Shove a book across a table and let go. It slows at a steady rate and stops, because the table pulls back on it with a roughly constant force while it slides. That force comes from microscopic bumps on the two surfaces catching and breaking as they slide past each other. Press the book down harder and the bumps grip harder: the friction grows in proportion to the normal force, fk=μkN.
The coefficient μk belongs to the pair of surfaces: roughly 0.7 for rubber on dry concrete, about 0.05 for a waxed ski on snow. A 20 kg crate on a floor with μk=0.30 feels fk=0.30×196=58.8 N whether it slides at 0.5 m/s or 3 m/s. Push with exactly 58.8 N and it keeps moving at constant speed; push harder and it speeds up.
Contrast this with static friction. Push a resting box gently and it does not move; that is static friction matching your push, up to a ceiling of μsN. Kinetic friction takes over only once the surfaces slide, and it has a single value, μkN, usually smaller than the static ceiling. That is why a heavy box is hardest to get started and then lurches forward. Before choosing a coefficient, check whether the surfaces are sliding past each other.
Think of it like this. Kinetic friction works like a toll charged per metre of sliding: the rate depends on what the surfaces are made of and how hard they press together, not on how fast you go through. Every metre slid costs the same energy, paid out as heat. The toll model fails for fluids and for very high speeds, where resistance does grow with speed.
Core idea. While two surfaces slide past each other, friction opposes the sliding with a force of μkN, set by the materials and the normal force and essentially independent of speed and contact area.