Past questions: Software Automation (Year 12)

1: Identifies the correct answer.

Supervised learning

1: Identifies ALL the items that are key principles of DevOps.

Rapid feedback; Automation of repetitive tasks; Collaboration and shared ownership.

2: True or false correct for all SIX check boxes. 1: Correct for FOUR of the six.

True: spam, two-party election outcome, university admission (all yes/no classification). False: house prices, height v weight, spread rate (continuous values: linear/polynomial regression).

5: Comprehensive discussion of the impact on workers with reference to the scenario. 4: Outlines both positive AND negative ways. 3: Outlines some ways. 2: Outlines ONE way OR identifies some ways. 1: Some relevant information.

The AI monitoring system can improve safety on site by reducing human error through real-time alerts and remote monitoring. It can also help supervisors track workers' compliance with PPE requirements and identify risky behaviours, potentially preventing accidents. However, some workers might feel their privacy is invaded, especially if the wearables continue monitoring during breaks or in private areas like bathrooms. There's also a risk of the system misidentifying safe behaviour as unsafe, leading to unnecessary alerts or stress which could cause distractions at critical times. Additionally, the long-term data collection could be used to improve training, but workers may worry about how the data is stored or used.

5: Explains how the neural network supports the VPA AND includes one positive AND one negative impact. 4: Explains how the NN supports the VPA OR outlines how it supports the VPA and includes one positive AND one negative impact. 3: Outlines features of a neural network AND/OR impacts. 2: Outlines a feature of a neural network OR an impact, OR identifies features/impacts. 1: Some relevant information.

Neural networks recognise natural language patterns to categorise common support queries, such as lost luggage, and use nodes to assist with processing. They assess language tone to gauge frustration or urgency, allowing the virtual personal assistant to recommend specific phrasing for customer service staff. For instance, calming language may be suggested for distressed customers. By tracking customer ratings and turnaround times, the neural network uses weightings to refine recommendations to prioritise high-frequency queries or improve suggestions for poorly rated interactions. While this system streamlines support processes and improves response accuracy, it may also impact the role of customer service staff. As the neural network learns from interactions, staff may rely more on its recommendations, potentially reducing their familiarity with emerging query trends or response strategies. Customers may come to expect faster, more precise service, raising the standard for human interactions.

6: Explains how human bias could affect the usefulness of AI solutions AND supports with reference to the video. 5: Describes how, and refers to a scenario(s) in the video. 4: Outlines how human bias could affect AI solutions and refers to the video. 2-3: Identifies some sources/examples of human bias AND/OR effects on AI AND/OR outlines how AI learns. 1: Some relevant information.

AI develops its ability by analysing vast amounts of data. The quality of these data can be influenced by human bias, and in turn, affect the usefulness of AI solutions.

Human bias may be explicit or implicit. Explicit biases are intentional. For example, a researcher may favour male physicists and only collect pictures of male physicists for training an AI solution, as shown in the video. Implicit biases operate unconsciously and may be shaped by social and cultural factors. Whether the bias was explicit or implicit, the choice of male physicists would suggest to the AI solution that physicists are male.

Human bias can occur in various stages of the development of AI solutions. As shown in the photo library example, training data needs to be diverse and representative, otherwise AI solutions may produce inaccurate or unhelpful responses. The process of labelling training data may also introduce biases as people may have different interpretations of the same data. The video shows that if most people draw shoes in particular ways, certain types of shoes will not be recognised as a shoe.

3: Describes how ML algorithms can be used in data analysis. 2: Outlines some features of ML algorithms relevant to data analysis. 1: Some relevant information.

Machine learning algorithms can be trained with a subset of the data to automatically identify patterns and trends within large datasets. The algorithms can handle complex relationships and nonlinearities present in the data allowing for efficient and comprehensive analysis. Machine learning can also be used for predictions and forecasting. These models can learn from new data over time, continuously improving their analysis and predictive accuracy and performance.

1: Correctly determines the outcomes.

3: Draws a correctly simplified decision tree. 2: Simplified tree with most of the logic. 1: Some understanding of the requirement.

3: Shows thorough understanding of how the effects of human bias on training can be minimised. 2: Outlines some effects of human bias on ML OR features of training. 1: Some relevant information.

Training data is initially selected by humans, who may unintentionally introduce their own biases. To minimise this, ensure that training data is representative of the problem the ML model will address. Reviewing data by individuals from diverse backgrounds could identify and reduce potential biases, ensuring fairer and accurate model performance.

3: Compares linear regression and KNN. 2: Outlines features of one or both. 1: Some relevant information.

Linear regression and K-nearest neighbours (KNN) are both supervised learning algorithms that make predictions from labelled data. The key difference is how they handle data: linear regression assumes a linear relationship between input features and the target variable, while KNN makes predictions based on the closest 'K' training examples, allowing it to handle more complex, non-linear patterns. However, KNN can be more computationally expensive than linear regression and is usually slower.

8: Discusses the proposal with reference to ML in DevOps, RPA and BPA from the slideshow. 6-7: Outlines advantages AND disadvantages with reference to at least TWO of these areas. 4-5: Outlines some advantages and/or disadvantages with reference to ML in DevOps and/or RPA and/or BPA. 2-3: Some understanding of the features of the proposal. 1: Some relevant information.

QuidantCon's proposal for the company suggests leveraging machine learning (ML) and automation to bolster security measures. One key benefit is the automation of security testing and vulnerability scanning, facilitated by integrating ML-powered tools into the DevOps pipeline. This approach ensures continuous security testing and provides rapid feedback to developers, allowing for early detection of security issues and reducing the risk of vulnerabilities being exploited. The frequent and automated testing improves the overall security position and enhances compliance with industry standards. Additionally, ML algorithms can analyse vast amounts of data to detect patterns and anomalies, which is crucial for proactive threat identification. This capability enables the company to identify security risks proactively, learn from past incidents, and predict future attacks, thereby enhancing their overall security measures.

QuidantCon's proposal for the company also raises some concerns. One key issue is the need for high-quality data for ML algorithms to work effectively. If the data is inconsistent or incomplete, the ML models may not perform well, leading to false positives or missed threats. Another concern is the complexity of implementing ML solutions, which can be challenging and may require specialised skills and training. This complexity could result in integration challenges, higher costs, and delays. Additionally, ML systems can be biased, potentially affecting security decisions. Security and privacy are also critical, as handling sensitive data requires protection against unauthorised access and breaches. These concerns emphasise the need for a careful and balanced approach when considering ML and automation in security practices.

1: Correctly determines the outcomes.

3: Correctly simplified decision tree. 2: Simplified tree with most of the logic. 1: Some understanding of the requirement.

8: Informed judgement about the use of AI in our society, comprehensive understanding of its use and effects, supported by information from the slideshow. 6-7: Broad understanding of the use of AI and its advantages and disadvantages, with some slideshow information. 4-5: Sound understanding; outlines some advantages and/or disadvantages. 2-3: Some understanding of AI. 1: Some relevant information.