Whatever your views on Artificial Intelligence (AI), it’s clear that it’s developing at a rapid pace. It’s going to revolutionise many aspects of our lives, and retirement planning will be no exception. As a financial planning tool, EvolveMyRetirement® is unique. It leverages the power of AI to help users find the retirement strategies that are best suited to them. Let’s take a look at how EvolveMyRetirement uses AI, making it such a powerful tool for intelligent retirement planning.
Monte Carlo simulation
Monte Carlo Simulation is the foundation stone of the AI used by EvolveMyRetirement. It runs multiple simulations of a particular strategy using random variables to model uncertainties. In this way, it determines the likelihood of different outcomes based on various factors; these include investment returns, inflation rates, and longevity. By running thousands of simulations, it can generate a range of possible outcomes. And by analysing the outcomes it can assess the probability of success or failure for a particular retirement strategy. This allows users to understand the potential risks and uncertainties associated with their plan and make informed decisions.
Most retirement calculators simply make straight-line projections, ignoring uncertainty. This approach is fundamentally flawed, since uncertainty is the one thing of which we can be certain. A few tools also use of Monte Carlo simulation, which is an important step forward. But by itself Monte Carlo doesn’t reach the level of AI. Incorporated into the EvolveMyRetirement approach is the concept of a Utility Function. This allows the program to compare the results of two or more different simulations, ranking them in order of fitness-for-purpose. So in this way, the program can try out various different strategies, and pick the best based on the utility function. This ability to make value judgements is the beginning of AI.
AI in Optimisation
It’s not enough to answer the question “What’s the risk of running out of money?”. Some retirement calculators treat this the their main purpose. But not so with EvolveMyRetirement. Even though it too can determine risk, it’s main purpose is to help the user arrive at their optimal strategy. As we’ve already seen, it has a utility function. This can determine if one strategy is better than another based on the underlying assumptions. So each time we use the utility function, we run a Monte Carlo simulation. In theory we’d need to run Monte Carlo simulations on every possible strategy to find the best one. But this would take forever. Alternatively we could randomly pick some strategies, and work out the best; but this approach is, well, random.
So we shortcut trial-and-error by using a type of AI known as a Genetic Algorithm. Although this starts off by picking random strategies, it then improves upon them via an evolutionary process. Each generation results in strategies with better utilities. This process converges towards the best strategy far faster than trial and error.
The Genetic Algorithm, combined with the Monte Carlo simulation, helps EvolveMyRetirement identify the most suitable strategy that aligns with the user’s goals and maximizes the chances of achieving a successful retirement outcome.
Large Language Model (LLM)
Large Language Models have been in the news a lot these days, starting with the rise in popularity of ChatGPT. Recently, EvolveMyRetirement has added an AI Assistant feature that leverages the power of an LLM. The AI Assistant is a virtual guide that can helps users understand the program and provide answers to their questions. We’ve designed and trained it to provide helpful information about features, functionality and best practices. And more recently, we’ve trained it to introspect. By this we mean that it can retrieve relevant information that the user has already entered into their plan, or the results of their optimised strategy.
Over time, as LLM technology improves, we expect that more of the features of EvolveMyRetirement® will become accessible via a natural language interface.