With Great Power Comes Great Responsibility | Brad Pilkington | TEDxDoncaster

Summary

In August 2020, an algorithm was used to assign A-level grades in the UK without exams due to school closures, resulting in nearly 40% of students receiving lower grades than expected, especially affecting underprivileged students. The public outcry led to a reversal to teacher predictions within days. The incident highlighted systemic educational inequality, with biases in grading based on socio-economic status and ethnicity. The rush to revert the algorithm's decisions obscured the need for addressing these deeper societal issues.

Takeaways

• 📚 On August 13, 2020, an algorithm was used to allocate A-level grades in the UK due to school closures, resulting in almost 40% of students receiving lower grades than expected.
• 🌐 Students from underprivileged areas were disproportionately affected by the algorithm's grading system.
• 🔍 The public rejected the algorithm, but it may have exposed deeper issues of educational inequality.
• 💡 An algorithm is a set of instructions used to solve problems, often developed using statistical analysis.
• ⚖️ Statistical analysis is used across industries to make informed decisions based on historical data patterns.
• 🤖 The accuracy of algorithms depends on the quality of the data they're based on; incorrect or incomplete data can lead to flawed outcomes.
• 🧠 Human behavior can be predicted with enough data, but algorithms lack the ability to understand the nuances and context of human experiences.
• 📉 The algorithm likely had some accuracy but failed to account for outliers, which is critical in high-stakes situations like grading.
• 📈 There was a lack of trust in the algorithm, reflecting a societal discomfort with relying on technology for significant life decisions.
• 🏫 The quick reversal to teacher predictions after public outcry led to logistical issues for universities and students.
• 🔄 The algorithm's failure highlighted systemic biases in education, such as racial and socioeconomic disparities in grading.

Q & A

What was different about the A-level results in 2020?

In 2020, students did not take any exams due to school closures and disruptions caused by the pandemic. Instead, an experimental algorithm was used to allocate grades based on merit and school ranking.

What was the reported consequence of using the algorithm for A-level results?

It was reported that almost 40% of students received lower grades than expected, with students from underprivileged areas being disproportionately affected.

What was the public's reaction to the algorithm?

The public dismissed the algorithm, claiming it was wrong, and demanded that teachers' predictions be used instead.

What does the reviewer believe the algorithm actually highlighted?

The reviewer believes the algorithm highlighted an ugly truth about education inequality in the country.

What is an algorithm according to the script?

An algorithm is defined as a set of instructions that can be given to a computer to solve a problem.

How can statistical analysis be used to develop algorithms?

Statistical analysis can be used to develop algorithms by looking for patterns in past data to predict future outcomes.

What is the potential flaw in algorithms as mentioned in the script?

The potential flaw is that if the data used to develop the algorithm is wrong or incomplete, the algorithm may tell the wrong story or make inaccurate predictions.

Why is human input necessary in algorithm analysis?

Human input is necessary to detect issues in the data that a computer cannot, such as legitimate errors or incomplete data, to ensure the correct story is told.

What was the main issue with the A-level algorithm according to the reviewer?

The main issue was a combination of mismanaged expectations from students and a lack of trust in the algorithm, along with the algorithm potentially not accounting for all factors affecting human behavior.

What was the immediate consequence of reverting to teachers' predicted grades?

The immediate consequence was that some universities had to offer students large sums of money to defer their admission, and less popular universities experienced a significant drop in enrollment.

What did the reviewer learn about critical thinking from their first role as a data analyst?

The reviewer learned that critical thinking involves considering every consequence of every action before it happens, which allows for preventative rather than reactive actions.

Why does the reviewer argue that dismissing the algorithm was a mistake?

Dismissing the algorithm was a mistake because it allowed societal biases and education inequality to go unnoticed, as these issues were not caused by the algorithm but were instead highlighted by it.