Recently, a movement to the use of automated, or “robo,” advisors has been occurring in the field of financial advice. So, it’s timely to look at the details of what is available, whether it is worthwhile and if it even works. Historically, investors generally had two choices: do it themselves or hire an advisor.

Today, there is a third choice—automation. Yet, for the most part, automation in the investment world has been around for quite some time, since the advent of asset-allocation models and modern portfolio theory.

Additionally, mathematicians long ago came out with systems to help generate outcomes, such as the Monte Carlo simulation and the portfolio stress test. These were used to show clients how a portfolio would behave during serious market duress.

But the results of these systems made us skeptical, because they only looked backward; these systems had no ability to forecast the future.

Robo-advisors, as we see it, are really no different from these quantitative tools of the past—with a few mathematical solutions added in to represent human intervention.

The tools robo-advisors use for allocation are based upon historical inputs to generate a recommended asset allocation for a client.

But in the outcome, the portfolio is always filled with what worked really well for the last decade or so. Granite Group knows that this is not the best way to invest your money for the future, as the results are based on backward-looking data.

Will robo-advisors alert you to sell or to change investments if the entire portfolio-management team leaves? Most likely not.

A second element to consider with robo-advisors is not talked about much: the process of due diligence (qualitative). While a robot can use index funds as its base case, anything outside of simple investing is not realistic, because a robot does not perform the due diligence needed.

In contrast, a human advisor (assuming he or she is doing the job right) can look “under the hood” at a hedge fund or a private investment. A robo-advisor cannot. Will robo-advisors alert you to sell or to change investments if the entire portfolio-management team leaves? Most likely not.

They cannot look at an organization to make sure it’s a sound business; and they can’t look into the eyes of a portfolio manager and know that there might be something deceitful in the data presented. These elements need to be considered in the final analysis.

To prove its case, Granite Group recently decided to customize risk tolerance with forward-looking inputs, so allocations would be aligned with clients and the markets. And what Granite found, with these changed inputs, were results very different from those of the original outcomes of the automated version.

The conclusion? This process can be done only by experienced professionals, not robo-advisors.

Simply put, using automation for part of your investment thesis can be a worthwhile part of your planning, but it cannot be the entire process. If an investor is looking to avoid allocation, operational and manager risk, a robot will never be able to help. And, in the current environment, when a portfolio or retirement plan is doing poorly, and personal involvement is needed, a robot is not even in the game.

Perhaps in the future, when human-like judgment skills are built into robots, robo-advisors may become more valuable. Until that day, however, using a robo-advisor alone may actually hurt your investments and therefore hurt you.

This article was originally published in the August/September 2016 issue of Worth.