This is an article published in 1997 by Aron et al., where they developed a procedure for to generate interpersonal closeness, aptly named the "fast friends" procedure. Here is the abstract:

A practical methodology is presented for creating closeness in an experimental context. Whether or not an individual is in a relationship, particular pairings of individuals in the relationship, and circumstances of relationship development become manipulated variables. Over a 45-min period subject pairs carry out self-disclosure and relationship-building tasks that gradually escalate in intensity. Study 1 found greater postinteraction closeness with these tasks versus comparable small-talk tasks. Studies 2 and 3 found no significant closeness effects, inspite of adequate power, for (a) whether pairs were matched for nondisagreement on important attitudes, (b) whether pairs were led to expect mutual liking, or (c) whether getting close was made an explicit goal. These studies also illustrated applications for addressing theoretical issues, yielding provocative tentative findings relating to attachment style and introversion/extraversion.

The questions used are available in this article in The Guardian:

What I find interesting about it is that the key to these procedure is to promote self-disclosure, talking about things that are important for people's identity in a non-threatening way using silly questions (e.g. When did you last sing to yourself? To someone else?). And it does so in a progressive way, starting with questions that require lower levels of self-disclosure and eventually getting into more personal questions (e.g. When did you last cry in front of another person? By yourself?).

I think it shows how easily we can like other people given the right conditions, and how important self-disclosure is for friendship development (there is plenty more research on that). Also, the importance of starting slow, but progressing steadily throughout a conversation.