Humans are industrious creatures in that we'll find something else to do
The trick with this round of automation is if we develop advanced enough AI to do the automation, there isn't going to be something else
I think you're both right, to some extent. There will be new types of jobs and perhaps new industries that either can't be done by machines, or that people would rather have (and pay a premium to have) humans do. But personally, I don't believe those will come near to matching the numbers of people who become unemployed, perhaps even unemployable, due to automation.
There are some hard-to-predict variables that confound discussions like this. For one, how fast is machine learning/AI/automation getting better? How soon jobs can be replaced depends on that. Will it be three years before we see significant impacts, or 5 or 10 or 15? Also, how fast will the adoption of new tech be? That depends on how costly the tech itself is, how much money it actually saves, how much retrofitting/redesigning is needed to incorporate it, and how good it is. I am sure that we will see companies that go out of business because they are ahead of the curve and try to deploy new tech before it's ready, as well as ones that wait too long and get out-competed by more automated businesses.
And if you can answer all those questions for AI/machine learning, for many jobs you will also have to answer them for robotics/sensors/physical environment components.
On top of those considerations, we can't really predict yet how automation might change business processes. There are certainly going to be things made obsolete by automation. There are also going to be things that are currently obsolete, but are still being done, that will be exposed by looking at whether a job/function/task can be automated. Very likely, there will be cases where one function or task prevents a job from being automated, resulting in that function or task being eliminated rather than holding on to human workers to do that bit. It won't always be possible, of course, but there will probably be a fair number of people surprised at losing their jobs because they thought it was "impossible" to get rid of some aspect of what they do.
How soon and how fast are the big questions, and that will of course vary by industry/job type. For something like self-driving cars, you would expect "how fast", i.e. rate of adoption, to be pretty fast once level 5 is reached since nearly 100% of any given driving job will likely be handled by the automation, and there are other advantages aside from the direct cost reduction of replacing the human labor (e.g. more than doubling the available operating time of vehicles in trucking, since human drivers have strict limits on hours of operation/service). Whereas if you have an automated system that can do 10 or 20% of a job, there is less incentive/advantage to adopt quickly.