With the pace at which things have been moving since 2022, AI researcher is a popular I have seen steadily show up across job lists. Their main role is to research new AI models, fine tuning, orchestration techniques and adapt/adopt to existing infrastructure within a company.

At this point, we are not there yet in terms of letting AI make business critical decisions based on its own outputs. Its meant to serve as a decision support system rather than a decision maker.

To minimize hallucinations, yes AI should be set up for deterministic behaviour (depending on your use case, for example, in recruiter, yes it should be deterministic so it produces the same evaluation for the same candidate every time). Secondly, having another AI check hallucination can be a good starting point, assigning scores and penalizing the first AI can also lead to more grounded responses.

When generating candidate summaries against job descriptions, how does your AI safeguard against potential bias crept into the training of the model? Or how do you address AI hallucination ?