It is not an easy task. When trying to convey a substantive message, there is always the risk of falling back on stereotypes, cliches, or jargon. The following dialogue summarises many of the dialogues that have taken place between Diplo’s communications and AI teams.
AI lab: We need some illustrations for the event in AI reporting.
Creative lab: Here it is….
AI lab: Hmm. We don’t really want to anthropomorphize AI. It can be problematic to ‘humanise’ AI.
Creative lab: Why? AI is regularly depicted in human form.
AI lab: We know. But it is almost fake news, or at least disinformation. AI does not have the free will and creative capacity of humans.
Creative lab: This is getting complicated. Everybody just draws robots. Ok. What do you want?
AI lab: A few years ago at the IGF in Berlin, we had a coffee machine as a depiction of AI. People were confused as they expected one of those ‘robots’….
Creative lab: Interesting. But this is not necessarily the best visual communication. Why don’t we ask AI to generate some images?
AI lab: Let me try
Creative lab: Not bad. But it has also given us a humanised image.
AI lab: It has. But it is wrong! The AI drawing algorithm relies on human input, which can be faulty.
Creative lab: People are always correct, especially when it comes to politics! (I’m joking).
AI lab: They are not. There is a long history of people (the majority of people) getting things wrong.
Creative lab: Ok. Let me prepare something. What do you think?
Creative lab: Or… something simpler…
AI lab: There is tension between AI and humans. It should be an interplay.
Creative lab: Is this interplay better?
AI lab: This one has a bit of tension. We often argue that humans and AI can coexist peacefully, but we don’t really know if that will be the case. It’s our aim, anyway…
Creative lab: This is getting complicated!
AI lab: It might be easier if we get together to figure out how we’re going to represent AI visually.
Creative lab: Beer later today?
This dialogue summarises many discussions we have had at Diplo. Would you like to join us for an online discussion on the problem of visualising AI, with all its historical, psychological, social, and cultural layers?
Register for updates on How should we visualise AI?