end result
We in contrast Smart Agent to a baseline of conventional voice-controlled AR assistants. We measured cognitive load utilizing the NASA Job Load Index (NASA-TLX), total usability utilizing the System Usability Scale (SUS), person desire on a 7-point Likert scale, and complete interplay time.
A very powerful discovering was a discount in cognitive workload. NASA-TLX knowledge confirmed that on a 100-point scale of psychological calls for, the imply rating for Smart Agent was 21.1, in comparison with a statistically vital distinction (𝑝 < .001) from baseline of 65.0. Perceived effort decreased considerably as properly (𝑝 = 0.0039). This means that the proactive system efficiently alleviated the psychological effort of formulating queries.
When it comes to usability, each programs confirmed good efficiency and there was no statistically vital distinction between the SUS scores (𝑝 = 0.11). Nonetheless, individuals expressed a statistically vital sturdy desire for Smart Agent (𝑝 = .0074). The typical desire ranking on a 7-point scale was 6.0 for Smart Agent in comparison with 3.8 for baseline.
The interplay time recorded from the second the immediate was triggered to the system’s closing response to the person’s enter was sooner for the baseline (μ = 16.4 seconds) in comparison with Smart Agent (μ = 28.5 seconds). This distinction is an anticipated tradeoff of the system’s two-step interplay stream, the place the agent first suggests an motion after which the person confirms it. Customers’ sturdy desire for Smart Brokers means that this trade-off is suitable, particularly in social contexts the place discretion and minimal person effort are vital.
