Conflicting answers between 2 AIs on interest gained from monthly vs at maturity

[alleylehmann]

Hi all,

I'm getting conflicting results between ChatGPT and Gemini when I ask the same question. I was wondering if anyone can help clarify which is the correct answer?

The question asked was if i have $250,000 in my savings account, and i have a choice between 6.15% interest compounding out monthly, or 6.30% paid at maturity, which is better for the duration of 180 days?

ChatGPT said the better option is the compounding monthly interest. The calculation was based on the forumula for compound interest:

A = P(1 + r/n)^(nt)

Where:

A = the future value of the investment/loan, including interest

P = the principal investment amount (the initial deposit or loan amount)

r = the annual interest rate (decimal)

n = the number of times that interest is compounded per unit t

t = the time the money is invested for, in years

For this case:

P = $250,000

r = 6.15% = 0.0615

n = 12 (compounded monthly)

t = 180/365 (180 days out of 365 days in a year)

So, A1 = 250000(1 + 0.0615/12)^(12*(180/365))

Compared to the 6.30% paid at maturity,

you're essentially investing $250,000 at 6.30% for 180 days.

A2 = P(1 + rt)

= 250000(1 + 0.063 * 180/365)

The results:

For 6.15% monthly: A1 = 250000(1 + 0.0615/12)^(12*(180/365)) ≈ $254,169.29

For 6.30% at maturity: A2 = 250000(1 + 0.063 * 180/365) ≈ $254,027.40

However, with Gemini, it's the other way around.

Monthly compounding:

• Principal amount (P) = $250,000
• Interest rate (R) = 6.15% per year (convert to monthly for compounding: 6.15% / 12 = 0.5125% per month)
• Time (T) = 180 days / 365 days per year = 0.5 years
• Amount (A) = P * (1 + r/n) ^ (n*T)
  • n = number of compounding periods per year (in this case, monthly = 12)
• A = $250,000 * (1 + 0.5125% / 12) ^ (12 * 0.5)
• A ≈ $257,678.38
  • Simple interest:
• A = P * (1 + RT)
• A = $250,000 * (1 + 6.30% * 0.5)
• A = $257,767.12

So ChatGPT says I'm $141.89 better off with monthly interest, while Gemini says I'm $88.74 better off with interest at maturity..

The funny thing is, if I take ChatGPT's answer and paste it into Gemini, asking Gemini to confirm which is the right one, it or GPT, it corrects itself and concludes that it was wrong, and GPT was correct.

But.. if I take Gemini's original answer, and paste that into GPT asking to confirm which is right, it then concludes that it was wrong, and Gemini was correct...

So both think they're right, then both think they were wrong and the other was right..

Thanks.

 

[4himnow]

@alleylehmann The time it took you to write this post could have been spent calculating it yourself. Don't waste your life trying to make sense of Chatgpt or any of the others. Even the creators don't know how they work.

You've illustrated the exact problem with these models, that the time taken to validate the results is longer than the time it takes to get the result. This only becomes useful to us when we can trust the results. These models have proven time and time again that they can't be trusted to the extent that validation is redundant.

Never use them for calculations only for language, because that's what they do. Now Chatgpt is trying to be smart by writing the python to solve the problem and running it, and sometimes you don't even see the script. So it can easily choose a function that isn't appropriate for the task, or is implemented badly.

In short, don't believe the hype, go outside and look at a tree or something useful.

 

[resjudicata]

@4himnow Yep I’ve seen ChatGPT give me wrong answers for something that should have been basic

 

[mpuhcrs]

@resjudicata What too many people fail to realise is that chatgpt is just putting words together that seem correct while sounding confident.

It's a confidence scam.

It will offer you a product that 'sounds' correct, but then it will say that the product should perform a certain task that it is incapable of. Because the product and the task sounds correct.

 

[alleylehmann]

@4himnow I should've mentioned that I'm not good at these calculations, and struggle with them, hence why I gave the two AI a shot. I don't mind doing it myself, but it's too complex for me unfortunately, which is why I was asking for help here.