I take it you are asking how much time a reviewer should spend on reviewing an individual paper.*
You should spend as much time as it takes to ensure that the paper satisfies the fundamental criteria for scientific inquiry that apply in your field. This is the minimum. Some journals or conferences also ask their reviewers to assess the novelty of the contribution and its topical fit with the venue. Personally, I believe a reviewer should go beyond a pure quality check and make constructive suggestions for improvement, if the paper has potential.
How much time is "as much as it takes"? This varies depending on your practice, on the paper, and on the discipline. From what I hear, reviewing math papers can take many working days. In political science, which is my own discipline, a thorough review takes me the better part of one working day. This is a mean with a large dispersion around it.
15 minutes are not enough time to read a paper, let alone to apply even a basic quality check, unless the paper is very obviously insufficient. Rubber-stamping every paper is not enough; arguably, it is even worse, because you don't even pretend to do your job. Both "approaches" simply bypass and sabotage the peer reviewing process.
The editors and readers, i.e. the scientific community, entrusts you, as reviewer, with the task of upholding the scientific validity of the research that is being published and disseminated. You and your colleague are breaking this trust. You are doing the community a big disservice. Please stop immediately to take on any more reviewing tasks.
If you are concerned about the time reviewing takes away from your own research, do less reviews rather than more superficial ones. Peer reviewing builds on reciprocity. If you want to publish in a refereed journal, you should "give back" and review. Since most journals require two reports, reciprocate by writing at least two thorough reviews for every paper that you publish in a peer reviewed venue.
If you are concerned that you lack the expertise to write a review, tell the editor that you aren't qualified, and they will find someone more suitable.
*An earlier version of the question asked how much time do you spend. Nobody is interested in how much time I or any one individual spends on reviewing a paper. Perhaps you're interested in some general patterns across many individuals, but I'm not aware of any research on this.