Using AI to invest

If good investing is predicting the future, then I struggle to see AI doing much better than a reasonably competent human investor. What is more, if the value of shares is partly a result of human emotion, how can AI reliably factor in a variable that is completely alien to it, that is, by definition, unpredictable?

If so, maybe the effect of AI on investing will be to enhance the value of emotions-possessing humans making investment decisions - ???

There also already seems to be a trust problem with AI eg Musk has said he wants to create a truthful version, what with current versions already spouting out pc gibberish (a problem affecting far from only AI).

Whatever, it is early days, certainly AI needs a lot of kinks ironed out of it first before I start taking advice on spending my money from sth that lacks emotion and, in some areas, is clearly putting pc before truth (& simple truths at that).

More broadly, being pc may make you popular and feel good in the short term but it wrecks your reputation in the long term.

If FT uses AI, use Musk AI.

Humans are flawed. Humans programme AI, therefore AI is flawed.

You can get 100% correct on a test yet that test is written within a Framework of imperfection.

We are a while off yet AI being what people may think it will be or what it can acheive.

1 Like

That’s the sound of a thousand YouTube investment gurus logging into ChatGPT


AI models relay upon huge data sets to create their content. At the moment ChatGPT uses data publicly available to about end of 2021 or something like that. The real interesting thing with AI is when they access private data sets in real time and use that to do their stuff. They could perhaps have access to live market data, financial experts opinions etc… and craft out some sort of response from that. It’s very early days with AI, we have only just started the journey into what will be possible, what will actually be delivered etc… exciting/worrying times ahead

quote=“PeteL, post:24, topic:67886”]
financial experts opinions

Do they exist?
Being an expert implies there’s a chance you will get a prediction correct :laughing:
I suppose there’s got to be a first time. :rofl:

I don’t like using emojis. I find it childish but I thought the idea of financial experts and in particular Economist getting a prediction correct …to be ludicrous.
6 months ago the bank of talking bollocks (also known as the Bank of England) predicted a severe recession almost as bad as the 1930 depression. Now they are suggesting a mild recession.

“The only function of economic forecasting is to make astrology look respectable.”

Oops I am back with another Galbraith!

“Economics is a subject profoundly conducive to cliche, resonant with boredom. On few topics is an audience so practiced in turning off its ears and minds. And none can say that the response is ill advised.”


Most of the training data Chat GPT will have access to will be pump and dump BS, because that’s what most of the “information” on the web is. It’s quite likely it will just amplify the existing pump and dump echo chambers


That’s true at the moment for sure with public data sets for the AI models. I think we will start to see private data sets in all sorts of business/function (not just financial) from well established companies with known knowledge and capability. That would help weed out the pump and dump folk somewhat… AI is so early doors at the moment, interesting times ahead which ever way it goes…

When you look at AI designed hardware you can see that the results are novel, not based on previous iterations or solutions.

I think that AI works in such a manner that it will avoid human made pitfalls, by excluding extraneous datasets.

If you look at what AI outputs for NASA, or the photo that was recently awarded a Sony World photography prize (but declined) then you can see just how creative it can be.

It will not surprise me if AI will need to be legislated as it will be so disruptive that the consequences could be terrific - or horrific!

1 Like