"i" prefixes are for SNPs that don't have rsids, so if 23andMe wants to assay them, they have to give them some kind of internal designation, which is what this "i"-number is. Unfortunately, that means these SNPs don't live in any kind of standard namespace. Internally, 23andMe maps the i-number to the reference flanking sequence, chromosome position, and anything else they need to know about the SNPs to put it on a chip (a lof of these SNPs come from the custom portion of the genotyping array) and interpret the data later. As far as I know, they don't make this data publicly available.
If you have a question about a specific variant, you can probably ask 23andMe tech support how to get a more standard ID. I mentioned this in one of my blogs and they provided the rsID for that particular internal ID:
If it exists, you can find the RS number by looking up the chromosome position of the SNP. Position data can be downloaded from the latest table from ncbi.nlm.nih.gov/SNP/ table name: b141_SNPChrPosOnRef. If no exact match is found, you can sometimes identify SNPs from the flanking regions.