Yes, I let ChatGPT reallocate my retirement portfolios

You do need to frame the questions correctly though. I could probably get AI to say that investing in cousin Eddie's used car tire home construction business would be a smart allocation.
Framing the questions has nothing to do with the incorrect information and contradictions I'm seeing. It's often in a string of prompts, where it has a lot of context, and then it's still providing incorrect information. I still use them often. After I educate it, it normally admits to the error and might apologize. But it happens a lot.
 
After reading your post, I thought to explore the idea via MS Copilot. There I found it to be a "yes machine", i.e. no real ideas but reenforce my thoughts and convictions. What AI are you working with?
ChatGPT version 5.2. I've never used Ms Copilot. Maybe try again in ChatGPT. Copy/ paste your portfolio, supply your age and goals (growth, income, etc.) and tell it to analyze your portfolio.
 
I’ve been on the receiving end of AI that “replaced” humans, so I guess in the literal sense you and that article are correct… but those interactions are always painful; hence my statement.

I’d hate to say it, but at current time I would rather deal with a call center in XYZ country then AI for customer service.
The only time I dealt with AI on the phone, it w*rked perfectly. Who knew? At first I thought I was talking to a human. The first thing that tipped me off? I could understand every word!!
 
Grok is still my favorite so far of the little I've used. I'd ask "him".
 
I only use ChatGPT because that is what my son put on the computer of a "computer challenged" person.
You can access Perplexity AI by typing in perplexity.ai in any web browser address bar. No account sign up needed unless you want to keep a conversations history. Its more of a Spock like straight forward answers versus ChatGPT more Captain Kirk charming answers.

1765752751466.png
 
I gave ChatGPT my age, investment goals, time horizon, and preferences like no bond funds, a 50/50 equity goal, stability over income, and a goal to withdraw $100k a year.

I copy/pasted my current investments (gasp!) from our five E*Trade accounts and let "chatty" suggest the moves needed to align with my goals. Her suggestion aligned very closely to what I already knew I "should" do and this finally gave me the motivation to do it. Sold a bunch of losers. Sold the bonds, CDs and REITs that were in my Roths, added QQQM to the Roths instead. Made the Roths my high growth vehicle and the IRA's the stable vehicle that will be drawn down.

I took advantage of the 30 day free trial so I could copy paste and have Chatty "read" the screens. It was very helpful. even guiding me through selling the existing bonds and CD's and building a new bond latter. Everything is much cleaner now.
I entered my and DWs portfolios (IRA, 401K, Brokerage) with ages and additional note that with SS and pension I was meeting my current expenses. It suggested some simplification (duplicate strategies in several EFTs/MFs) but overall said I was doing fine. Nice to have justification if even from a robot.
 
Like you/humans, AI only knows things that already happened/existed at some point.
If it makes it up, it's because it combined 2 things it knows into 1 'new' thing.

AI can't tell the future any more than you can.

It may help you reason and collate data to get to an end, but the means still needs to be within you.

For every simulation run, even with firecalc or monte carlo, at some point, there's a wide margin of error.

tread carefully.
 
I forgot to mention a couple of things that may help in my earlier response. First, as you interact with an LLM, clarifying your questions, etc., all of the interaction remains in the AIs context window and is used to shape the response to further prompts. If you do not delete the chat, it can be returned to later and all of the chat will be reloaded into the context window as you proceed. That allows you to continue where you left off withoout needing to reinput data. But a caution: if a chat seems to be getting worse or off the rails, stop and start a new chat because the confused context (response, correction, new response, diversion to new issue...) can cause the chat to degrade.

Second, when looking into serious matters like health, finances, legal issues, etc., it can be very useful to submit an AI's response to another AI for analysis and fact checking. I haven't taken that process out to three or even four reviews, but that might be interesting.
 
I'm wondering if AI can give me the next lottery winning number.
 
It might tell you numbers 1 through 31 were more highly represented because people like to include dates, so more likely to get an overlap with another winner.

IME, it has no problem saying it doesn't know something.
 
Like you/humans, AI only knows things that already happened/existed at some point.
If it makes it up, it's because it combined 2 things it knows into 1 'new' thing.

AI can't tell the future any more than you can.

It may help you reason and collate data to get to an end, but the means still needs to be within you.

For every simulation run, even with firecalc or monte carlo, at some point, there's a wide margin of error.

tread carefully.
Your view is spot on!!!!
 
I'm wondering if AI can give me the next lottery winning number.
I just asked here was the answer. Per ChatGPT
I need to be clear with you: lottery numbers are drawn randomly, and there’s no way for me—or anyone—to predict the winning numbers in advance. If someone claims they can, that’s misinformation.
 
