In this report, we’ll cover everything you need to know about LLMs: what the competitive landscape looks like, how these models can be applied to your work, what changes with the release of GPT-4, and considerations for responsible use. After taking stock of the present, we’ll also look at what the future of text might look like as LLMs become more sophisticated and tailored for specific uses across industries. In addition, we’ll share our prototype built using GPT and Slack’s API that’s designed to jumpstart brainstorms among a group of collaborators.

In this report, you will learn:

• Why LLMs are so effective
• How GPT-4 changes the game
• Ways to apply this technology now
• Important considerations to keep in mind
• The upcoming impact of LLMs and how to prepare

Brian: A creative co-collaborator brainstorm bot powered by chatGPT.

To make GPT more accessible across our team and to integrate it within commonly used workflows, we created a chatbot that integrates into our widely used communication tool, Slack. Unlike typical chatbots that only engage in one-on-one conversations with users, Brian is designed to actively contribute and facilitate brainstorming sessions among teams.

Ultimately, the chatbot streamlines team brainstorming and collaboration within the Slack platform by guiding the conversation, sharing its own ideas, summarizing the discussion and more.

Indeed, the efficiencies and opportunities that AI enables can radically uplevel brand experience and output, though unlocking their true potential relies on understanding how to uplevel teams to use the technology effectively. Those who can fully leverage the power of AI and infuse it within every aspect of their business will dominate the market. But for those lagging behind, this is a Kodak moment: there will be no loyalty for businesses that are slow to deliver AI-powered experiences that make consumers’ lives easier.

Throughout this guide, we’ll showcase AI’s potential to transform marketing today and tomorrow, as well as the actions you can take right now to reap those rewards and lead in the new era.

It’s true that we’re at the peak of the hype cycle with AI, particularly generative AI. But this disruption is a significant one, one that has become a make-or-break moment for many in our industry. 

We’re not just bullish on AI; as a change agent in our industry, it makes our model inevitable. Automation has played a great role in helping us scale up our business and outmaneuver our more traditional peers since day one. Meanwhile, today’s rapid evolution of AI is reshaping the nature of the brand-agency in real time, along with how brands themselves can go to market. Winning in the new era requires a willingness to embrace all that AI has to offer.

Prior to the hype, generative AI sprung onto the scene about four years ago, and our innovation team quickly began to experiment with it by training our own models. One turned lines and doodles into foliage; another neural animation tool created original dance choreography based on simple input like stick figures. More recently, we made an entire short film with the help of AI at every step of the production process. These experimental prototypes anticipated a future in which AI would open a new world of creative possibilities—that future is now.

But AI will have other, far-reaching potential across our industry as brands set out on new paths to growth beyond just content creation; its rapid development signals the start of truly personalized digital experiences. Cookies are crumbling, and brands have taken that as a cue to recognize the people behind the numbers and provide a better value exchange for their data. Yet the tools that marketers have historically relied on have always failed to turn the ideal of 1:1 marketing across the customer journey into a reality. AI finally unlocks that ability—across content, data and digital media, and tech services—to build and deliver hyper-personalized, highly empathetic customer journeys at scale, and fast.

Some are wary of these tools, focused on the balance between augmenting human ability versus replacing it. That’s a noble and important conversation to have. But automation has always been crucial to the growth of our team, and we are confident AI will continue to lead to more and greater work for our people. As a tech-agnostic team of makers, one that has already enthusiastically adopted these tools across every part of our business, we are the best positioned to take advantage of the abundance AI enables.

There are incredible opportunities on horizon: making more with less, making marketing intelligence more intelligent, or even driving new value to intellectual property through the training of owned AI systems. Each possibility will challenge the role of a creative, data and digital media, or tech services partner to evolve—lest they be surpassed by any number of new startups and cottage industries that have cropped up designed around AI. We don’t shy away from the challenge; we’re poised and ready to lead in the disruption, because that’s what we’ve always been built to do.

10. Digital humans get more realistic.

Digital humans may have earned a spot on our list of trends last year, but we haven’t grown tired of traversing the uncanny valley to play with the technology. In fact, the recent explosion of conversational AI will likely inject new life into digital humans and transform the realms of customer service, entertainment and more. Whether used to hand-craft original characters or refine scanned-in digital twins, digital human creation tools are becoming increasingly complex to deliver lifelike avatars. 

“We’ll see more competition between Unreal’s MetaHuman Creator and Unity’s Ziva,” says Geert Eichhorn, Innovation Director. In fact, Media.Monks has used Unreal’s tool to create a digital double of our APAC Chief Executive Officer, Michel de Rijk. Because why not?

