Currently in practice, some companies are already marrying the AI/ML with AQL principles to the automated tool (CMM, OMM, etc.) data and providing faster, cheaper, and more reliable quality inspection reports.
Lower skilled people load/unload the tools and do preventive maintenance.
Higher skilled people are freed to focus on reworking the non-conforming/reject parts if possible or to improve process.

Michael,
I agree with Philip Tuet's thinking: it's a matter of time until the majority of metrology operations are done by AI controlled machines (rather than humans).
The field stands to be revolutionized because everything that is produced has some sort of metrology.
The term Quality Lab will become a thing of the past, less skilled operators will be a) either re-trained to do different operations (e.g. loading/unloading as pointed out by Stephanie), b) kept in a niche operations (example: manual jobs within R&D, where weight of industrialization is smaller, many operations are customized and require human manipulation) or c) simply moved to other kinds of jobs.
In my view, the future of the field will be mostly (r)evolutionized by the following 3 concepts:
1) Live and closed loop: metrology machines will work in close connection with production/manufacturing lines, with different levels of inspection depending on risk/product/business (example: cars, luxury and pharma @ 100%; office utilities at 50%?), and providing a close-loop response that influences/adjusts production settings.
2) Automatic operations: think Quality Lab meets production line. A series of robot arms, conveyor belts, cameras and lasers (e.g. OMM) and contact sensors (e.g. CMMs) will analyze parts without any need for human intervention. A strong contender to these automatic operations is actually Virtual/Digital assistance, where operators are supported by the likes of Google Glass - these will delay the evolution of automatic operations and in my point of view will be used for specialized situations though.
3) Digital operations: Today many labs (most FMCG anyway) still collect data manually in one way or another. Consider how unreliable it is to collect data in physical lab notebooks, or even collecting data manually and filling in ELNotebooks... We today's digital age its' a safe bet to say Cental Lab Computers or LIM systems will be the heart of data.
4) Big data: Today we run maybe a hand full of critical tests on parts. In an automatic world with huge computing power the number of assessments we can generate will be immense, and that will (finally) generate big data in metrology which with the help of AI and machine learning will advance troubleshooting, fine-tune processing and scale economics.
These concepts will evolve in parallel until a (series of) major corporation(s) will combine it all.
I'm actively developing the first 3 concepts in my lab, leading the way and role modeling to the remaining labs in my company. Today its a couple of engines or conveyors that semi-automatize one test or its a Rasberry-pie that collects the data that 20 operators generate... Tomorrow it's a series of robotic arms, sensors and pc software that run an efficient lab, with a single operator filling in the 'orders'.
Looking forward to develop the Lab of Tomorrow, not so that its human-free, but that humans can have a higher added value.

I definitely agree that there's already a lot of work being done in terms of automation, but (in my observations) many processes are still manual. Couple this with the fact that some companies require 100% inspection on increasingly-complex components, and it seems as though metrology often becomes the bottleneck. Would be very interested to hear other's thoughts on this.

If i understand the question corectly, Then i have to say that to calibrate, check and adjust your CNC, CMM & OMM's etc. then...
You have to do the physically measurement jobs and evaluate the data and adjust the machines on its live condition parameters - there is no other way to collect the real microns and remove the faults by adjustments in live conditions.
Yes you can collect the faults by other systems over time and online etc. - but never ever reach precision on the machines nor make the corrections if you do not have the laser optical measurements systems to do so - and do it smartest in Space.
The real fault is that the people requiring the inspections do not know which systems to use for what and can most often not see the advances you get by using the special systems which unique will give all live conditions in real.time.
Most people today still hang on brandnames and use those instruments in the job - which often is decade old systems and requires far to much work and handling - be smart use new technologes & ways of work.
Theres a reason to that smaller companies survive and do good jobs!
My suggestion is that you look closer to these unique OEM's to reach the altitudes in precision the right way. But it requires you to understand their systems down to details to get the unique preformance they really gives you in advances.
Optodyne.com, D-Test.eu, HamarLaser.com, APIsensor.com ( all these will help you by unique Patents giving U more... )
Can we also get some links from above mention alternative ways of work into this discussion ?
Thanks! - Jannick