The proliferation of robots in social interactive scenarios presents challenges for successful adoption. Trust and social perception have emerged as dimensions that predict effective interactions. Social robots are not widely used, thus evaluations are mainly based on media coverage. Research suggests that actual interactions could affect these impressions. Whether social perception or trustworthiness vary if robots exhibit faulty behavior (and apologize) has not been widely researched. In a pilot study with a first small sample (N = 22), we investigated possible changes in these dimensions before and after a collaborative interaction (i.e., completing a puzzle) with the social robot Navel in two conditions: Navel gave either correct or intermittently faulty instructions. Results show that social perception of Navel was enhanced after the interaction for anthropomorphism, competence, and sociability/morality, and negative attitudes towards robots decreased. Trustworthiness did not differ between the conditions. We provide first, albeit preliminary, insights in the effect of interactions with the social robot Navel on social perception, trust, and associations with faulty behavior.