Do You Need an AI Financial Assistant?
Introduction
The topic of the year in 2024 is Artificial Intelligence. It's been all over the news constantly, with more interest and investment globally than ever before. All the big tech companies are either building their own proprietary AI models, making microchips for processing these tools or partnering with AI companies to bring solutions in house.
But where are we actually in terms of progress with these?
Large Lanuage Models
We finally saw something resembling a pass of the Turing test which means that it could adequately pass as a human to another human. Now, that's not to say that we have reached what's called Artificial General Intelligence, where computers are actually thinking for themselves. But what we do have is advanced models, called large language models, or LLMs.
These are essentially scraping all of the available information on the internet and generalising it into a pool from which data and inferences can be drawn. Of course, this means that the models won't stand up to a great deal of scrutiny on deep logical problems, but they are certainly more than adequate for solving surface level problems, or answering questions that you might previously have asked of a search engine.
How useful are LLMs and AI right now?
It's a good question. The answer is that they are able to perform almost all simple tasks and hold continuing conversations where they are using the inputs from previous messages to hone their answers for the future.
So they're constantly improving - not just at a macro level, but also on a micro level during your interactions with them.
Training a Model
This means that you can essentially 'train' a model to improve its answers. By telling these models what of their output is relevant and what isn't, you're telling it what to focus on and how to improve. Now, where it gets interesting is when you tell it to take on a persona, and "talk to me as if you are a financial adviser" - it will then use information on the internet to decide what a financial adviser would sound like and what sort of things they should say, and use those to build its answer.
Now, that's pretty good if you're talking about something where there is a generally accepted standard, which we are.
How can I use an AI assistant myself?
All this begs the question - how can you actually use it to improve your life. The simple answer is that you can just start testing out Chat GPT, or one of its competitors, with simple questions and problems.
You can ask it to write emails for you, input data sets for it to make inferences from, or simply ask it to summarise large pieces of text.
It will do all of these with no problem. However, it can also provide information that has more practical applications, and that's where Strabo comes in., as your digital asset manager.
An AI financial assistant
We started playing around ourselves with an AI financial assistant - given that you are syncing balance and transaction information at points in time with the Strabo dashboard, you are essentially putting together a basic picture of what your finances look like over time.
What if we could input this into an AI model, and allow it to generate some inferences and suggestions, based on a 'target persona' with a specific risk mandate. It would use your age, appetite and tolerance for risk and a few other things to determine what a sensible portfolio rebalancing or tax optimisation strategy might look like.
Now, that isn't to say this could replace, or even compete with your financial adviser. But having a mini-financial assistant that has access to some of your financial history to give you an informed co-pilot when making decisions? Now that could be a game changer.
Strabo AI
With that in mind, we're releasing the beta of Strabo AI, a model that's been trained to provider information as a financial adviser would, based simply on the inputs you provide when signing up to Strabo.
We've nicknamed it Strabonic, and we're pleased to report that it's going live in November. You can ask it basic questions about financial literacy - things like "What is an ISA", or questions about the market like "Is now a good time to invest in stocks?" but also more direct questions about your portfolio like "Am I adequately diversified?"
It will take the information you've provided to us and use this to compile a picture of what your finances could look like in future, and how to get there.
How safe is it?
We're operating the model on an opt-in only basis for now, so that only users who consent to their financial information being used in the model will be able to proceed. This is to ensure that we aren't sharing any information with Open AI without user consent.