AI – Thinking outside the bot

AI for knowledge producers and communicators

Two years ago, we began researching the potential use of AI in organizations that produce knowledge. We were the tech-savvy note at some conferences and events, the splash of color amid the things that truly matter.

We were met with curiosity and skepticism until ChatGPT appeared. That’s when everything changed. Suddenly, we understood that the change was serious. This wasn’t just a footnote in our lives, but life itself.

Now that the excitement is settling, at Sociopúblico we are moving from phase 1 (thinking) to phase 2 (prototyping).

The first tool we are building is an assistant for disseminating knowledge. For now, it only takes a more or less lengthy document – a policy brief, a technical note, a research book – and turns it into communication pieces for different channels, such as a press release, a blog post, social media updates, and a newsletter. It does this with three simple yet complex safeguards: 1) it doesn’t invent anything not in the original document, 2) it respects the voice of the organization endorsing the ideas, and 3) it optimizes the text according to the rules favored by each channel.

Achieving this is proving to be quite challenging, and it doesn’t always perform perfectly. However, the tool holds its own, and it teaches us.

Thanks to it, today we know much better than a few months ago how to achieve good automated communication results.

  • We know that brand manuals will have to change because AI doesn’t respect them. We are rewriting them for it, leading us to rethink what a brand manual is.
  • We know that for AI, distinguishing between instructions of form and content is very difficult. When we ask it to present an idea humbly, it says, “I present this brilliant idea humbly.” We are having quite a bit of fun with its mechanical literalness, and our prompts are gradually getting around it.
  • We know that for AI, explaining relationships between data with percentages (there are 20% more doctors in one country than in another) is easier than with ratios (“in one country, there are 3 doctors for every 20 children; in another, only one”). As the latter method sometimes communicates better, we are teaching it to use it.

In short, we are making progress.

Is it worth continuing to bang our heads against the wall until this comes out perfectly? Yes. Because today, organizations dealing with the most pressing public issues of our time also spend a lot of time communicating their ideas or explaining them to agencies like us.

That can change: with AI, we can process knowledge and translate it into thousands of communication pieces (and languages) in a short time, with a fraction of the resources we used to need. All that time and effort saved can go into more research, better thinking, and new ideas. The impact is immeasurable, for established organizations that will gain new resources and for smaller ones that couldn’t even afford a communication team before.

These days, we are going to Berlin to present what we’re doing at DGAP’s European Think Tank Conference, together with Wuppertal Institute, our partner to build and test these new ideas. We’re taking our MVPs with us to receive feedback and keep learning. 

Journeys are moments of transformation. This one has just begun.

More cases

Cornell UniversityGeorge Mason University

A platform to explore migrant rights around the world

Learn more

A new website for innovative journalism

Learn more