I feel compelled to answer this question, because I disagree with most of the answers here.

The common answer here seems to be: yes it will kill jobs, but only the jobs low on the food chain, and it will create more jobs to at least offset the jobs it does kill.

I disagree. I think AI will kill jobs, and over time, AI might kill most “jobs” as we know them. I think that people are somewhat complacent in regard to the economic impact of AI, and will likely be ill-prepared for the changes we have to adapt to in the not-so-distant future.

First, let's start with the comparisons to machinery and automation. They did indeed put factory workers out of work. In that respect, I agree AI today is similar in many applications, replacing workers who have less specialized skills, perhaps call center operators, office assistants (in a limited extent), and maybe soon, taxi drivers and truck drivers. But I would argue that AI is fundamentally different from machinery or most of the other analogies commonly made when answering this question, because AI is growing and is unlikely to stop growing. It's growing in breadth (of applications and industries), in geographic and economic scope, and in power (it's capacity to address increasingly complex tasks). A more fitting analogy would be machinery in an automobile plant that not only made the parts one day, but then learned how to assemble them the week after, and then how to design cars a year later.

I think that there is little that is out of reach for advanced AI of the future. Let's leave the question of the AI singularity alone for now. Instead, I think that deep learning efforts at Google and elsewhere are making AI systems learn faster and faster, with a growing rate of acceleration. Advanced AI can now address increasingly complex tasks including medical diagnosis, stock market trading, weather prediction and human behavioral modeling. Very soon, it will be able to take the place of certain types of teachers, and find a role in education. It can already deal with complex systems in software and mathematics, and seems to be only limited in applications that require interactions with the physical world (sensors are still imperfect), and with people.

So without projecting too far into the future, we can ask the question, what jobs will NOT be killed by AI? Jobs that involve labor are already (or soon to be) replaced. Jobs that require logical reasoning are being replaced, albeit at a slower rate. What are the qualities that humans have that cannot be captured by AI? Perhaps creativity, emotional responses? So perhaps researchers in academia will survive longer than most, and artists (though AI stylistic mimicry is already quite impressive and their results enjoyable), and counselors/psychologists/case workers, and decision makers like CEOs who cannot be predictable or error prone. And hopefully software engineers and algorithm designers who develop AI systems.

This leaves a very, very small portion of today's jobs intact. Many say: we just need to train people to fill higher level jobs created by AI, e.g. Programmers, ML researchers. But this is no easy feat. The US educational system is struggling to meet the demands created by advanced technology today. This failure is in part responsible for the economic divide plaguing the country today. The challenge of educating the public for a job market that is both decreasing in scope and increasing in complexity, is nothing we've ever faced before.

I am far from original in this opinion. But if I had to guess, I would say that AI will put far more people out of work than we can (re)train in time. That will have significant economic repercussions as corporations seeking to minimize cost will do so at the cost of rising unemployment. There will need to be dramatic changes in social policies in order to avert large scale economic disaster, first dramatic rises in minimum wage, then eventually some type of universal income/welfare. It's the kind of change that requires significant leadership in our government, something that seems woefully absent today.

EDIT December 25, 2022

I think that the one thing I would add today is that I am quite surprised at how fast AI is now displacing jobs that many considered driven by creativity. Artists are not actively fighting back against AI art produced by diffusion models like Stable Diffusion and MidJourney, especially when they can be prompted with the names of artists you want them to mimic. I do think that as this displacement happens, human artists will still be needed to inject new forms of creativity into these models, but the art ecosystem will look dramatically different from what it used to be.