It really depends on the exact job and industry. In research, I doubt that AI will replace humans in the next 100 years (because AI is a fancy term for what essentially is old statistical algorithms of pattern prediction). Yes, you can automate repetitive tasks (and yes, driving a car is a repetitive task) but you cannot automate creativity, and research requires creative thinking.

In other areas, for example medical areas, AI could help speed up some procedures, such as diagnostics (e.g. cancer tissue recognition) but for ethical reasons humans will still have to check the results, at least for the next 10-20 years.

I think the industries in which AI would have the most impact are computer-related. "Real" jobs, like builder, nurse, kindergarten teacher etc. cannot really benefit from AI right now.

I Kindly refer you to the thoughts presented and discussed by some AI experts at an MIT Conference which focused on “preparing workers for the era of artificial intelligence”, the summary is given at the following link of the MIT News:

http://news.mit.edu/2019/ai-work-future-congress-1122

I kindly refer you to the following article on “Artificial Intelligence market analysis and forecast to 2025”

https://www.marketsandmarkets.com/Market-Reports/artificial-intelligence-market-74851580.html?gclid=EAIaIQobChMIhaWxuuyY5gIVGpzVCh0rEQm0EAMYASAAEgJLuvD_BwE

Some scientists believe that the greatest fear about AI is singularity (also called Artificial General Intelligence) (AGI), a system capable of human-level thinking.

The following link summarize the opinion of 373 experts on the issue of AGI singularity by 2060.

https://blog.aimultiple.com/artificial-general-intelligence-singularity-timing/

AI impact on the future workplace will be the same as every other invention throughout history. To put context to my reply let us look at what is the underpinning of the assumptions of AI being a "job destructor " and why it is not well founded

1) The start of the question of AI replacing jobs started at around 2012 when the “jaws of the snake” was noticed and AI was tagged as the source. A good analysis refuting this point can be found in:

https://thebreakthrough.org/articles/its-not-about-the-machines

Also, at the time the analysis carried out did not take into account previous dislocations of jobs by technology which can be found even on the first industrial revolution. A takeaway from looking at previous dislocations is that it occurs only for short periods of time until it rebalances due to retraining and employee adaptation

2) AI has already gone through several AI winters and most likely will happen again. A counter argument can be found at:

https://www.forbes.com/sites/tiernanray/2019/09/12/no-there-will-be-no-ai-winter/#691bf73646a5

While industrialization can be argued that has taken place (as the article does), it has not been quantified in the article and contrasted with previous technological adoption curves making the argument suspicious to say the least. Further, the article does not go deep into the technical aspects of AI besides saying the magical words of "deep learning" and being adopted in industry (how many and is it meaningful?). The magical words do not justify the argument

3) Deep learning, for all its merits is just pattern recognition with some improvements. Yet, it is being taken as Artificial General Intelligence (AGI). There exist a huge gap to jump from one concept to the other.

AI in its current incarnation has its merit, but it is not the universal game changer that the hype makes it out to be. A sober look will show opportunities as is expected, but it should be accompanied by the right expectations.

I feel that many things cannot be automated at the present stage of AI research. It is difficult to fully model human thinking and communication. These topics are still heavily being researched upon. We are able to replace technology with AI only if we are able to completely understand human thinking and behaviour. Therefore, I felt AI can at most replicate certain repetitive task but complex tasks still need to be achieve by human.

Automation is an important manufacturing tool. It is tireless and repeatable. While not error free (machines do fail), the error rate is less than humans and more importantly machines don't "work around" problems. So the cost reduction one can expect comes from getting closer to the zero rework asymptote all the way through to the customer delivery. However, we still need people to handle the unexpected, repair and even create the machines.. Even with AI.

One thing humans have done to rise above stronger and faster species is accumulate knowledge. We do that by teaching our children what we have learned and teaching each other what we learn now. For AI to achieve equality and even surpass humans I believe it must to do that. What AI can do right now is process big data and with a lot of equipment for quickly that the human mind. That makes one wonder about AI and the "threat" it poses.

As far are replacing workers, sure. 120 years ago I am sure that a group of blacksmiths having copy at the 19th century Starbucks lamented about the automobile and why it wasn't a good as a horse and what it would do to their industry. And it did all but eliminate blacksmiths. But is make tire makers, wheel makers, engine builders, chassis builders and more.. None of which existed at the time. The difference between now and then is the speed at which innovations such as AI permeate our lives. Cars to nearly 50 years, AI will take 5 if something slows it down. This creates a gap in training. There just isn't enough time to retrain the existing workforce for the new jobs and careers that evolve with these new technologies.

I have many more thoughts on this subject, but I'll stop here.