AI in Social Services: Balancing Promise with Peril

The use of AI in social services offers benefits like reduced backlogs and enhanced service delivery. However, it can also cause harm, perpetuate biases, and threaten privacy. Emerging tools and regulations aim to reduce AI-related harm. A trauma-informed approach and thorough evaluation are crucial for responsible AI usage.

In recent months, the implementation of AI systems within social services has spotlighted both its transformative capabilities and significant risks. A standout case involved a Victorian child protection worker using ChatGPT, which made an alarming misclassification of a 'doll' as an 'age-appropriate toy.'

As AI technology becomes more prevalent, its unchecked use can exacerbate existing societal inequities. This is evident in the detrimental effects of some recommender systems, which have shown bias in job advertisement distribution and alarming prenatal recommendations. Likewise, recognition and risk-assessment systems raise substantial concerns over privacy and discrimination.

Researchers advocate for a trauma-informed approach, underscoring the need for social services to evaluate AI systems judiciously. A newly developed toolkit aims to facilitate this, guiding service providers in safe and ethical AI usage.

(With inputs from agencies.)