by Vaughn Hannon, The Bezark Company, introduction by Adam Bezark.
The technological advances of the past few years have brought both promises and challenges to the LBE landscape. In this three-part mini-series, Adam Bezark and members of The Bezark Company team share their perspectives on how advances in generative “AI” and xR may shape our experiences and the tools we use to create them.
I had the pleasure of speaking at SATE EME 2024, where some of the industry’s best minds explored the challenges and promises of the future. Our industry works on generational time scales, and we’re tasked with not just responding to the latest developments but anticipating how they will play out years—decades—down the road.
Our creative technologist, Vaughn Hannon, has some thoughts on the seemingly overnight success of generative AI and shares his long view on this new tool’s true potential:
The rise of generative AI and machine-generated content
There’s a scramble within the creative community to understand the rapid rise of machine-generated content and what it means for the people who make a living crafting stories and building worlds.
Most people call it Artificial Intelligence (AI)—it’s not. Rather, the words, drawings, photographs, and songs that are being pumped out are the results of large language models (LLMs), and the creative world has been caught off guard by their sudden emergence, quick advancement, and seemingly boundless generative abilities.
The themed entertainment industry consists of some of the world’s most creative people, and this new development has, understandably, unnerved many of them. Every new technology that enters the mainstream brings with it a certain amount of fear, uncertainty, and doubt.
The machines know nothing and understand nothing but produce convincing and sometimes impressive material based on our input. They pump out images, music, video, and 3D models with the most minimal text prompts. This endlessly generated art seems to be getting better every week, and there’s real concern that the value of human creativity is going to plummet.
From Deep Blue to Transformer
In 2017, a handful of Google engineers released the Transformer architecture to the world. Seven years later, all of the latest text, image, audio, and video-generating machines are built upon Generative Pre-trained Transformers (GPTs) utilizing large language models to whip up content in seconds.
Twenty years before the advent of Transformer-based machine learning, IBM’s Deep Blue beat chess Grandmaster Gary Kasparov. A computer had beaten the best chess player in the world. The chess world reeled, assuming there was no point in humans playing any further. Don’t worry – people still play chess and now use these powerful machines to help develop new strategies.
In 2016, Deep Mind’s AlphaGo beat top player Lee Sedol in a series of Go matches. Go is a complicated game with an impossible number of possible moves. The Go world reeled at the human’s defeat. Don’t worry – people still play Go, and the machine-learning algorithms have taught us new strategies and even resurrected old strategies that were thought to be outdated.
Generative AI as a tool
It feels inevitable that machine-generated art and ideas will flood the world, but if we can learn anything from the Chess and Go communities, it’s that these machines are just tools. Like the printing press and desktop computers before them, they are assistive and empowering.
Remember, the machines are not thinking. They know nothing, but they are fast and can aid in ideation and prototyping in ways we haven’t seen before. Just as they have with each technological advancement, the landscape of work and career will change not only in the creative fields but across all industries as the possible applications for machine learning are wide-reaching.
Creatives should not fear the generative capabilities of machines but harness them. They help us fail fast so we can succeed sooner.
While it’s great for headlines and flashy news bits, generative art is the least interesting thing that will come out of all of this. There’s growing concern that the focus on LLM-based technologies is pulling resources from real advancement. There have already been major announcements and breakthroughs for protein folding, material discovery, molecular dynamics, medical imaging, understanding whale language, etc.
The ability to feed incredibly large datasets into these algorithms is a boon to the scientific community and should prove beneficial in the not-too-distant future.
Overcoming challenges
None of this comes without challenges. Jobs are going to shift as we adapt to these new tools. Energy consumption while training and running these models is a huge concern. We may very well be in the midst of another hype cycle, and these advancements that feel like huge leaps may hit an unforeseen barrier that stalls progress for another 10 years.
Techno-optimists see solutions coming to the energy problem. More efficient hardware and increased low-to-no impact energy generation might make this technology more sustainable. Whatever happens, companies should be proactive in educating employees about these available tools and how to use them effectively, securely, and responsibly.
Yes, the big players creating these LLMs have a lot to say about Artificial General Intelligence and the inevitability of the machines doing almost everything, but they need to pump up that inevitability to satisfy investors and markets. Ignore their bluster. This may be the beginning of another tectonic shift in human/computer interfacing, but we will all do well to focus on what’s available now and how to use these generative tools as another brush, instrument, or pencil, in our trusty and worn backpacks.
