Welcome! Bioinformatics is just a term that describes an approach to doing biology. People who use ( DNA sequencing/genomics/protein informatics + programming ) to answer biological questions are after the same answers as the the people who use ( pipettes/microscopes/molecular biology + a lab notebook ).
My best suggestion is to learn about some biological concepts that are interesting to you, see how people have applied bioinformatics to that branch of biology, and think if there is something you might be interested in doing to work on that question. It is critical to have a strong understanding of the underlying biology to get to the core of new findings using bioinformatics.
A lot of the hardcore algorithm development going on today is in genome assembly with long + noisy DNA sequencing reads, aligning long+noisy reads to genomes and trying to figure out the differences between the maternal and paternal haplotypes, working on evolutionary questions with population-level genetic variation data, and figuring out how things are related to each other with phylogenetics, to name a few.
As far as a machine-learning type approach which you seem to be hinting at in your question, one very hot topic at the moment is single-cell sequencing. This is a set of technologies that allows us to get DNA, RNA, or protein sequencing from a population of single cells to look at heterogeneity in tissues, timepoints, et cetera.
Good luck! If you have some free time on your hand this website, ROSALIND, is a fun project Euler-type website where you can learn some biology while programming bioinformatics solutions.
