Page 55 - AI Ready – Analysis Towards a Standardized Readiness Framework

AI Ready – Analysis Towards a Standardized Readiness Framework

(continued)

Examples Potential AI impacts
Characteristics
1. Awareness and training
2. Trust and security
3. Driver certification with awareness of the technology and usage of the
technology.
4. Upgrade of training for both drivers and pedestrians.
5. Adoption level including consent to provide data.
6. Human feedback, surveys, training, and simulations.
7. Supervision and oversight [68]
8. Usability of visualization apps, websites, and mobile apps for advisories.
Human factors, 9. Involvement of local communities, taking into account traditional knowl-
trust, interop- edge from communities [51]. Deployment of tinyML models in a farm [67],
erability, and for example, to provide local conditions and assist farmers in deciding
standards when to plant crops.
10. Local accent handling e.g. Khmer ASR (speech engine), TTS (fast speech),
a chatbot (sentenceBERT for finetuning), for accent handling [2], dialects
and math, physics and chemistry formulae [43].
11. Expert annotation of the anomalies and irregularities in public service
documents [57], [58], plant leaf images [60], and other public information
related to agricultural water usage [71].
12. Signed consent by operators to the State actors for collection and use of
video data [1].
13. Verification of answers provided by Question-answering models [27].

1. SRC (source) - real-time - is the driver, RSU (roadside unit) sensors,
Weather conditions
2. C (collector)
3. PP (preprocessor)
4. M (analytical model)
5. P (policy)
6. D (distributor)
7. SINK (where the action is applied)
Example:
1) Data collection – conversion – quantization/aggregation/range-checking –
Open data: network – model – users [15]
Data handling 2) Data format Conversion of sensed data to formats to protocols such as
LoRa. [48]
3) Local contextualization and mapping to local regulations and policies.
4) RAG-based backend database with dynamic update capabilities.
5) APIs for asynchronous update of warning systems and endpoints for
consuming generated predictions [50]
6) Streaming formats for camera data for fire detection [51]
7) Satellite data formats and APIs for heat sensing and detection of wildfires
[51]
8) Historical records of procurement-related text documents, mostly in local
language [57], [58]

48