Which phase of the AI data life cycle presents the GREATEST inherent risk?

The correct answer is D. Training. The training phase is where a model learns its weights and patterns from data. Any bias, poisoning, or quality defects present in training data become permanently encoded in the model and propagate into every future inference. Errors introduced at training are the hardest to…

AI Security Risk Management

Question

Options

AMonitoring
BMaintenance
CPreparation
DTraining

How the community answered

(21 responses)

A
5% (1)
B
14% (3)
C
24% (5)
D
57% (12)

Explanation

The training phase is where a model learns its weights and patterns from data. Any bias, poisoning, or quality defects present in training data become permanently encoded in the model and propagate into every future inference. Errors introduced at training are the hardest to detect post-deployment and the most costly to remediate, often requiring full retraining. Preparation, monitoring, and maintenance phases carry risk, but the training phase has the highest inherent risk because flaws at this stage are foundational and systemic, affecting all downstream outputs.

Topics

#AI Lifecycle Risks#Model Training Security#Inherent Risk Assessment

Community Discussion

No community discussion yet for this question.

Full AAISM Practice