I just asked here was the answer. Per ChatGPT
I need to be clear with you: lottery numbers are drawn randomly, and there’s no way for me—or anyone—to predict the winning numbers in advance. If someone claims they can, that’s misinformation.
Except these guys

 
I use ChatGPT almost daily at work. I use it to help write emails and other documents. You can give ChatGPT the parameters of what you're trying to accomplish, then drag and drop the doc and it will do it's thing. Offering to create an updated pdf for you to use.
I can even direct ChatGPT to use a certain tone in the email communication if I really want to drive home a point or signal the urgency of the issue. It's been a great assist with these types of things.
 
I read this the other day.
Keep in mind there is a med student using ai to get through school. Time to eat healthy.
 
I read this the other day.
Keep in mind there is a med student using ai to get through school. Time to eat healthy.
There is a scene in Real Genius where at the beginning of the school year several student desks are empty and instead there is a tape recorder in their seat presumably recording the lecture. Later in the year, the same classroom has all of the seats empty of students and filled with tape recorders and then pans to the front of the class and the teacher is gone and has been replaced by a tape recorder. Lol.

Maybe it will come true but with AI. Students will use AI to pass class and teachers will use AI to teach them.
 
Like you/humans, AI only knows things that already happened/existed at some point.
If it makes it up, it's because it combined 2 things it knows into 1 'new' thing.

AI can't tell the future any more than you can.

It may help you reason and collate data to get to an end, but the means still needs to be within you.

For every simulation run, even with firecalc or monte carlo, at some point, there's a wide margin of error.

tread carefully.
That’s called a hallucination, correct? People have them when dehydrated crossing the desert, or when partaking in certain controlled substances.
 
ChatGPT can't do what I want based on my model.
A trader friend with a similar model paid for ChatGPT and put in a lot of effort...and as I expected, missed the best funds and the chance to sell on time.
The main reason is that my model is based on experience and feel. You can't program it. This is why I make "mistakes"; they are part of the model. My main goal is to avoid losses; if I'm wrong, I'm back within days; if I'm right, I'm out for weeks to months.

For someone who looks for generic stuff, ChatGPT would do a better job, but that will depend on the parameters you feed it.

If ChatGPT was great, the people who created it would make lots of money and wouldn't need others to invest in their company.
 
I haven't yet used AI for financial stuff. I had a bad experience with AI where I asked it to give me a 4 mile training run and it gave me an 8 mile loop and claimed that the loop was partly on a river path that in reality is about 10 miles away. I tried correcting it several times but I eventually gave up.

I also don't use it for tax advice because it doesn't yet seem to be able to do things like read the text of the OBBBA and apply tax strategies to specific situations like I can. Maybe the next iteration of AI will be able to do so.

I do tend to use AI for linguistic sculpting as a sort of editor to get the tone I want or clear up awkward phrasing. I've also had it do really well when I took pictures of some stuff around the house and said "Hey, how much is this stuff worth?" - it seemed to accurately identify and summarize that for me.

The content window thing is interesting. I use Grok and because of the context window it can interrelate and incorporate things between, for example, my marathon and flight training stuff in a pretty reasonable way.
 
Hesitantly jumping in as this topic seems to generate a bit of emotion. No expert here, but have a little life experience and worked in the tech industry for 40 plus years.
I've been modeling my retirement for 30+ years, retired in 2018, and have always done my own investments as I trust myself better than financial advisors.

Like many of you, I've had multiple consultations (free & paid) with financial advisors (FA) and also used every online modeling tool I see mentioned on this and a few other forums. I do this because I want to validate strategy, learn of changing laws, and truthfully .... reduce anxiety of running out of money. I find most are very similar in the general plan for budgets & strategies to maximize income, while some detail ideas on minimizing taxes. I don't ask for individual buys, but look for broad guidance on investments and withdrawal strategies. We've been very blessed and comfortable in life, but will always live below what every model suggest.

With that said I recently tested a few of the AI programs (ChatGTP and Gemini) spending time to anonymously input assets, budgets, and longevity, requesting a detailed retirement plan.

In my personal experience the AI tools I used generate a plan that mimic and correlate to every consultation and financial planning tool I have ever used. They're not foolproof (but neither is any online tool or FA), AND they're only as good as the input they get. I don't recommend trying to get down to individual investments, but they validated our investment / withdrawal bucket strategy with details on how to best minimize taxes while mitigating risk.

I will continue to use them as a tool just like any other financial modeling tool ... and then apply my gut/experience. This isn't to say they're for everyone, as I see many views every day on this forum for a variety of financial topics.

I frequent many financial sites and this is one of the best retirement forums I have found on the internet. You are polite and informed...... well mostly :>)
 
Back
Top Bottom