Strategically, the Human Services Summit positioned AI as a set of General Purpose Technologies that can simulate and perform complex analysis and decision-making. Technically, Summit attendees looked at AI as encompassing machine learning via data and algorithms, deep learning via neural networks, and generative intelligence, via contextual learning and adaptation. When put in practice, AI can be leveraged to increase efficiency in operations, improve effectiveness of service delivery, and interact with customers in new ways.

During the Summit, Sharad Goel, Professor of Public Policy at Harvard Kennedy School, presented an overview of the transformative power of AI. Following Goel's session participants explored how they, as human services leaders, should prepare to harness the potential of AI, while proactively navigating possible perils. During the Ideation Session, Summit participants reflected on ideas that emerged during summit sessions, shared their own experiences, and developed action steps human services leaders can take to judiciously and effectively adopt AI to improve capacity in human services.

More specifically, they broke into small groups to discuss the following questions:

• Where could advances in AI improve capacity in human services?
• What steps must we take to prepare our organizations to adopt AI
• What “guardrails” must we establish to leverage AI in a way that is ethical, effective, and secure?
• What messaging or narrative should we develop to accelerate adoption of AI?
• Where do you envision your organization in 5 years, as it relates to AI?

Below are highlights from those conversations.

Three Imperatives for Utilizing Artificial Intelligence

Below are three Imperatives, and their respective action items--as identified by Summit attendees--that practitioners can follow to achieve meaningful results with Artificial Intelligence immediately.

Imperative 1: Improving capacity in human services with AI.

1. Identify low-risk, high value areas to introduce AI.

   Begin to use AI in the low-risk, high value areas to help employees, clients, and others get comfortable. Examples include license application and oversight, budgeting, writing legislation, administrative rules, and policies, updating contact information, and developing workflow reports.

2. Free up the capacity of experts.

   Focus on areas that will enable staff to dedicate their time to more value-add work. For example, AI could do an initial draft of care reports, allowing staff to dedicate additional time to meeting with clients.

3. Design supportive AI to enhance customer-facing operations and interactions.

   Identify opportunities for AI to act as a co-pilot, aiding staff in delivering better and more customized services to customers. For example, AI could assist in eligibility determination, answering questions, applications completion, language translation, or matching community members with services.

4. Build new capacities to serve clients when, where, and how they prefer to be served.

   Offer 24/7, judgement-free self-service options to assist in areas such as addressing questions, providing service referrals, offering information on the benefits cliff and self-sufficiency, and helping through the intake process.

Imperative 2: Leveraging AI in a way that is ethical, effective, and secure.

1. Consider acceptable error rates and bias.

   Aim for better error rates and less bias than the humans currently doing the work, recognizing that AI will increase efficiencies and therefore have a greater impact.

2. Build in redundancies to protect against system failures.

   Develop a plan to ensure that systems will continue to function and operate if AI fails or goes down. And ensure AI is not used to automate outdated policies and practices, and initially focus on more transactional processes with AI serving as a “co-pilot.”

3. Create new policies to guide ethical use.

   Establish new polices around appropriate use, AI ethics, data permissions, content ownership, child protection, cyber security, and consent and disclosures pertaining to the use of AI.

4. Develop systemic guardrails and guidance at the community and program level.

   For example, at a systems-level a human services organization could decide that AI would never make a negative eligibility decision, and this approach could evolve over time for each community and program as more data is collected.

Imperative 3: Accelerating adoption of AI through messaging, narratives, and approaches

1. Engage the people most impacted in design and development.

   Allow the staff and customers who will be most impacted by AI to shape when and how it should be used and serve clients by introducing tools that are always available, easy to use, quick, and judgement-free.

2. Educate constituents about AI use in daily life.

   Inform employees, customers, partners and others about the applications and tools they are already using, which they may not realize are AI-generated.

3. Clarify that the aim is to support, not supplant decision-making by humans.

   AI can be used as a tool to inform, not decide. For example, it could assist with writing reports, helping staff prepare for a visit, teaching policies, and supporting the intake process.

4. Highlight the benefits for Human Services staff.

   AI can help to address worker shortages by augmenting staff capacity, reducing workloads, and allowing staff to focus on more rewarding and impactful tasks.

Generating Capacity through AI Infographic