The surface brightness of a source is by definition the flux density per solid angle; the surface brightness sensitivity of a telescope is, analogously, its point source sensitivity divided by the beam solid angle.$^{\dagger}$ Interferometers have smaller beam solid angles than dishes with the same area (Frayer 2017). Given that the solid angle is $\Omega\sim\theta^2$, the ratio of the surface brightness sensitivity between an interferometer with longest baseline $b$ and a single-dish telescope of diameter $D$ is proportional to
$$f\sim\left(\frac{\Omega_{\text{int}}}{\Omega_{\text{dish}}}\right)\approx\left(\frac{\theta_{\text{int}}}{\theta_{\text{dish}}}\right)^2=\left(\frac{\lambda/b}{\lambda/D}\right)^2=\left(\frac{D}{b}\right)^2$$
which is a filling factor of sorts. For a single dish, you could say that $D=b$ because the longest "baseline" is the telescope diameter; for an interferometer, we have $b\gg D$, and so the ratio is quite small.
This means that long baselines can be both a blessing and a curse, and it's why you shouldn't always opt for the longest baselines when conducting interferometry. As an example, the Very Large Array has four different configurations, with $b$ ranging from about 1 kilometer (D) to 36 kilometers (A). You might want to observe when the VLA is in the A configuration if you just care about angular resolution; you'd want the D configuration if you care about surface brightness. The D configuration's resolution is worse by a factor of 1/36, but its surface brightness sensitivity is better by a factor of (1/36)$^2$=1296.
$^{\dagger}$The point source sensitivity tells you how dim an object the telescope can detect; the surface brightness sensitivity tells you how low a surface brightness a source can have before it becomes undetectable.