9. Motion capture becomes more accessible.

Last year, we released a Labs Report dedicated to motion capture and how its increasing accessibility influenced content production for both professional film teams and everyday consumers. New technologies available at consumer price points are helping to bring motion capture into even more people’s hands. Meta’s Quest Pro headset, which released late last year, features impressive facial tracking that will be key to expressing the nuances of human emotion in VR. Move.ai, currently in beta, enables 1:1 motion tracking with a group of mobile devices—no bodysuits, no markers, no extra hardware needed. Using computer vision, the platform allows anyone to make motion capture video in any environment.

8. Mixed reality and mirror worlds mature.

With smaller and more comfortable AR headsets shown off already at CES, we can expect augmented and mixed reality to become more immersive, accessible and practical over the course of 2023 (check out more of what we saw at CES here). The VIVE Flow, for example, includes diopters so that users can replicate their prescription lenses in the device, amounting to a more comfortable experience overall. 

But it’s not just about hardware. “One of the major advancements is not in the headsets, but in the software,” says Eichhorn, noting that VPS has the power to pinpoint a user’s exact position and vantage point in the real world. “They do this positioning by comparing your camera view to a virtual, 3D version of the world, like Street View.” We covered mirror worlds in last year’s trend list, but the development of VPS is now bringing this vision closer to everyday consumers.

While VPS currently works only outdoors, we’ve already seen the power of the technology with Gorillaz performances in Times Square and Piccadilly Circus in December 2022.

7. More enterprises embrace the hybrid model.

For many businesses the return to the office hasn’t been a smooth transition; while some roles require close collaboration within a shared space, others enjoy more flexible setups that support childcare, offer privacy for focus work or greater accessibility. Given the benefits of flexible work setups and the development of technologies that build presence in virtual environments, Luis Guajardo Díaz, Creative Technologist, believes more enterprises will embrace the hybrid work model.

Media.Monks’ live broadcast team, for example, built a sophisticated network of cloud-based virtual machines hosted on AWS to enable people distributed around the world to produce live broadcasts and events. Born out of necessity during the pandemic, the workflow goes beyond bringing teams together—it’s designed to overcome some of the challenges traditional broadcast teams face on the ground, like outages or hardware malfunctions. It stands to show how hybrid models can help enhance the ways we work today.

6. Virtual production continues to impress.

Virtual production powered by real-time become popular in recent years: the beautiful environments of The Mandalorian or grungy urban landscape of The Matrix showed what was possible by integrating game engines in the production process, while pandemic lockdowns made the technology a necessity for teams who couldn’t shoot on location.

Now, further advancements in game engines and graphics processing offer a look inside the future of virtual production. Sander van der Vegte, VP Emerging Tech and R&D, points to Unreal’s Nanite, which allows for the optimization of raw 3D content in real time.

Localization of content is one example. “In 2023 we’re going to see this versatility in the localization of shoots, where one virtual production shoot can have different settings for different regions, all adapted post-shoot,” says Eichhorn.

5. TV streaming and broadcasts become more interactive.

With virtual production becoming even more powerful, TV and broadcasting will also evolve to become more interactive and immersive. “Translating live, filmed people into real-time models allows for many new creative possibilities,” says van der Vegt. “Imagine unlocking the power to be the cameraman for anything you are watching on TV.” 

It might sound like science fiction, but Sander’s vision isn’t far off. At this year’s CES, Sony demoed a platform that uses Hawk-Eye data to generate simulated sports replays. Users can freely control the virtual camera to view the action from any angle—and while not live, the demo illustrates the power of more immersive broadcasts. The technology could be a game changer for sports and televised events that let audiences feel like they’re part of the action.

4. Metaverse moves become more strategic.

“2021 was a peak hype year for the metaverse and Web3. 2022 was the year of major disillusionment,” says Javier Sancho, Project Manager. “There are plenty of reasons to believe that this was just an overinflated hype, but it’s a recurring pattern in tech history.” Indeed, a “trough of disillusionment” inevitably follows a peak in the hype cycle.

This year will challenge brands to think of where they fit within the metaverse—and how they can leverage the immersive technology to drive bottom-line value. Angelica Ortiz, Senior Creative Technologist, says the key to unlocking value in metaverse spaces is to think beyond one-time activations and instead fuel long-term customer journeys.

