Have you heard self-driving cars are going to replace normal cars in 5 years? If you have, so did I—in fact, I first heard this about 7 years ago. Which is funny, because self-driving cars still aren’t really here, except in some very small fringe cases that are still rare enough to be newsworthy.

The reason those promises of only autonomous vehicles didn’t appear (and why many other jobs haven’t been automated into obsolescence) is that these predictions of jobs being replaced by robots tend to be much more complicated in reality than in theory. Perhaps it’s because we have simple labor models in our heads where we see a person doing a task by route and then we imagine a robot coming in and doing that task. In that case, it’s easy to imagine the jobs going away.

Alternatively, maybe we’re obsessed with factories. It’s undeniable that factory jobs have disappeared as more processes within factories have been done more by robots and less by people—but ultimately, replacing those tasks-by-route has not meant widespread unemployment in the manufacturing sector. Manufacturing jobs are actually going up, and have been for several years. The story here is much more complicated, and recent economic studies indicate that looking at automation in a vacuum gets you a very distorted picture of what’s really going on.

So let’s take that conceptual framework and apply it to Wall Street. Are finance jobs going to be automated away? Many newbies to the industry might not realize that this was a very serious question being asked in the years after the 2008-2009 crisis, and the discussion reached a crescendo among the rise of the robo-advisors in the mid 2010’s, where traditional financial advisory was seen as a dinosaur that was destined for extinction.

Much has happened since then. A recent study shows 85% of investors would prefer a human financial advisor over a robo-advisor. Last year, two robo-advisory firms (including the second-largest one, Wealthfront), settled with the SEC to get past charges of fraud. To make things even worse, Wealthfront has had to cut fees on its risk-parity strategy product due to competition and terrible performance. Robo-advisors’ AUM growth slowed substantially in 2017, and there’s little reason to see the trend getting better.

Beyond the obvious preference for humans over robots, the robo-advisor world is at risk of failing to see the unique and complex nuances that exist with every single investor. For instance, one family of $2 million net worth may have different needs than another family with a $2 million net worth if the second family unexpectedly has a child, or a desire to change careers, or a sudden family illness…there are many variables that a robot could theoretically manage to answer somewhat competently, but the current crop of A.I.-driven financial advisory firms has not been designed to provide.

This is not to say that the financial sector’s job market is healthy. Quite the contrary. While many sectors have recovered since 2009, the financial sector has not, over a decade later. Many jobs have disappeared, and the opportunities that remain are much more competitive.

But the kinds of jobs that have disappeared are very telling. Mortgage advisors are a shadow of what they used to be—which makes sense, if we’re comparing to the housing bubble of the mid-2000s. M&A bankers are rarer, because M&A activity is down. Sell-side equity analysts have downsized because of new regulations (e.g., MiFID II). In fact, a quick look at almost any sub-sector in the financial sector over the last years demonstrates a logical and non-robotic answer to why the jobs are down.

If we want to look at the jobs that have been replaced by robots, we need to go back further. Stockbrokers aren’t really a thing anymore, but being a stockbroker was a lucrative job in the 1980s. What happened? Discount retail brokerages. The active trading pit of the NYSE has become a TV studio for CNBC—again, because the jobs were replaced by robots. And this makes sense, because the value of these jobs ultimately was in the ability to process transactions between buyers and sellers in a way that is easily automated; the low-hanging fruit was plucked by robots decades ago.

If this sounds like good news/bad news, that’s because it is. The robots aren’t coming for the jobs in finance—at least not now. But other dynamics are. Changes to regulation, market demands, market conditions, and efforts to control bubbles are all going to make the financial sector healthier, but smaller. And that means the aspiring financier needs to understand how the market is changing to survive.