Branding in the AI age - a guide to machine learning marketing

Branding for the AI field doesn't need to use the same old science - learn how top studios have bucked the trend in marketing machine learning to the masses.

"We didn't want the brand to feel cold or technocratic, and didn't want to rehash common visual tropes, like amorphous networks of dots and lines or weird Jude Law-ian robots."

Ritik Dholakia is talking to Digital Arts about the common visuals associated with the branding of companies in the artificial intelligence and machine learning field. Managing partner and founder of New York's Studio Rodrigo, Ritik had a chance to buck the trend with a recent branding project for Spell, a cloud-based platform offering individuals and organisations access to the AI and deep learning capabilities usually reserved for big corporations.

Working with Spell CEO Serkan Piantino, Ritik and team wanted to create a visual system that balanced technical and trustworthy qualities with approachability, all the while communicating the potential of machine learning to the uninitiated. The results can be seen in the floating pixel forms and Tron-like environments as seen on the Spell homepage.

"From purely a design perspective," Ritik explains, "we wanted to create a fundamentally approachable design language, that could both accommodate more playful tropes, like some slight visual references to 80s video games, with a more rigorous and precise language for the functional UI."

Studio Rodrigo's handiwork also extended beyond the visual and into the written.

"We tried hard to make the copy very plain language and non jargon-y, as open and inclusive as the visual references," says Ritik.

"So much of this approach derived from our collaboration with the Spell core team - who, for a bunch of super smart technologists, were committed to creating an approachable and inclusive visual identity that could help de-mystify AI."

Human but not humanoid

The demystification of AI drove the very human face to the marketing campaign behind DeepCube, creators of a deep learning framework designed to give solutions with greater ease.

Put on show in industry events around the US and Israel, digital signage advertising DeepCube registered the face of each and every observer using a combination of camera tech and AI. Thrown back at them was a unique logo to be then remembered each time the person returned to the signs; a logo for their eyes only, in other words, totally unique to the viewer and no other potential customer.

The design behind this deep learning feat came from New York/Seattle studio Rainfall, creatives who were also dissatisfied by the current state of branding for the AI industry.

"Brands for machine learning platforms are currently stuck in a clichéd world of brains, circuit boards, and atoms," as complains Marc Anderson, CEO and designer for Rainfall.

"I think that branding for machine learning needs to, and eventually will, become more abstracted much like identities for other industries, where more focus is placed on the outcome of the product depending on its intended use case. That said, I think that a little humanity in any identity is good, especially for machines and a future that many are scared of."

"For DeepCube the humanity in the identity was born out of some existential conversations we had in the studio around what it might mean if we gave a machine the ability to define its own personality," he continues.

"At the same time, we knew that the resulting 'personality' shouldn’t be humanoid in its visual form, as those experiments have already been done and are… terrifying."

The DeepCube logo generator was created in-house at Rainfall as opposed to DeepCube itself (as one might be forgiven for thinking), its development and design teams being used to UX and digital product design.

"We knew that we wanted a solution that was more akin to a digital experience than a physical 'logo' so the teams had a number of discussions about how technology could inform the actual output of the identity and ultimately the mark," Marc says.

"With many design projects we like to allow for the possibility of a unique result that is not specifically planned, and for DeepCube that meant designing a system in code through which the logo would be constructed, rather than designing the logo itself."

When asked whether other design studios can start thinking about making use of AI in their design work to make ambitious projects of their own, Marc points out how how bits of machine learning are already built into tools like Adobe’s Creative Suite, as powered by the Sensei framework.

"The tricky part is that despite the machine’s 'intelligence' it still requires a great deal of oversight to perform its task consistently well," he says.

"For DeepCube, that meant explicitly training the neural network model to consistently recognise inputs (in most case faces) and then performing a number of experiments to determine how well that trained model would operate when processing new individuals.

"The initial setup of the overall system was not difficult, but we had to set a very clear goal as to what we were “solving for” with deep learning, and what characterised a successful outcome."

No more boards, brains or bots

Both Rainfall and Studio Rodrigo have looked past the tired visual tropes that signify AI or 'complicated' technology in order to grab attention using different and interesting visuals - the spur behind all branding, in other words.

"Our most valuable takeaways," says Rodrigo's Ritik, "are looking past the more tired visual tropes that signify 'AI' or 'complicated technology' and trying to build an identity that gets to the mission of the company and also tries to be clear and specific about the benefits of the technology to the end users and buyers."

The DeepCube project though arguably blurs the line between branding and contemporary art, much like AI blurs the boundary between science and magic to most outsiders.

"DeepCube’s identity was motivated largely out of the client’s request that it communicate the dynamic nature of machine learning, where the platform is not a set of predefined rules, but instead a framework through which continuous input of information will result in more educated outcomes," Marc explains about the brief given over to Rainfall.

"Through several conversations we agreed that in its very nature the company of DeepCube was not 'static' - therefore it should not have a static mark or a 'logo' that was designed once and appeared exactly that way across all materials.

"The challenge for our design team was to combine that desire for continuous transformation with the obvious need for the mark to have enough uniformity that its consumers would recognise it as DeepCube regardless of what form it took.

"It is therefore a piece of art in the sense that we experimented with the medium to create a form of expression using digital and analog tools, but the edges of our canvas were the established rules of branding and a need for consistency," he stresses.

The field of AI may be a new and burgeoning field, but the same rules of branding apply to this wild frontier, then. That said, it wouldn't be too wild to let AI guide your designer eye.

"I view the relationship between deep learning and design as a framework through which implicit inputs can create explicit outputs (or vice versa)," Marc concludes.

"That is something that design studios can and should start looking into now if for no other reason than the possibility that an unpredictable result will lead to a creative breakthrough they might not have foreseen."

Read next: How AI is powering groundbreaking Photoshop and Premiere plugins

Comments

Elsewhere on IDG sites

Read Next...