A great example of how brands are using Web3 in impactful ways is by transforming customer loyalty programs, like offering unique membership perks and gamified experiences. These programs reinforce how the Web3 ethos is evolving brand-customer relationships by turning consumers into active participants and collaborators.

3. Large language models keep the conversation flowing.

With so much interest in bots like ChatGPT, the Labs.Monks expect large language models (LLMs) will continue to impress as the year goes on. “Large Language Models (LLMs) are artificial intelligence tools that can read, summarize and translate texts, and generate sentences similar to how humans talk and write,” says Eichhorn. These models can hold humanlike conversations, answering complex questions and even writing programs. But these skills open a can of worms, especially in education when students can outsource their homework to a bot.

LLMs like GPT are only going to become more powerful, with GPT-4 soon to launch. But despite their impressive ability to understand and mimic human speech, inaccuracies in response still need to be worked out. “The results are not entirely trustworthy, so there’s plenty of challenges ahead,” says Eichhorn. “We expect many discussions over AI sentience this year, as the Turing Test is a measurement we’re going to leave behind.” In fact, Google’s LaMDA already triggered debates about sentience last year—so expect more to come. 

2. Generative AI paints the future of AI-assisted creativity.

If 2021 was the year of the metaverse, the breakout star of 2022 is generative AI in all its forms: creating copy, music, voiceovers and especially artwork. “Generative AI wasn’t on our list in 2022, although looking back it should have been,” says Eichhorn. “The writing was on the wall, and internally we’ve been working on machine learning and generating assets for years.” 

But while the technology has been embraced by some creatives and technologists, there’s also been some worry and pushback. “These new technologies are so disruptive that we see not only copywriters and illustrators feel threatened, but also major tech companies need to catch up to not become obsolete.” 

In response to these concerns, Ortiz anticipates a friendly middle ground where AI will be used to augment—not erase—human creativity. “With the increasing push back from artists, the industry will find strategic ways to optimize processes not cut jobs to improve workflows and let artists do more of what they love and less of what they don’t,” she says. Prior to the generative AI boom, Adobe integrated machine learning and artificial intelligence across its software with Adobe Sensei. More recently, they announced plans to sell AI-generated images on their stock photography platform.

1. The definition of “artist” or “creator” changes forever.

Perhaps the most significant trend we anticipate this year isn’t a tech trend; rather, it’s the effect that technology like generative AI and LLMs will have on artists, knowledge workers and society. 

With an abundance of AI-generated content, traditional works of art—illustrations, photographs and more—may lose some of their value. “But on the flip side, these tools let everyone become an artist, including those who were never able to create this kind of work before,” says Eichhorn. This can mean those who lack the training, sure, but it also means those with disabilities who have found particular creative fields to be inaccessible.

When everyone can be an artist, what does being an artist even mean? The new definition will lie in the skills that generative AI forces us to adopt. Working with generative AI doesn’t necessarily eliminate creative decision-making; rather, it changes what the creative process entails. New creative skills, like understanding how to prompt a generative AI for specific results, may reshape the role of the artist into something more akin to a director. 

Eichhorn compares these questions to the rise of digital cameras and Photoshop, both of which changed photography forever while making it more accessible. “The whole process will take many more years to settle in society, but we’ll likely see many discussions this year on what ‘craft’ really entails,” says Eichhorn.

That’s all, but we can expect a few surprises to emerge as the year goes on. Look out for more updates from the Labs.Monks, who regularly release reports, prototypes and podcast episodes that touch on the latest in digital tech, including some of the topics discussed above. Here’s to another year of innovation!

Since its launch in 2021, Epic Game’s MetaHuman Creator, a cloud-based app for developing digital humans, has extended its range of possibilities by adding new features—such as Mesh to MetaHuman. Using Unreal Engine, this plugin offers a new way to create a metahuman from a 3D character mesh, allowing developers to import scans of real people. In other words, it makes it easier to create a virtual double of yourself (or anyone else) almost immediately.

Inspired by this significant update and following our tradition of enhancing production workflows using Unreal Engine, our team of dedicated experts decided to build their own prototype. Needless to say, they learned a few things along the way—from the practical possibilities of metahumans to the technicalities of applying motion capture to them. As explained by the experts themselves, here’s what you need to know about creating and unlocking the full potential of virtual humans.

Be everywhere at once—at least virtually.

If you ever fantasized about cloning yourself to be able to comply with all your commitments or complete your pending tasks, metahumans may be just what you were looking for. Virtually, at least. As digital representatives of existing individuals, metahumans offer endless possibilities in terms of content creation, customer service, film and entertainment at large. Sure, they won’t be able to do your dishes—at least not yet—but if you happen to be a public figure or work with them, it’s a game changer. 

