PSY205 Analyzing AI Bias: Prejudice in Language Technology and Its Impact in Singapore

Section A (95 marks)

Answer the question in this section.

Question 1

(Word Count: including in-text citations and excluding your reference list, shall fall between 1,800 to 2,000 words.)

First, examine the following excerpt from Hofmann et al.’s (2024) abstract to understand how their findings apply to the phenomena of prejudice:

Hundreds of millions of people now interact with language models, with uses ranging from serving as a writing aid to informing hiring decisions … Here, we demonstrate that language models embody covert racism in the form of dialect prejudice: we extend research showing that Americans hold raciolinguistic stereotypes about speakers of African American English and find that language models have the same prejudice, exhibiting covert stereotypes that are more negative than any human stereotypes about African Americans ever experimentally recorded … Language models are more likely to suggest that speakers of African American English be assigned less prestigious jobs, be convicted of crimes, and be sentenced to death.

Finally, we show that existing methods for alleviating racial bias in language models such as human feedback training do not mitigate the dialect prejudice, but can exacerbate the discrepancy between covert and overt stereotypes, by teaching language models to superficially conceal the racism that they maintain on a deeper level.

Develop your understanding of social psychology in the digital world as you apply theories of prejudice to analyse the extent to which biases of language technology, as uncovered by Hofmann et al. (2024), are relevant to the local Singapore context.

Additional writing tips:

For a more comprehensive essay, your analysis could also:

• discuss how sources of prejudice are relevant to raciolinguistic bias in AI.
• address the last sentence of Hofmann et al.’s abstract by proposing empirically-based recommendations to reduce AI’s dialect prejudice and to enable fair applications of language technology.

(95 marks)

References (5 marks)

The References lists the full reference of all sources cited or referred to in your assignment. In-text citations also need to be presented accurately in APA format.

• 5 marks: In-text citations and the reference list are presented in APA format with no errors. None of the sources cited in-text are missing from the reference list and vice-versa. A minimum of 5 academic sources are cited.
• 4 marks: In-text citations and the reference list are presented in APA format but there may be one or two minor stylistic errors. There may be sources cited in-text that are missing from the reference list and vice-versa. A minimum of 5 academic sources are cited.
• 3 marks: Formatting errors are evident in the in-text citations and/or the reference list. There may be sources cited in-text that are missing from the reference list and vice-versa. A minimum of 5 academic sources are cited.
• 2 marks: Major formatting errors are evident in the in-text citations and/or the reference list. There may be sources cited in-text that are missing from the reference list and vice-versa.
• 1 mark: Effort was made to present a reference list even though there may only be one item listed and it contains formatting errors; in-text citations may be missing.
• 0 marks: In-text citations and the reference list are missing.

English competency (Deduct up to 5 marks)

No marks are allocated to English competency, but a penalty of up to 5 marks will apply for errors.

• 1–2-mark penalty: Minor errors present, such as typos and spelling or punctuation mistakes.
• 3–5-mark penalty: Major errors or a fair number of minor errors present, such as spelling or punctuation mistakes, or use of colloquial and/or non-academic writing style.

The post PSY205 Analyzing AI Bias: Prejudice in Language Technology and Its Impact in Singapore appeared first on Singapore Assignment Help.