Getty Images
How AI can help businesses circumvent inflation
AI can potentially help businesses avoid -- and even counteract -- inflation. Machine learning may be better at this than large language models.
The buzz, hype, enthusiasm, fear, loathing and even panic surrounding AI right now is daunting. AI's practical uses, such as helping to forecast and even tame the issue of inflation, are significant, however, -- when businesses use the correct AI tools.
The advent of OpenAI's GPT-3 and ChatGPT, for example, upended previous conventional wisdom about the nature and role of AI. Even AI scientists did not foresee the degree to which large language model (LLM) applications would surpass previous AI applications across a variety of metrics. As a result, people are turning to AI for help with just about everything.
Within the financial sector and the context of dealing with inflation, using the right AI tools is one key consideration. The other is using them in the right ways.
How AI can and can't assist in financial forecasting
However good ChatGPT and other LLMs appear, it's important to remember several key facts. First, they are not good at everything. They may be able to pass a basic conversational Turing Test or eke out a passing score on a bar exam, but they can produce outputs that are badly reasoned, badly structured and even badly written.
Also, they are prone to hallucinations, or answers that are simply false. When pressed, most LLMs stick to their answers and even cite fabricated proof in the form of made-up statistics or quotations. Without significant intervention in training and oversight forcing AI products to stick to actual facts and their implications, this tendency to mix lies with truths is a major impediment to enterprise use of LLMs.
Thankfully, AI comprises many branches of development besides LLMs.
The financial services industry has long been at the leading edge of applying AI. Financial services and fintech companies have focused mainly on machine learning (ML) for analysis and predictions. ML developers get an algorithm to learn the important parameters in a data stream by feeding it lots of related data points and giving it feedback on its answers to questions. The goal is to build a model that not only gives good answers but also adapts to changes in the data stream -- one that can thereby navigate financial data to find insights.
One branch of ML that is especially important to the financial services industry is deep learning. Deep learning applications use a neural network model that has multiple layers of analysis tied together. Financial services companies can train their deep learning models with enormous volumes of stock market behavior to develop algorithms that help them predict how the markets will move next, for example. In the insurance industry, they already train models on huge amounts of demographic, policy-holder and weather data to predict where crop, fire or flood insurance rates need to rise to address a shifting climate.
Offsetting effects of inflation with AI
AI doesn't make it possible to completely avoid inflation. If a supplier raises prices, a business leader experiences the effects of inflation. ML applications can help offset the effects of inflation by improving a company's efficiency, however.
Consider a retailer. Given access to information about orders, warehouse stock and shelf stock on the one hand, as well as sales, customer and demographic information on the other, an ML system could learn to optimize. It could determine how much ordering, transit and storage of warehouse stock would be needed to save money while also achieving the company's sales goals within a given amount of time. This same concept of optimization through AI can apply to a manufacturer.
A retailer or fast-food company might use ML to help it find not just materials and transport efficiencies but also locational efficiencies. Specifically, an ML model can be trained to determine where a new location should go or whether an existing location should expand, shrink, close or move. Service providers can even use such models to optimize the numbers, skill sets and placement of staff.
Machine learning helps with the effects of inflation, but quantum computing and quantum AI may play a much bigger role in solving such problems in the future. For several problem types, quantum AI will allow businesses to find solutions faster. Quantum AI will also allow machines to find not just better solutions to some optimization challenges but the provably best solution. This future may be closer than it appears, given how quickly generative AI has taken the world by storm.
