STEP 2: PROBLEM DEFINITION, EVALUATING AI/ML SUITABILITY

Does AI present a viable/suitable solution to the given problem?

 The context and environment in which an AI program or policy resides is key in determining its efficiency. For this reason, co-creating a viable and suitable AI solution must include the involvement and consideration of the entire ecosystem surrounding the problem. Local stakeholders must be involved in defining the problem and assessing whether AI is an appropriate and suitable solution in the given context. This can be done through an evaluation of alternatives to AI, an assessment of assets and liabilities, and through landscaping what has been done in the past in order to extract key learnings. During this step, it is crucial to understand the context, what is needed, and who the key stakeholders are. Refer to this content to explore tools and resources that can help determine whether AI presents a viable solution to a problem in a given context.

Please find below a legend of what can be found within the framework:

📚Resources - e.g. reports, articles, and case studies

🛠Tools - e.g. guidelines, frameworks and scorecards

🔗Links - e.g. online platforms, videos, hubs and databases

❌Gap analysis - tools or resources are currently missing

👥 List of stakeholders which should be included in the specific decision point

  • 👥 Local populations, end users, local government, academia and local universities.

    📚 UNESCO Indigenous protocols for AI - A paper outlining a set of defined principles and protocols to involve local indigenous populations in the design of AI solutions.

    📚 The Africa-Canada Artificial Intelligence and Data Innovation Consortium‍ (page 4) - Consortium that engages closely with local community leaders and policy-makers to co-develop research questions and solutions relevant to local needs.

    ❌ Resource assessments to determine whether AI is viable and sustainable over time, assessments of funding and technical expertise required to maintain and update solutions over time.

  • 👥 Academia and international organisations

    ❌Knowledge sharing platforms, hubs, or databases of use cases

  • 👥 Ecosystem actors, local government, local entrepreneurs and businesses, local and international software developers.

    🛠 USAID Scorecard: "Is ML Worth Trying Out as an Approach to Solving Our Problem? " (see 'Managing Machine Learning Projects' pg. 24) - guidelines and checklist to assess whether ML is a suitable solution to a given problem.

    🛠 Nethope and USAID: AI Suitability Toolkit - tool to increase NGOs' internal expertise and capacity to evaluate, develop, procure, and use AI /ML in their work to ensure that people in need are aware of the technologies that affect them and their communities. The toolkit includes materials on AI / ML capabilities and on how to evaluate suitability of AI/ML for programs and projects.

    🔗 ML Hub - open source platform making artificial intelligence (AI), machine learning (ML), and data science accessible to everyone

    ❌ Benchmark for performance of existing AI/ML programs.

  • 👥 Program developers, managers, implementers, suppliers.

    📚 Digital Impact Alliance’s 9 Principles for Digital Development - Guidelines designed to help digital development practitioners integrate established best practices into technology-enabled programs and help them succeed in applying digital technologies to development programs.

    🛠 Nesta: Collective Intelligence Design Playbook - Playbook designed by Nesta to help design and deliver a collective intelligence project. It helps understand how to harness the best ideas, information and insight to address a complex issue and introduces activities to orchestrate diverse groups of people, data and technology to achieve goals.

    ❌Step-by-step guidelines to define AI/ML project management and timeline.

Is AI/ML a feasible and desirable solution to implement in this context?