Free Authentic IAPP AIGP Actual Exam Questions - Question 6 Discussion
Please use the following answer the next question:
A mid-size US healthcare network has decided to develop an Al solution to detect a type of cancer
that is most likely arise in adults. Specifically, the healthcare network intends to create a recognition
algorithm that will perform an initial review of all imaging and then route records a radiologist for
secondary review pursuant Agreed-upon criteria (e.g., a confidence score below a threshold).
To date, the healthcare network has taken the following steps: defined its Al ethical principles:
conducted discovery to identify the intended uses and success criteria for the system: established an
Al governance committee; assembled a broad, crossfunctional team with clear roles
andresponsibilities; and created policies and procedures to document standards, workflows,
timelines and risk thresholds during the project.
The healthcare network intends to retain a cloud provider to host the solution and a consulting firm
to help develop the algorithm using the healthcare network's existing data and de-identified data
that is licensed from a large US clinical research partner.
Which of the following steps can best mitigate the possibility of discrimination prior to training and
testing the Al solution?
A vs C? More diverse data could reduce bias before training, not just assess impact.
Option C makes sense since an impact assessment can help flag potential discrimination risks in the data and use case before any model training, letting them address issues proactively.
It’s C because an impact assessment helps identify potential bias in the data and use cases before any training happens, which is crucial to prevent discrimination from the start. Getting clear on risks early beats waiting for audits later.
B imo, having an external audit early on can uncover hidden biases or gaps the internal team might miss, especially before any model training begins. It adds an objective layer of scrutiny that’s valuable here.
C/D? An impact assessment (C) sounds solid for spotting issues early, but a bias bounty program (D) could catch hidden problems by inviting diverse testers. Not sure if they’re ready for that step yet though.
Maybe B makes sense here—getting a third party to audit the data and methods early could catch bias issues before training even starts, especially if the team hasn’t done a deep dive on dataset balance yet.
C vs B, impact assessment feels more proactive before training than an audit.
Option C seems right since doing an impact assessment upfront helps spot potential bias before training the AI. Just grabbing more data or audits feels like later steps to me.