Disrupt Development

View Original

AI for Civil Society

AI for Civil Society

In this blog Erwin Fisser shares about the process of cocreating a workshop on practically using AI tools for NGO and civil society professionals. What steps did Disrupt Development take with their client CIVICUS to come to a workshop that answered the most important questions the communications team had at that point?

You have probably noticed that this ‘Artificial Intelligence’ thing has become quite a big deal over the past few years. Everywhere you look there seems to be a scientist, tech entrepreneur, journalist or futurist telling us about how it’s going to change our world. In good ways and not-so-good ways. While many of us are either dreaming about how AI will bring more equal opportunities and solve complex social issues or are having nightmares about how AI will make an end to privacy or become a tool for oppression, CIVICUS* also had a pragmatical question; ‘How can our communications department make use of AI right now?’. Together with the CIVICUS communications team, Disrupt Development consultants Irene Siaw and Erwin Fisser put together an online (the team is spread out across the globe) half days introductory workshop to explore this question.

What is AI really?
While at first sight it appears a pretty straightforward question, there is actually quite bit more to it. It actually already begins with the definition of Artificial Intelligence. What do we mean by it? Even in the scientific AI sector this debate has not been settled and this also showed in the outcomes of the survey we sent around to the CIVICUS team members. Unfortunately, we were not able to settle the debate once and for all. But we did come to the conclusion that even though during the workshop it would be useful to provide an overview of the different types of AI, we would zoom in on generative AI; AI tools that are able to create text, images, audio and video.

Opportunities or threats, which wins?
Obviously, for an NGO like CIVICUS there are many worries about potential risks attached to AI, like surveillance, increasing economic inequality or a concentration of power in the hands of non-democratic entities. By zooming in on generative AI, of which there are many tools and uses known and freely available, the workshop gained focus and it would actually become a workshop, instead of a lecture or a debate session. Not that generative AI does not have its own unique set of hopes and fears. It is well known that generative AI has bias issues, where its outcomes tend to be more favorable to the groups that currently hold more economic and political power. There are also worries about generative AI providing inaccurate outcomes, violating copyrights, or being an IT security threat. The question is then, do these threats and worries outweigh the potential benefits? In other words: should we be using generative AI at all?

Understanding instead of prohibiting
Instead of prohibiting the use of generative AI until all threats and dangers are known and contained, you can also take the route of helping people understand where these threats and dangers come from, which will provide them with knowledge to make better decisions themselves. So during the workshop, we gave a lot of attention to explaining how generative AI works. How is a Large Language Model (LLM) able to take a question (prompt) from a user and then come back with an output (in the form of content) that is so impressive that it make you think that ‘the machine understands you’? Obviously, it does not make sense to explain that in a technical way. The biggest challenge there is to decide, what does a professional working at and NGO or civil society organisation need to know, so that they understand enough of the technology behind the computer screen? So, for example, they don’t just know that they should always check ChatGPT’s answers (even though ChatGPT’s makers do everything they can to make you feel like it’s not necessary), but they know why they should always check. And, on the more positive side, when you understand how generative AI works, it helps to think of potential uses for it, without being dependent on, continuously outdated, lists of new tools that have been developed by others. So, thinking about potential uses during the workshop ourselves was a valuable part of the workshop that provided added insight but also deepened the learning experience.

Learning and doing
It’s one thing to tell people about how generative AI tools try to convince you that they are an actual person, it’s another thing to have them experience it. By having the team work with ChatGPT (as the best accessible tool with the most well-known user interface), trying out various assignments with different prompts and evaluating their experiences together, the theoretic part of the workshop came to life. It’s a bit like explaining how a really cool magic trick works and then seeing it performed before your eyes, you might have a better understanding of how it works, it’s still amazing.

After the workshop, we created a document for the team so they could go over what we had worked on together, but also share with other interested colleagues. We thought it was a great journey and we are confident that CIVICUS is a bit better equipped for the future.

 

*CIVICUS is a global alliance of civil society organisations and activists dedicated to strengthening citizen action and civil society throughout the world