We’re at the point now where the more forward-thinking teams and companies are starting to build and roll out their LLM-powered, branded chatbots. They’re trained on the company, product data and content and ready to converse with customers about their needs and interests.
From the discussions I’ve been having and witnessing, it’s all very exciting and brimming with potential. But, putting my old marketing hat back on, there’s one element that doesn’t quite sit right yet. For understandable reasons, many brands seem to be leaning towards human-like avatars for their interfaces. Thanks to amazing tech like Unreal Engine’s MetaHuman and a range of simpler-but-still-impressive online tools, a lot of these AIs are getting human faces. There’s nothing inherently wrong with that, but from a branding perspective, it’s a dropped ball.
A brand’s AI avatar is going to be a major brand asset. They are only going to get more important as the tech improves and usage grows. It requires careful thought and collaboration with the branding team to do right.
You know the playbook; make it ownable, identifiable, distinctive and consistent with your other assets. There have been some pretty good examples. OpenAI’s ChatGPT has a white blob that morphs between a circle when listening, a thought-bubble when thinking and a row of 4 stretching dots when speaking. Google, Microsoft and Amazon all have identifiable icons that presumably will rotate, bounce, blink and wiggle accordingly when their interfaces go live.
Non-human avatars are more versatile and give a brand more options for how they’re used and evolved over time. One question a brand should be asking themselves right now is how long they are willing to commit to their avatar and whether a short-lived one is the best idea. AI is improving at an extremely rapid rate after all.
I have a feeling the humanoid approach will get tired pretty soon, anyway. It’s definitely slightly awkward looking at one when it’s still, waiting for you to respond. They don’t work as well from a branding point of view, and they offer very little differentiation in the short term. Plus, there’s a more serious reason.
Some users will be fooled into thinking they are talking to a real person, and even if no harm is caused by that, it’s still an area of confusion brands should want to avoid. Plus, bad actors will definitely be using human-looking AI to commit fraud, so why would a brand even want to risk any perception that they are attempting to deceive people? No, a responsible brand is aware of humans’ natural inclination to anthropomorphise non-human entities and goes to efforts to demonstrate that they are not willing to exploit that.
My advice is to take a step back, think longer term and involve your brand team in designing your avatar. There’s a wealth of new possibilities and potential.
Written by Paul North, Head of AI at big group.