Re: (Score:2)
You can laugh at them, but if they can sell the use of their application, it will turn them a profit even if it doesn't work.
Re: (Score:2)
If packaged properly, you could probably sell them a "past stock price predictor" AI.
Re: (Score:2)
You jest, but I think also hint and why a LLM should easily give out accurate investment advice to nearly anyone. You can pretty much do it with a simple decision tree, so an LLM should have no problem. Just ask their age, and have them go 100% into low fee index funds (with a standard large/mid/small cap & international blend) until they are around 50-55, and then start rebalancing a bit into bonds at that point. How hard could that be for ChatGPT?
Al Czervik (Score:2)
Oh holy shit (Score:2)
Jesus Forkin' Christ, they're going to have AI driving their investments?
What am I worrying about, nothing could possibly go wrong!
Re: (Score:2)
Re:Oh holy shit (Score:4, Funny)
Don't worry, the model was trained on many thousands of thoughtful Reddit discussions related to investing.
Re: (Score:2)
You jest, but training model on actual investment decisions and hindsight of how they turned out would be a way to go.
Problem is that this software would struggle to identify real world issues for data input to analyze. It would be limited to what can be found in internal databases + online. I.e. this sort of LLM based approach would be severely constrained by what bank's internal data analysts can input into it.
Re: (Score:2)
Actually no. There was some research a while ago that found that in investment, the past is not a predictor for the future. Hence anything "trained" is useless.
Re: (Score:2)
And there has been other research that found it *is* a predictor of the future. What you find depends on what model you are using and how your filters are set. Perhaps there really is a patter than an AI could pick out. They're known for picking out answers based on pieces of the info that people just ignored. (And, of course, for being quite fragile when beyond their training dataset.)
I don't think it's really safe to bet either way in principle. But I suspect that JPMorgan is planning on selling acce
Re: (Score:2)
Not really. Or rather only if you ignore that the big investment companies are using exactly that and hence they pretty much eliminate the effect for everybody else.
I do agree that this is just something to be sold to the stupid. There are always plenty of stupid people with money the can "invest". And if they could not afford to lose it, let society pay for the damage.
Re: (Score:2)
Re: (Score:2)
I'd be willing to bet a garjillion dollars that
1) the whole mess suddenly crashes and burns, and
2) a literal ton of the investor's money disappears into thin air, and
3) no one will ever spend a minute in jail because of it, and
4) version 2.0 launches under a new name as soon as they can round up more suckers.
Re: (Score:2)
Jesus Forkin' Christ, they're going to have AI driving their investments?
What am I worrying about, nothing could possibly go wrong!
No, you got it wrong. They're going to use a chatbot to tell other people what to invest in, then watch it closely to see if something semi-worthwhile is in the stream and use it to drive their overall profit. This pretty much epitomizes the self-fulfilling prophecy desires of the big investment firms, if they can just get enough suckers lined up to listen to the chatbots.
Re: (Score:2)
Jesus Forkin' Christ, they're going to have AI driving their investments?
No, they're going to have AI driving other people's investments. You know, the less-sophisticated lower-end investors whom they can lose without much grief.
You can be sure that their own money, and that of their high-value clients, won't be invested according to the advice of a new ChatGPT flavour. At least not until after the rubes finish the shakedown cruise and prove the concept by having consistently good gains. Only then will the high-rollers start buying in.
They need ChatGPT (Score:2)
Jeffrey Epstein is not around any more to help JPMorgan with investment advice.
The new Shiny! (Score:1)
Anyone wanting to use an AI to replace some analysis a human is doing it to save money on human resources and make money from the customer faster.
They'll both come to the same bad advice so why pay a human to do it?
Re: (Score:2)
And that is very much it.
Holy shit this will be amazing (Score:1)
Whoever can weaponise the robots to write the most bullshit to pump or dump their target stocks will make a fortune. HFTs will have to integrate LLMs natively to attempt to decipher all the bullshit. The stock market will become a total random number generator.
I mean, if the whole thing didn't fall over at the first hurdle, which seems like the most likely outcome.
Re: (Score:1)
Sounds fine to me. I invest based on fundamentals - calculating a value I think the company is worth based on PE ratio, net tangible assets per share and dividend yield compared to govt bond price, or a combination of these, so the and and the more noise, the better chance I can pick up some underpriced stocks or sell some over - priced stocks.
Sometimes I have been suckered by IPO hype on one stock or another, and invariably been burnt, but when I an disciplined and pay attention to the fundamentals (using
simple to code (Score:2)
follow index,
profit
Yeah, um, nah, I'll pass (Score:2)
See, this is another example of why most investment advice is bullsh*t. Oh, you're a certified financial advisor meaning you give advice that's "in my best interest"? What if you decide that being independently wealthy isn't in my best interest because you think I'll get into politics or activism instead of working my ass off to fund government tax coffers until I die? Or how about this: "Or fees are structured so we do better when you do better"? Yeah, um, what if you don't care if I do better because