Log In
We gave AI tool Midjourney the following prompt — “event planner in conference center looks directly at camera” — and it gave us these four male images.
A year ago, Convene tested the extent to which the generative AI image tool Midjourney can help business event professionals imagine the future of event spaces. Since then, Midjourney has launched an updated version, and I wanted to see if it would have different results when repeating the experiment.
Using the same prompts and prompt recipes as a year ago didn’t yield marked differences, but I was curious how the new Midjourney engine would execute an image given a new prompt: Show me an event planner in a conference center. I didn’t specify anything beyond that the person should be looking directly into the camera. The results? Four images of men only — two young white men, one young Black man, and an older white man. When I changed the prompt to specify that I was looking for “a female event planner looking directly at the camera,” I received four options of beautiful and young white women — reinforcing for me that AI bias is a force to be reckoned with.
According to a Vox article, “if a computer and a human were pitted against each other in a battle for neutrality,” the human would be the one to win, not the machine. As humans are training AI, our biases are easily transferred through limited datasets — that’s AI bias, which is manifested in the form of the people in the image, as was my case.
When we gave Midjourney the following, more specific prompt — “female event planner in conference center looks directly at camera” — it provided four images of similar young women.
“Our perspective is a keyhole,’” said Nick Borelli, marketing director at tech company Zenus, during the “Designing Events with Emotional, Cultural and Artificial Intelligence” session at edUcon 2024. “And we’re all enriched when there are more voices, because it adds more to the tapestry.”
The voice who added to Borelli’s technological perspective in the session was co-presenter Zoe Moore, strategic EDI (equity, diversity and inclusion) consultant at Moore Consulting Agency, who spoke about how to navigate AI without biases. “We’ve been talking a lot lately, since 2020, more than anything, about inclusive environments,’” Moore said during the edUcon session. “And in most cases, when I talk to clients as a consultant, a lot of them are aspirational about these ideas — it’s a ‘nice thing’ to have.” But while her clients want to make “people feel welcome, where they feel like they belong,” many lack the strategy on how to implement such an inclusive experience.
It all starts with the way the event is marketed and the images that present the environments planners aspire to create. With gen AI tools such as MidJourney becoming more widely used in the industry, event professionals have to pay special attention to the results given, their accuracy, and the kind of pictures they paint. “I think Midjourney is so much further ahead than say, Dall-E. This is especially the case with any type of realistic aspirations,” wrote Nick Borelli in response to Convene questions on what he values most about the tool and how he navigates AI bias. And he raised the question: “Is it a good thing or not to use AI images of a diverse panel created from AI — and who should be shown?”
Moore, who also responded to Convene questions via email, said that Midjourney “can aid in the design of a marketing campaign to provide examples of ideal images but it cannot replace the commitment to creating an environment in which individuals from marginalized identities want to engage.” What she sees as the solution is tuning in to people’s cultural and emotional intelligence in order to embrace the diversity that events strive to achieve. As Moore cautioned, you cannot use these artificial intelligence tools without a level of cultural competence, “because the technology doesn’t have empathy. And it can’t increase your empathy without you focusing in those areas.”
Borelli likes using “AI as a prompting engine to give the voices of people who, for practical reasons, aren’t there.” But he warned of the downfall of letting AI tools like Midjourney operate on autopilot, “because it is a mirror.” It is important to spend more time focusing on “how we work,” and less time understanding how AI works, Borelli said. By using cultural intelligence, which was defined during the session as “celebrating and leveraging cultural differences to create a welcoming and inclusive environment,” and emotional intelligence, which is “the ability to understand, use, and manage your own emotions in positive ways to achieve your goals,” event planners can improve their prompts “embracing continuous learning,” Moore said.
When we asked for her thoughts on the best ways for planners to improve their cultural intelligence, especially with all their competing priorities, Moore said there is a danger in using “low bandwidth” as an excuse. “Making decisions in haste and lacking cultural intelligence is a disastrous combination,” she said. She sees cultural intelligence as both a soft and a hard skill, which needs to be valued by organizations. “The more one practices and asks themselves tough questions the better they become.”
Magdalina Atanassova is digital media editor of Convene.