A company developing and deploying its own AI model would perform all of the following steps to monitor and evaluate the model's performance EXCEPT:

The correct answer is A. Publicly disclosing data with forecasts of secondary and downstream harms to stakeholders.. Publicly disclosing data with forecasts of secondary and downstream harms is not a required or typical step in monitoring an AI model's performance. Monitoring focuses on internal tracking, incident response, and scheduled human review, not public disclosure of harm forecasts.

AI Governance Program

Question

A company developing and deploying its own AI model would perform all of the following steps to monitor and evaluate the model’s performance EXCEPT:

Options

APublicly disclosing data with forecasts of secondary and downstream harms to stakeholders.
BSetting up automated tools to regularly track the model's accuracy, precision and recall rates in
CImplementing a formal incident response plan to address incidents that may occur during system
DEstablishing a regular schedule for human evaluation of the model's performance, including

How the community answered

(40 responses)

A
83% (33)
B
5% (2)
C
10% (4)
D
3% (1)

Explanation

Publicly disclosing data with forecasts of secondary and downstream harms is not a required or typical step in monitoring an AI model's performance. Monitoring focuses on internal tracking, incident response, and scheduled human review, not public disclosure of harm forecasts.

Topics

#model monitoring#performance evaluation#incident response#post-deployment

Community Discussion

No community discussion yet for this question.

Full AIGP Practice