How 'conscious tinkering' can kickstart AI in your organization

How do you integrate AI to drive meaningful, sustainable impact? If you follow conventional wisdom, you might formulate an AI strategy, identify major AI use cases aligned with your goals, allocate ample resources, and start complex projects. This classic top-down model presumes several things: that management recognizes AI's potential, that viable use cases are clear, that employees are excited to change their workflows, and that the organizational infrastructure – governance processes, data management, and IT systems – are ready for these projects.

In reality, this is almost never the case. It certainly wasn’t the case at Ashoka. With offices in more than 30 countries and a network of approximately 4,000 Fellows, Ashoka is the world’s largest network of social entrepreneurs. After initial experiments in 2022, we created a dedicated AI Lab in September 2023. The motivation: don’t wait for an organizational AI strategy; instead, start tinkering to make an AI strategy possible.

So far, with two staff members, we have created 25 prototypes. These include obvious ideas like a chatbot with access to Ashoka’s knowledge and a tool to find collaborators among our global staff. It also includes unexpected apps, like a roleplay experience that asks colleagues to explain Ashoka’s vision to a group of 8-year-olds in a real-time conversation.

Today, AI applications save the organization time equivalent to several full-time employees, and a comprehensive AI strategy is starting to take form. We call our approach “conscious tinkering”. The name comes from the report AI for Impact: The PRISM Framework for Responsible AI in Social Innovation by the World Economic Forum that features our approach. We like the term and have adopted it internally and believe that it could be relevant for other organizations as well.

Conscious AI tinkering addresses risk aversion, limited resources, resistance to change, and a lack of AI strategy, by starting small and iteratively building on successes. A virtuous cycle of three simple steps allowed us to foster innovation and accelerate AI readiness without incurring large costs or risks.

Ideas for prototypes can emerge from any part of the organization. If a prototype does not generate the value we were hoping for, we only invested 40-80 hours to learn something. If it does, there is data to justify a bigger investment, and excited colleagues to make it happen.

The more prototypes we created, the more AI-ready Ashoka became. For example, an initial prototype might require the legal team to figure out a specific data privacy issue or the IT team to provide a new feature on a server. With these improvements in place, the next idea could be more demanding in terms of AI readiness than the previous one. Each cycle also encourages more teams to experiment with the technology, which helps with adoption.

Speed and pushing AI readiness is balanced by the “conscious” part of conscious tinkering. It emphasizes that prototypes must stay within the boundaries of what the organization can manage reliably and responsibly. If there are no governance processes to ensure legal and ethical compliance, you cannot create prototypes that process sensitive data. If the IT infrastructure is not robust, you cannot build an application that supports critical business processes. Prototypes should only slightly improve AI readiness in one or two areas while respecting current capabilities. This is how the approach generates fast innovation without causing catastrophes.

Our organizational culture makes this experimentation easy. As we work towards a world where everyone is a changemaker, we also select staff who embody this entrepreneurial spirit. Finding new ideas and collaborators is never a problem.

Experimentation implies that not every prototype justifies larger investments and management attention. A few of them do, though. Three different prototypes performed analyses against our database of social entrepreneurs and their innovations: one wrote benchmarks to help the evaluation process of new potential members of our network; another identified potential allies for any goal (“find social entrepreneurs who push policies against lead exposure in Asia”); a third curated media briefs for journalists who want to make their stories more constructive and solution-oriented. Each analysis saved a colleague a considerable amount of time – up to two full days of work in the case of benchmarks. We realized it was time to invest.

We engaged an external development studio to create a professional user interface for a unified application, upgraded our IT infrastructure to ensure the application was accessible to our global staff, and started the development of an ambitious monitoring system that automatically collects updated information on our 4,000 Fellows and their organizations.

For a network of social entrepreneurs, it might sound obvious that we should create a suite of AI tools to analyze data about our members. However, when we asked senior colleagues which AI applications would add the most value to the organization, this was not what we heard. Instead, the project emerged from small experiments with different teams that, unknowingly, were pointing in the same direction.

Conscious tinkering draws on well-known approaches like Agile, Lean Startup, and Design Thinking, which advocate starting small, testing ideas, and learning rapidly. It also echoes Kaizen, the Japanese philosophy of learning through small, continuous improvements. Perhaps the biggest inspiration came from community projects that promoted digital inclusion and co-developed solutions with future users. This made the solutions more relevant, helped with adoption, and ensured that users have a sense of ownership. Conscious tinkering adapts these principles specifically for AI, offering a flexible and responsible way to embrace innovation.

If you are an innovator or “intrapreneur”, don’t wait for management directives to start engaging with AI. Identify repetitive tasks and ask your IT teams to quickly build something for you. Or create a prototype yourself as a side-project (using platforms like make.com if you cannot code).

For managers, create an environment where tinkering is recognized and celebrated. Start hackathons, give innovators some time off their regular jobs to create prototypes, and showcase these efforts within your organization. Then use the golden nuggets to inform your strategy, help with change management and take them to the next level.