Chondroitin is what cartilage is made of, so taking it puts back what your joints have worn away.
Cochrane pooled 43 trials in more than 9,000 people and found chondroitin better than placebo for short-term pain, but almost all of that came from trials it rated poor quality. On the one outcome it rated high quality, 53 people in 100 taking chondroitin reached a meaningful pain reduction against 47 in 100 on placebo: a difference of six people in a hundred. The benefit became uncertain whenever the analysis was narrowed to trials that were well concealed, large, or funded independently of a manufacturer. The American College of Rheumatology recommends against chondroitin for the knee and hip but conditionally for the hand, on the strength of a single trial. None of this evidence measures cartilage being rebuilt; what it measures is pain.
Chondroitin's honest result is not zero, which is what makes it more interesting than the supplement aisle version. On the strictest evidence it is about six people in a hundred, for pain, and for the hand rather than the knee.