Lossy audio refers to an audio format that encodes audio information with some loss of information.
Encoding into a lossy audio format uses minimization and compression techniques (like removing certain frequencies in the music) to produce audio that is much smaller than its lossless counterpart, at the loss of some audio fidelity (though this is often imperceptible).
Formats such as MP3 and AAC are common lossy formats used in digital music.
What gets discarded is not random. Encoders work from psychoacoustics, the study of what human hearing actually registers, and drop the detail a listener is least likely to notice: quiet sounds sitting underneath loud ones, frequencies at the edges of hearing, information masked by whatever else is happening in that instant. At a high bitrate a good encoder is genuinely hard to distinguish from the original, which is why 320 kbps files have been standard on dancefloors for two decades.
The part worth understanding is that the loss is permanent and it compounds. Every time a lossy file is decoded and re-encoded, the encoder throws away detail again, this time working from an already damaged version. A 320 kbps file made from a 128 kbps source carries the quality of the 128, not the 320, and nothing in the filename or the file’s own properties will tell you this happened. That is the single strongest argument for getting music from official sources rather than wherever it happens to be available.
If you are not sure about the file’s quality, you can use software like Fakin’ The Funk to check the real quality.