Implementing your first AI project

This guidance is for courts that have identified a specific AI project. Refer to our "Understanding your court's AI readiness" guide to learn about and navigate the entire AI maturity spectrum. 

Change management strategy

Change management is about helping people adapt to new ways of working while ensuring that transitions happen smoothly and successfully. Effective change management is vital to the success of any technology implementation project. Strong change management can help ensure that the new AI technology has stakeholder buy-in, improves the job experience for personnel, enhances the quality and efficiency of the court's work, and improves experiences for court users.

Key components of change management include: understanding people's needs; clear and open communication; providing training and support; addressing fears, concerns, and resistance; and measuring success and making adjustments.

A change management strategy should:

• Assess stakeholder readiness and impact
• Develop communication strategy and use participatory design
• Develop phased implementation plan
• Develop continuous improvement and optimization strategy

Discover more guidance for your change management strategy.

Project scope & resource assessment

A critical task in preparing to implement an AI project is analyzing the project's costs, as well as the benefits and savings that may result from the innovations. The first step in this analysis is articulating exactly what the new AI technology should accomplish. 

Project scope includes identifying the specific tasks that the AI system will perform, such as automating scheduling or summarizing documents, and clearly stating the expected outputs. The court should also establish specific metrics for project success. 

Project costs can be grouped into three broad categories: 

• Direct costs
• Indirect costs
• Intangible costs 

Benefits from AI implementation can take multiple forms. Many court systems expect increased productivity, as AI is well-suited to handle routine and repetitive tasks and can reduce error rates. AI may also allow for operations to scale without a corresponding increase in resources. AI innovation may also lead to improved service delivery, especially for underserved populations. 

In addition to examining the costs and benefits of AI integration, it is important to consider potential risks, including vendor lock-in, ethical concerns related to data usage, and the challenges posed by emergent behaviors from AI systems. 

Review examples of cost & benefit analyses.

Build or buy?

Court leaders must navigate complex decisions about whether to purchase commercial AI software or develop custom solutions in-house. The decision to build or buy an AI tool must balance the pros and cons of each approach. 

Generalist AI tasks are often better suited for buying pre-existing software solutions:

• Document redaction
• Transcription services
• Summarizing tools
• Calendar management and scheduling optimization
• Translation services

Specialist AI tasks may often be better suited for building custom solutions in-house. 

Some examples of use cases that fall under this category:

• Jury pool fairness analysis
• Workload forecasting for judges or clerks
• Case triaging
• Evaluating outcomes for equity

Consider build or buy and use our decision-making tool.

Vendor engagement & procurement

It is vital that courts carefully select vendors and craft procurement and contract terms that protect the court and court users.  The vetting and negotiation process must reflect the court's technical requirements and its legal, ethical, and operational mandates.

When comparing vendors and identifying those best suited to serve court needs, courts should consider:

• Vendor expertise
• Security & compliance
• Customizability & integration
• Scalability & performance
• Transparency & explainability
• Validation & measurable outcomes
• Ongoing development & innovation commitment
• Support & engagement
• Cost transparency

The following vendor behaviors should be regarded as warning signs not to proceed with the vendor: vague or evasive answers, inflexible solutions, and absence of a clear support plan.

When the court has vetted and selected an AI vendor, it is important to negotiate contract terms carefully to protect court data, operations, and legal interests. 

The following are some especially important terms to negotiate with care:

• Data ownership & use
• Intellectual property (IP) rights
• Compiled & source code
• Termination, audit & exit provisions
• Model drift & change management

Get an AI vendor licensing checklist and learn more about vendor relationships.

Download "Implementing the First AI Project"