LambdaCDM, for the currently preferred cosmological density Omega_0 and cosmological constant Omega_Lambda, predicts that the Universe expansion decelerates from early times to redshift z~0.9 and accelerates at later times. On the contrary, the cosmological model based on conformal gravity predicts that the cosmic expansion has always been accelerating. To distinguish between these two very different cosmologies, we resort to gamma-ray bursts (GRBs), which have been suggested to probe the Universe expansion history at z>1, where identified type Ia supernovae (SNe) are rare. We use the full Bayesian approach to infer the cosmological parameters and the additional parameters required to describe the GRB data available in the literature. For the first time, we use GRBs as cosmological probes without any prior information from other data. In addition, when we combine the GRB samples with SNe, our approach neatly avoids all the inconsistencies of most numerous previous methods that are plagued by the so-called circularity problem. In fact, when analyzed properly, current data are consistent with distance moduli of GRBs and SNe that can respectively be, in a variant of conformal gravity, ~15 and ~3 magnitudes fainter than in LambdaCDM. Our results indicate that the currently available SN and GRB samples are accommodated equally well by both LambdaCDM and conformal gravity and do not exclude a continuous accelerated expansion. We conclude that GRBs are currently far from being effective cosmological probes, as they are unable to distinguish between these two very different expansion histories.