Randomization of subjects to treatment and control groups does not always guarantee a valid causal inference in experiments.
The video talks about common problems in experiments: unfair comparison, demand effect, non-consequential decision, and manipulation checks. In the end, the lecturer shows a hypothetical experimental vignette study, which demonstrates a worse-case example of how the design of an experiment can induce a demand effect.
Link to slides: https://osf.io/r7ymt