Page 26 - CT_AI_Class-8

Ask
AGENT
OrangeAI
What challenges are associated with deploying AI models in real-world environments?

Study

Monitoring and Maintenance

After deployment, the next important stage is Monitoring and Maintenance. This stage ensures
that the AI model continues to perform accurately over time.

Monitoring means continuously checking the performance of the AI system. Maintenance means
updating and improving the model whenever needed. During monitoring, the following questions
are checked:
 Is the system working correctly?

 Is the model giving accurate predictions?

 Has accuracy improved or decreased?
 Are there new patterns in the data?

 Is the model still useful in real-life situations?

For example:

After some time, the model's performance may decrease because:
 Students may start using new gadgets like smartwatches or gaming consoles

 Online classes may increase screen time

 Students' daily routines may change

 New social media platforms may increase gadget usage

 Academic patterns may change

These changes can affect model accuracy. Therefore, regular monitoring is necessary.

If the model performance decreases, improvements can be made by:

 Adding new student data
 Removing outdated data

 Retraining the model

 Fixing prediction errors

 Adding new features (gaming time, social media usage, etc.)

 Improving model accuracy
Monitoring and maintenance ensure that the AI system continues to perform efficiently and

remains useful over time.

24 Artificial Intelligence (CT & AI)-VIII