Page 61 - AI Ready – Analysis Towards a Standardized Readiness Framework

AI Ready – Analysis Towards a Standardized Readiness Framework

Table 3: Analysis of use case scenarios (continued)

Use case Scenarios Type of sensor/ Deployment as Thresholds for Experiments and
actor Integral/add-on validation and Specifications for
controls compliance validation�
Early Classification Smart- To be studied Pilot deployments in
Healthcare detection and ensemble phone-inte- based on local hospitals [68]
Applica- of critical models grated model requirements
tions
conditions deployments. and ethnicity
Live Sign Patients and Mobile inte- Dialects Zimbabwe’s Primary
Language healthcare grated GenAI related to sign Healthcare [77]
Translation professionals. and synthesis languages [77]
Gestures, Applications
audio, and text

Accessi- Audio-vi- Accessibil- Mobile Chinese Sign Smartphone
bility sual speech ity features integrated language OS-based Informa-
recognition hosted in application tion Accessibility
smartphones with Multi- Solutions [2]
modal Large
Model with
audio and
visual inputs
AI-based Weather Add-on: Ray [80] Regional State Grid
clean sensors, Power tune hyperpa- [80]
Clean energy plant measure- rameter tuning
Energy prediction ments, and
solar panel
monitoring
Target CCTV camera Add-on: [1] Network operator
detection image infor- AIGC-based internal deployment
mation generation of [1]
Video training data
Analysis samples, use
of annotation
formats, feder-
ated learning
Training Assisted Editing tools Assisting bots Not Proposed sandbox
and content for content are addons for available for African contribu-
knowledge generation creation editing tools tors [27]
creation
Resource Natural Overlay Thresholds Digital twin-based
optimiza- Language (add-ons) for anomaly sandbox
tion and Processing based GenAI, detection, in
5G/6G autonomy (NLP) module digital twin 5G/6G, slice
connectiv- and Recom- resource
ity
mendation configurations
engine solu-
tions

Table 3 covers the analysis of different scenarios under different use cases in clause 4. For each
scenario, the type of actor, controls, thresholds for validation and compliance, and experiments
for validation are studied.

The type of actor in each scenario determines the interaction between AI model and the external
world. For example, in the scenario where the aim is to detect driver distraction, drowsiness
detection relies on cameras that can track eyeball movements whereas the monitoring lane

54