For this, an auditor should have a deep knowledge of development, design and security standards, user needs and ethical considerations, with latest algorithms updates.
Areas to be covered by technical audit While there are methods that developers and researchers can leverage to gain information for effectively auditing systems, such as the scraping approach (i.e., issuing repeated queries and observing system behavior) or code audits (i.e. using tests to understand what vulnerabilities may exist in the source code),[1] it is sometimes the case that users' insight is critical to understand problems with a system (e.g., concerning the ethics of artificial intelligence).
As such, there are methods of soliciting (nontechnical) user perceptions and feedback to support a technical algorithm audit.
[1] Crowdsourced or collaborative auditing is another approach, in which users are considered as testers, sometimes specifically hired to do so, to provide feedback about system design and behavior.
[1] With a trend toward user-centered, ethical systems (e.g., to avoid issues in which harm or bias may go unchecked due to lack of expertise or knowledge that only a diverse range of users can surface), incorporation of users' feedback in the auditing process is becoming increasingly common.