We have ChatGPT. Here’s what it will take to get to ‘InvestmentGPT’
Check out all the on-demand sessions from the Intelligent Security Summit here.
From a Google killer to the heralding in of a humanity controlled by AI, the new ChatGPT application has been making waves — and even scaring people, too — since it was introduced late last year.
The logical culmination of advanced machine learning technology, ChatGPT seems to know almost everything — as well as how to express itself in an extremely clear and erudite manner.
But while ChatGPT knows a great deal, it’s not likely to be any better than any human (or any AI system for that matter) at developing a sustainable, profitable investment strategy. This is because ChatGPT is not trained on statistics, but on specific texts, ideas and descriptions. This means that it is incapable of analyzing investment data accurately enough to make reliable decisions on investments, markets or the economy as a whole.
To accurately predict market trends or investment value, AI tools need to not only be trained on data, but go a step beyond the current capabilities of most machine learning (ML) systems. That is, developing predictive models that are trained to update themselves in response to new events that were not part of the original training models.
Intelligent Security Summit On-Demand
Learn the critical role of AI & ML in cybersecurity and industry specific case studies. Watch on-demand sessions today.
Regurgitated financial advice
Trained on a huge array of databases, information sources, journalism sites and social media from around the world, ChatGPT is essentially ML on steroids. With access to so much data, neural network-based algorithms to match data patterns and advanced natural language processing (NLP) abilities to communicate its findings, ChatGPT is able to write essays or articles, do homework, write music, write and debug code, take tests (and cheat on them), woo potential beaus, and much more. And all this as well as, or even better than, humans are capable of doing.
ChatGPT can also provide general financial advice. Still, according to experts, the AI system’s advice is typical of human asset managers and financial advisors. When asked, “What is the best strategy to make money in the stock market?” ChatGPT responded:
“There is no one-size-fits-all answer to this question, as the best strategy for making money in the stock market will depend on a number of factors, such as your financial goals, risk tolerance, and investment timeline.”
With that, it stated that there were some basic principles that investors should follow: diversifying one’s portfolio, investing for the long-term, using dollar-cost averaging and performing research before investing.
“Overall, the key to making money in the stock market is to have a well-thought-out strategy and to be patient and disciplined in your approach,” the system concluded.
AGI: The upgrade AI needs to “play the market”?
That’s a far cry from what we may expect from an advanced system that has access to far more information than the average investor and the most sophisticated analytical tools on the planet. But given the current limitations of ML — especially the fact that learning models can only be built on currently available data — ChatGPT’s financial advice is in line with what should be expected from such systems.
Until ChatGPT and other ML-based analytic systems get a substantial upgrade, they will remain unlikely to outdo human analysts. That upgrade would require a far more flexible modeling system — one that enables the system to change its predictive model in response to new events that could skew existing predictions.
Artificial general intelligence (AGI) systems, for example, could provide the upgrade that AI needs to “play the market,” providing not only more humanlike thinking processes but also allowing those processes to take into consideration a far greater amount of data than humans could deal with at one time.
Armed with huge amounts of data and advanced, flexible analytic systems designed to adjust predictive models as required, AGI-based systems would be a much better bet for investment predictions than current AI systems — including ChatGPT.
“What can (or will) be” capabilities
AGI is still largely under development, but data scientists are working on enhancing current AI technology to enable better investment predictions. The process, of course, is incremental — but more advanced algorithms are being developed, based on the trading experiences of quant funds, which use complex mathematical models to make predictions.
Quant funds rely largely on electronic trading, with millions of trades executed at one time, supplying more data for ML models to develop more accurate predictions. The main difference between these technologies and ChatGPT is that the latter relies on “what is,” while AGI and advanced mathematics-based ML analyzes data sets to develop models of “what can (or will) be,” making them far more appropriate for investment purposes.
AGI and mathematics-derived advanced ML will — eventually — enable better and more accurate investment predictions; it’s just a matter of time before scientists are able to build out the advanced data sets needed to train AI to make accurate investment predictions.
Until then, let’s use current-generation ML-based systems like ChatGPT for the many things it is very good at. “InvestmentGPT” is still in the future.
Anna Becker is CEO and founder of EndoTech
Welcome to the VentureBeat community!
DataDecisionMakers is where experts, including the technical people doing data work, can share data-related insights and innovation.
If you want to read about cutting-edge ideas and up-to-date information, best practices, and the future of data and data tech, join us at DataDecisionMakers.
You might even consider contributing an article of your own!