Typesetter. Fax machine repairman. AOL technician. Watchmaker. Punch card processor. Film developer. Switchboard operator. Typewriter repairman. Encyclopedia salesman. Paperboy. Tinker, tailor, candlestick maker.
(Ok, the last three were older.)
But it's hardly an exhaustive list. Neither was pizza delivery boy for new jobs. We also have oncologists, data scientists, app developers, and (modern) industrial engineers.
(If you want more, type into google "careers that didn't exist" and it will offer suggestions... 20, 10, even 5 years ago. simplyhired shows 72,000 "social media" jobs right now. Even though probably half of them are spam, it's thousands, for a phrase that we just made up a few years ago.)
In the 90s, manufacturing dropped like a rock, but overall, employment ticked up. So did median household income. The shift was primarily due to trade liberalization. The recent recession decimated jobs across many fields, as you'd expect from a drop of jobs due to flagging demand (driven by pressure towards deleveraging, not unexpected given high personal debt to GDP ratios).
So automation isn't displacing workers yet, other factors are. If the automation dystopia hasn't even begun, how do we know it's starting? We go on a hunch, and say technology will probably displace all jobs at some point in the future, maybe five years from now on no evidence, because it feels right?
Will new jobs and new sectors always replace the old? All of economic history suggests yes. Armchair speculation suggests no.
Go with whatever your gut tells you, and forgive me for opting for a more conservative reliance on metrics. Sometimes metrics are wrong, but that's where I hang my hat. I will have your back if we reach 35% unemployment in a few years (or ever). I hope you'll pledge to ditch linear economic forecasting if we hit sub-6% unemployment by 2017.