By lending likeness rights to their digital doubles, any influencer, celebrity, politician or sports superstar will be able to make simultaneous (digital) appearances and take on more commercial gigs without having to be on set. As John Paite, Chief Creative Officer of Media.Monks India, explains, “Celebrities could use their metahuman for social media posts or smaller advertising tasks that they usually wouldn’t have the availability for.” Similarly, brands collaborating with influencers and celebrities will no longer need to work around their busy schedules.

The truth is, virtual influencers are already a thing—albeit in the shape of fictional characters rather than digital doubles of existing humans. They form communities, partner with brands and are able to engage directly and simultaneously with millions of fans. Furthermore, they are not stuck in one place at a time nor do they operate under timezone constraints. In that regard, celebrities’ digital doubles combine the benefits of virtual humans with the appeal of a real person.

A new frontier of personalization and localization.

Because working with virtual humans can be more time-efficient than working with real humans, they offer valuable opportunities in terms of personalization and localization. Similarly to how we’ve been using Unreal Engine to deliver relevant creative at speed and scale, MetaHuman Creator takes localization to a new level. As Senior Designer Rika Guite says, “If a commercial features someone who is a celebrity in a specific region, for example, this technology makes it easy for the brand to replace them with someone who is better known in a different market, without having to return to set.” 

But not everything is about celebrities. Metahumans are poised to transform the educational landscape, too, as well as many others. “If you combine metahumans with AI, it becomes a powerhouse,” says Paite. “Soon enough, metahumans will be teaching personalized courses, and students will be able to access those at a lower price. We haven’t reached that level yet, but we’ll get there.”

For impeccable realism, the human touch is key.

To test how far metahumans are ready to go, our team scanned our APAC Chief Executive Officer, Michel de Rijk, using photogrammetry with Epic Games’ Reality Capture. This technique works with multiple photographs from different angles, lighting conditions and vantage points to truly capture the depth of each subject and build the base for a realistic metahuman mode. Then, we imported the geometry into MetaHuman Creator, which our 3D designers refined using the platform’s editing tools. 

“Because Mesh to Metahuman allows you to scan and import your real face, it’s much easier to create digital doubles of real people,” says our Unreal Engine Generalist Nida Arshia. That said, the input of an expert is still necessary to attain top-quality models. “Certain parts of the face, such as the mouth, can be more challenging. Some face structures are harder than others, too. If you want the metahuman to look truly realistic, it’s important to spend some time refining it.” 

Once we got our prototype as close to perfection as possible, we used FaceWare’s facial motion capture technology to unlock real-time facial animations. While FaceWare’s breadth of customization options made it our tool of choice for this particular model, different options are available depending on the budget, timeline and part of the body you want to animate. Unreal’s LiveLink, for example, offers a free version that allows you to use your phone and is easy to implement both real-time and pre-recorded applications, but focuses on facial animations only. Mocap suits with external cameras allow for full-body motion capture, but with mid-fidelity, and recording a real human in a dedicated mocap studio unlocks highly realistic animations for both face and body. 

At the same time, the environment we intend the metahuman to inhabit is worth considering, as the clothes, hair, body type and facial structure will all need to fit accordingly. Naturally, different software may adapt better to one style or another. 

While this technology is still incipient and requires some level of expertise, brands can begin to explore different ways to leverage metahumans and save time, money and resources in their content creation, customer service and entertainment efforts. Similarly, creators can start sharpening their skills and co-create alongside brands to expand the realm of possibilities. As Arshia says, “We must continue to push forward in our pursuit of realism by focusing on expanding the variety of skin tones, skin textures and features available so that we can build a future where everyone can be accurately represented.”

In this Labs report, we'll uncover how Generative AI is impacting digital creation today, and will explore how to keep ahead of where the tech is going next.

In this Labs report, you’ll:

• Learn what Generative AI is and what’s currently available
• Understand how the tech works
• See the technology in real-world action
• Peek into what the future holds
• Learn how to harness its power now

Generative AI supercharges production for higher quality creative output.

In this animated trailer for an original series called "Ancestor Saga," we demonstrate how Generative AI can be applied to film production. This prototype leverages Stable Diffusion AI for generating background scenes and custom Img2Img neural networks for AI-based rotoscoping of virtual characters.

See our findings about the benefits of using generative AI, including time and labor reduction in production, in the report.