Which of the following is the PRIMARY reason IS auditors must be aware that generative AI may return different investment recommendations from the same set of data?

The correct answer is C. Computational logic is based on probabilities. Generative AI models do not produce deterministic outputs - they generate responses by sampling from probability distributions over possible next tokens. This stochastic nature means the same input can yield different outputs across runs, especially when model temperature or…

AI Risk Management and Controls

Question

Options

ALimitations can arise in the quantification of risk profiles.
BNeural node access varies each time the process is executed.
CComputational logic is based on probabilities.
DServers are reconfigured periodically.

How the community answered

(38 responses)

A
3% (1)
B
3% (1)
C
95% (36)

Explanation

Generative AI models do not produce deterministic outputs - they generate responses by sampling from probability distributions over possible next tokens. This stochastic nature means the same input can yield different outputs across runs, especially when model temperature or sampling parameters introduce randomness. For financial recommendations, this non-determinism is a material audit risk. Option A (risk quantification) is a downstream consequence, not the root cause. Option B is inaccurate - neural network weights are fixed after training; the randomness comes from probabilistic sampling, not variable node access. Option D (server reconfiguration) has no bearing on output variability.

Topics

#Generative AI#Probabilistic models#Output variability#AI audit considerations

Community Discussion

No community discussion yet for this question.

Full AAIA Practice