We've seldom added jobs so steadily, and yet we're terrified by AI. Why?

Since 2013, experts have repeatedly warned that millions of human jobs soon could be obliterated by rapid advances in artificial intelligence. Say goodbye to loan officers, financial planners, restaurant workers, hotel desk clerks and even sports referees. In this doomsday version, our world of work is about to be crushed by faster, better and cheaper AI systems.

And yet -- it's not happening. Not even close.

From 2013 to 2018, the U.S. economy added 12 million net jobs, with further growth since then. Even supposedly vulnerable professions are expanding. According to Bureau Labor Statistics, we now have an extra 11,000 credit counselors and loan officers; 16,000 more personal financial advisers -- and 3,000 more umpires, referees and other sports officials. Each of these narrow categories' job growth amounts to 3% to 15%. In something as broad as food preparation and serving, total U.S. employment has surged more than 1.2 million.

If a galactic battle between humans and AI-powered machines is taking shape, we're doing just fine so far. As I've written before, even routine jobs require a lot of human finesse. Social skills such as empathy can keep us employed for a long time. What's weird, though, is the degree to which AI-related job anxiety remains so intense, even though hardly anything in current employment trends justifies such fears.

It's time to take a closer look at this disconnect, with two key questions in mind. First, why do the alarmists still hold center stage, even if their message of doom seems overblown or premature? Second, does artificially heightened anxiety translate into a useful adrenaline rush -- or a debilitating panic attack?

Let's start with the engineers and computer scientists who have the clearest view of what AI can achieve. They're right to be excited (and maybe even nervous) about machines' growing ability to make "intelligent" decisions in complex areas. Even so, their job-loss predictions often overlook the degree to which supposedly automation-ready fields actually require a lot of hard-to-copy human skills, too.

A case in point: the outlook for financial advisers. Their jobs have shown up on soon-to-be-automated lists since this 2013 Oxford report. Yes, AI systems can build clients' model portfolios far more nimbly than a human adviser can. But human advisers actually do a lot more. They get to know clients in a much deeper way, understanding the emotions and irrational urges that can wreck a sound financial plan.

Big spending decisions like buying boats or choosing second homes aren't just quantifiable goals that can be plugged into a spreadsheet. They are aspirations that need to be explored with the right blend of sympathy and skepticism. In those realms, humans have a durable edge. That's why firms such as Charles Schwab, Edward Jones and RBC Financial keep hiring people. Even if tomorrow's financial planners use AI to help them build model portfolios, keeping the personal touch alive is central to their jobs.

AI's most ardent champions tend to wave aside such subtleties, and for now, they've got the microphone. Still, it's heartening to see people with wider horizons -- like Google's chief economist, Hal Varian -- take a more measured perspective.

As Varian remarked in an Australian public appearance last year, “There are many tasks that make up a job, and automation can eliminate some of the dull, repetitive tasks, but it doesn’t generally eliminate entire jobs. Most jobs, even those we think of as relatively low-end jobs, are much more complicated than we realize.”

Media coverage of AI's impact deserves a closer look, too. Do a Google search of scary phrases like "job-killing robots," and you'll find that that they've seeped into many hundreds -- perhaps thousands -- of headlines in publications such as Inc., Fortune, Vox and Business Insider.

Some of these usages are tongue-in-cheek. Others are meant as eye-catching conversation starters that lead into much more nuanced examinations of AI's limitations. Still, incessant use of extreme imagery, for any reason, can distort pubic perceptions. I'm guilty of one such instance of fooling around myself, in the way I titled this talk at Purdue University's "Dawn or Doom" conference last November. (Sorry. I'll be more careful in the future.)

The real problem with letting alarmist language crowd its way into the headlines, is that it becomes harder for more balanced analysis to win recognition. In the past two years, both McKinsey and the World Economic Forum have released detailed analyses suggesting that while artificial intelligence will greatly change the nature of the work we do, overall trends might tilt more toward job creation than job destruction. Such reports deserve wider audiences, yet they had trouble attracting the catchy headlines that invite a frenzy of readership.

Finally, I'm not sure what to make of employers' enthusiasm for AI as a driver of workplace transformation. In optimistic moments, I'd like to think that well-designed AI systems will accelerate growth, create new business opportunities and provide productivity-enhancing tools that will help workers get more accomplished and earn more money for their trouble.

But after an evening's worth of reading recent investor-call transcripts on Seeking Alpha, I'm not quite so cheerful. Everyone's got an announced AI initiative. Some of these seem to be moving forward rapidly; others sound like relatively empty slogans meant to impress Wall Street. What's most visible is the representation that artificial intelligence -- or perhaps even just talking about it -- is a game-changer.

The more that workers (both high-paid and low-paid) think their jobs are at risk, the less likely they are to press for higher pay, even in an economy with less than 4% unemployment. It also becomes easier for employers to scale back their commitments to full-time workers in favor of shorter-term contractors with fewer benefits, because no one can be sure any job will last long.

It's exciting to think about ways that well-built AI systems can make our jobs -- and our lives -- better. The McKinsey and WEF models present a bracing but palatable model of the long-term adjustments and transformations that this next wave of technology will bring. The more that we get caught up in Armageddon scenarios, though, the harder it will be to steer our way through this evolution.