How Performance Art Used AI to Highlight the Ongoing War in Ukraine
Just recently, humanity passed the one-year anniversary of the Russian invasion of Ukraine. It’s a grim landmark, and one which many wish had never come to pass. In lieu of being able to outright stop the fighting, many brands and agencies found ways to draw attention to the continued need to support Ukraine, creating campaigns and poignant reminders that peace has not yet been found.
Among this count was Performance Art, who, after discovering that AI like the enormously popular ChatGPT was trained on pre-September 2021 data, saw the perfect opportunity to draw attention to both the support effort, and the existence of data bias.
Armed with this insight, the agency, in partnership with the Humanitarian Coalition, created ‘A Prompt for Ukraine’, a 90-second video which depicts a conversation with a ChatGPT-like AI. Showcasing an interaction with the bot in which an unknown user asks questions about Ukraine, the bot provides jarring, outdated responses when describing some of the country’s now war-torn locations, like the Irpin Bridge and the city of Mariupol.
Serving as a poignant reminder of how the war has transformed Ukraine, the video finishes by prompting viewers to donate in order to make a difference, and has since launched across digital and social media platforms, supported by the hashtag, ‘#APromptForUkraine’.
LBB’s Josh Neufeldt sat down with Performance Art CCO Ian Mackenzie to discuss how this campaign came to life.
LBB> Why was an AI-related approach the best way to draw attention to the one-year anniversary of the Ukraine war?
Ian> Imagine you could have a conversation with someone who doesn’t know anything that’s happened in the past 18 months. That’s pretty much the situation we’re in with ChatGPT. Given natural language processing’s high position in the news cycle, and this time-based data blind-spot, it was quickly the winning idea. Our clients at the Humanitarian Coalition got behind it in a big way, and we just pushed it forward with haste.
LBB> The realisation that ChatGPT’s AI language model uses pre-2021 data is an interesting one. Was this a key information point that influenced the way the project developed? Tell us about the creative process!
Ian> This is a great example of the difference between an observation and an insight. The creative team (Jackson Kemp and Ben Playford) observed - through using it - that ChatGPT’s knowledge cut-off was September 2021 - five months before the war in Ukraine started. That’s the observation. Their insight is higher order: ‘Sometimes, what’s not there tells you more than what is’. That said, this is one of those ideas that emerged more-or-less fully formed. The job then was to make it happen, and fast. From inception to in-market was less than a week.
LBB> Let’s talk about the spot itself. What was the writing like for the prompt questions you asked? Were there any others that didn’t make the final spot?
Ian> We started by asking some general questions about ‘the beauty of Ukraine’ and its people. We found those responses to be interesting, but thought there was more tension in asking it specific questions, which we could then counterpoint with specific images. For example, we asked about a specific bridge over the Irpin River we knew had been destroyed. This created a clear line between what the chatbot knew, and what has actually happened during that time.
LBB> The spot depicts a conversation with a ChatGPT-like bot. As such, what did it take to achieve this interaction, and as a whole, how long did it take?
Ian> The entire production came together lightning fast. We developed the storyboard and script in tandem, and had an editor (Fi Alverez at School Editing) working in real time as we developed the creative with our clients at the Humanitarian Coalition. The actual prompts and responses in the spot are verbatim from real interactions we had with the platform.
Worth noting, these AI chatbots aren’t Google. They don’t serve up the same response every time, and they’re building a conversation through context of what’s come before in the conversation. What was crystal clear, however, is that it didn’t know, or wasn’t prepared to speak to anything about the war in Ukraine.
LBB> The photos are very poignant. Where did you find them, and did you write the questions around the photos, or search for the photos after writing the questions?
Ian> We feel the same way. There is so much powerful and profound war photography from journalists who are putting their lives on the line to tell the story of this war. We didn’t have a lot of time or budget, so we weren’t able to use rights-managed editorial imagery. Instead, we worked with photography available on royalty-free platforms, most of which didn’t include people in the photography… which turned out to be a bit of a creative gift. Images of war-ravaged Ukraine without people in them underscore our insight: ‘sometimes what’s not there tells you more than what is’.
LBB> The music in the spot really adds to the feel of the campaign. Who did you work with on this aspect?
Ian> Creativity, once again, loves constraints. With so little time to get to market, we went with our gut, found a stock track that delivered on the key story beats, and then worked with our long-time collaborators at Grayson Matthews Group to fine tune the timing and sound design. Our editor did a great job of working with us to sync impactful moments in the track to the video’s main story beats. The result is powerful and emotive.
LBB> Do you have any memorable lessons learned from the making of this new campaign?
Ian> As an agency with a heavy creative data bias, much of our most significant work involves platform builds, data/tech integrations, and in general, has lots of moving pieces – all of which takes time. Sometimes, it takes years. ‘A Prompt for Ukraine’ is a creative data idea, but made fast. That’s been good learning for our team.
LBB> As a whole, what has the initial response been like?
Ian> The early response has been great. For a relatively tech-y, AI data bias-based idea, the message seems to be widely understood, appreciated, and felt. We’re hoping it contributes to fighting compassion fatigue as the war drags into its second year. We know that every dollar donated to the Humanitarian Coalition through this campaign will be put to good use in the region, providing safe transport, shelter, food, hygiene kits, basic sanitation and more. Please consider donating here.
LBB> What did the chance to be involved in this project mean to you?
Ian> We’ve heard from members of the Ukrainian community that this work resonates and is appreciated. That means a lot. And, when you see the relief work that happens under the Humanitarian Coalition’s mandate… it’s just a huge privilege to participate in this small way.
Also, data bias is one of the defining issues of our time, and will only become more so as AI platforms push to the centre of our experiences. Of course, biases aren’t necessarily good or bad - they just are. What is bad is not recognising them. Anything we can do to help deepen the conversation on this topic means we’re making a positive contribution. Proud of the team and this project for that.
LBB> Just passing the war’s 1-year anniversary, is there anything you’d like to say?
Ian> Donations make a difference. Please give what you can at https://www.humanitariancoalition.ca/ukraine-crisis.
_jpg.webp)
