The question is really not well defined for two reasons:
What would you accept to consider being an observation? It seems to me that one accept to consider something to be an observation only if it can understand it extremely simply. Well, the methodology of observation in evolutionary biology are not always very simple to understand for non biologists.
I don't know what is a change in kind.
Because the question is poorly defined, it makes it almost impossible to answer. And that is the common problem with creationist arguments. Their arguments are not scientific and therefore cannot really been argued for or against. Just want to say also that evolutionary biologist do not need non-scientist to fight against their work. Scientists create their own criticism. Scientists do not form a political party that fight for one point of view just for the purpose of defending it. They think neutrally and objectively.
Let's assume change of kind means change in the DNA sequence. Change of kinds have been observed in real time in almost all lineages that exist including humans. These changes are not only observable at the genetic level but also at the so-called phenotypic (loosely speaking the phenotype is way the organism looks like) level. We can for example think of disease in humans that are due to de novo mutations.
If by change of kind, one mean "adaptation", then adaptation to new environment have been observed in tons of lineages again. In a lab it can be observed in bacteria within the course of a month. It has been observed in many mammals, birds, plants and unicellular eukaryotes.
If you accept observation based on slightly more complex method such as molecular clocks and comparative genomics, then you'll find thousands of example of "change of kind", again in roughly speaking all lineages of the tree of life.
Very recently I worked on the speciation of clownfishes. This speciation was dated and we observed morphological and genetic adaptation to specific environments.
P.s. I haven't looked at the video completely but only some randomly selected part for a total duration of about 30 seconds! At first sight I'd like to say: Please don't fall into the trap of the media. One making interviews of people that are sometimes identified and sometimes not and that select for some part of their answer and that ask poorly defined question can really easily say anything from his records.
I welcome editing and I welcome anybody wanting to add references. One can find tons of them!