The second half of the video is about observer's. Including the question of whether they can randomly fluctuate into existence. Sean Carroll says that quantum physics says that "if you are not looking at a system it "looks like" a wave. When you look at it it "looks like" a particle. He then relates that to space. To paraphrase, not looked at; just a wave configuration of the vacuum, "nothing happening". Looked at; there is a non zero probability of seeing things eg. "an observer, a brain, a galaxy".
That sounds to me like an incomplete model. The unidentified unseen wave configuration can be related to the 'data pool' of potential sensory information, which is not actual events happening or objects existing but signals with the potential to be converted to sensory information. But although nothing -discernible without looking- is happening within the potential sensory information, there are still happenings occurring between existent things that are the sources of 'encoded' events within the 'data pool'. Change is happening, foundational time is passing even when the observer is not receiving and processing information from (potential sensory information) signals.
Sean's discussion ends with talk about Boltzmann brains and the likelihood that most observers are products of random fluctuations. The kinds of things I regard as observers are organisms with functioning sensory systems, devices that are able to receive and process signals from the external environment into a 'sensory 'product, and sensitive materials that react to stimuli forming a product. With the exception of the sensitive materials, these seem too complex and too similar to be products of random fluctuations. Some silver nitrate could 'randomly' form due to particular circumstances but it us far from a thinking brain. It can produce a product due to 'light' exposure but can not itself analyze or be aware of it. Sensitive materials, 'sensing' devices and organisms sensory systems could be considered as a kind of spectrum of observers that can be ordered according to the complexity and completeness of observation they perform.