I asked ChatGPT to simplify and sort out my investments

fizio

I have been looking at sorting out the isa/pension investments for me and my OH as they are currently across 3 providers (Vanguard - OH, interactive investor (me) and L&G (me) (ex-employer master trust in drawdown). Total approx £1m and not really used for day to day as decent DB in play.

My allocation for the last 5 years (since I took early retirement) is majority S&P 500 and Global/Developed world trackers - so lots of overlap. I am wanting to de-risk a bit and get to 80/10/10 for Dev World, EM, short-term gilts. Focus is on low cost and simple.

Fair play to ChatGPT as it run all the numbers across a variety of options from all vanguard to keeping all 3 platforms to transferring out of the master trust into interactive investor. It then did me a document of what I need to do with which account/fund across them all to get to a 80/10/10. In terms of funds it advised to stay vanguard for all 3 and keep ii for flexibility to go outside vanguard in the future. It also looked at sipp transfer Cashback offers and concluded ii was still a good bet as the platform savings would make up for the lower Cashback over time.

I am sure there are other things I can do but I am actually quite pleased with the advice and after sufficient double checking I am likely to go ahead.

dunstonh

Why did it pick Vanguard? Was that linked to your instruction?

Vanguard don't have the best tracker funds in many areas. They have one of the widest ranges but cost wise, there are often better.

Are you matching FTSE benchmarked funds or mixing up MSCI and FTSE - ideally, you shouldn't mix as Asia/emerging are different between the two benchmarks.

daveyjp

Ask the question again and see if the response changes.

artyboy

A broader question, but is it really wise to be going open kimono with one's finances with any AI tool?

JoeCrystal

https://forums.moneysavingexpert.com/discussion/comment/81893022#Comment_81893022

It is the nature of the model that you will get different answers, unless it is extremely very well short question with elementary answer, you will always have different answers since it is all about the probability of the next most likely word. The models does have it uses but I would not trust it although like a wikipedia, it could be used as the most simple way of guiding you towards some informations.

Marcon

Looking at your previous posts, you've asked a lot of basic questions - which is exactly the sort of thing this forum can assist with, provided you can sort out where the person answering knows what they are talking about, as opposed to those who merely like to think they know.

AI can and does get it horribly wrong, or more often subtly wrong - the latter being much harder for a novice to spot. Much depends on the questions you ask, and again, it's hard for a novice to phrase them in a way which gives any sort of sophisticated answer, let alone 'advice'. This thread is a case in point:

https://forums.moneysavingexpert.com/discussion/comment/81869519#Comment_81869519

If you're happy you can ask the right questions to ensure your £1m is in good hands, then for you the system has worked. Just as well, since I don't believe that the Financial Ombudsman is going to help anyone who tries to complain about poor advice they've obtained via AI!

phlebas192

Out of interest, have you ever tried asking ChatGPT about a subject where you are very knowledgable? If so, how accurate were its results?

phlebas192

https://forums.moneysavingexpert.com/discussion/comment/81893029#Comment_81893029

Wikipedia has the advantage that for all but the most obscure of subjects you are tapping into communal wisdom. Doesn't mean it is right, but at least it is not so hopelessly wrong that someone hasn't corrected the error.

fizio

I am quite happy with vanguard and also happy with the 3 fund ( dev world, em, short term bond combo) and I asked for the 80/10/10 mix to be created from my current mix - which already had some of the right products in there already. the analysis of fees and Cashback etc was a bonus. My main point was that, in the old days, I would have got to broadly the same result by spending many hours googling and on forums whereas ChatGPT does all the legwork as long as you tell it what you want and are specific, following up with fine tuning so its a to and for to get the answer that I was looking for.

I have used it to design a cctv system for a remote site as it can easily tell which camera systems suit the specific requirements (eg number plate tracking) which network provider has a strong signal in the area, which telco is doing multi-sim deal and how to position the cameras and floodlights to get the best coverage and result. It also gave me teh best prices and part numbers for Screwfix, amazon and the wiring diagrams etc so saved me a ton of time. Never done cctv before and it was all pretty flawless.

I have also used AI for a load of stuff and, after a fair mount of to and fro, it comes up with pretty good answers. Obviously you have to anonymise the data and not share personal/specific info as well as doublecheck the answers with reliable sources (such as hmrc for tax matters) but its my goto tool for most things as its just quicker and easier than without it -I am just getting it to design me a home battery storage system in a specific way so I can get some accurate quotes.

Its def not perfect by any means and all results have to be treated with a degree of caution as it does miss things but overall its a big help in getting to the result

Aylesbury_Duck

I used AI to do something similar for me, but used it to test and verify a plan I'd already drawn up. It was really helpful to challenge my thinking and was correct on almost all points. There was one element it got wrong, and the way it dealt with it was a reminder of how fallible AI can be and how you can't entirely rely on it.

One of my pensions is split across two funds, call them A and B, with A being the more risky and lifestyling moving money from A to B as I get older. I intended to stop lifestyling and rebalance back towards to A to suit a longer investment horizon than the plan was set to. The suggested mix from AI seemed counterintuitive, so I probed more and it continued to justify its suggestion after three subsequent prompts. At that point, I asked if it had mistakenly mixed up A and B and it immediately responded that yes, it had. This was despite me having referred it to the fund factsheets which clearly stated the respective risks. The mix-up was one thing, but the continued defence of its ensuing advice in the face of further questioning was interesting.

Grumpy_chap

If the OP has a £1m investment portfolio, it probably warrants obtaining professional (paid for) advice with regard to investment strategy if the OP is not able to undertake that assessment themselves.

OR, had the OP already determined the choice with Vanguard and simply using AI to validate the pre-determined outcome?

With a portfolio of the size the OP has, small differences at the input end could make large differences at the output end.