As everything in our lives — including entertainment — turns digital, we’re starting to recognize the negative impact that digital consumption and technology have on the environment. For instance, Carbon Trust recently reported that streaming just one hour of the newest binge-worthy TV show requires the same amount of energy as boiling a kettle for six minutes. It might not sound like much, but it adds up quickly, especially as content consumption spiked in 2020 and our time spent engaging with digital media is expected to increase another nine minutes this year.
Businesses and consumers alike are looking for ways to mitigate their carbon footprint, and people aren’t just gaining awareness around sustainability when it comes to consumer and enterprise technology. What happens behind the scenes is just as, if not more, important.
Artificial intelligence (AI) has been instrumental in transforming a vast array of industries, especially influencer marketing, product placement and other forms of branded entertainment. Innovation in a wide application of predictive analytics, neural networks and custom algorithms has revolutionized these traditionally relationship-based marketing functions — delivering results and unlocking monetization opportunities for creators and brands.
However, the processing power required to drive those outcomes uses incredible amounts of energy, to the point where AI can be detrimental to the environment. On top of the environmental costs, running massive AI models is also monetarily expensive. Smaller organizations may not have the funds needed to keep up with the industry giants, creating a barrier that stifles innovation in our industry.
So how can we reconcile the need for AI to power the future of entertainment-based marketing with the drain it can have on environmental and economic resources? It might seem like a simple answer, but we need to dedicate time and talent to building more green, sustainable, accessible AI. That isn’t as easy of a task as it sounds, but it’s imperative for driving our industry forward while protecting our environment, especially as we face a growing climate crisis.
Using Less, Accomplishing More
Over the past decade, companies have competed with one another to build the most powerful AI systems with little regard to the energy cost associated with doing so. Over time, it has become the status quo to feed data into complex, iterative models that require massive amounts of computing power rather than creating more efficient models that don’t drain environmental resources. Recent studies have found that the power needed to sustain the top AI models has doubled every three and a half months and that training just one deep learning model can generate carbon dioxide emissions equal to the lifetime of five cars.
We’re at a turning point where we can’t be irresponsible with the tools we’re utilizing just to say that we have the fastest, strongest technology. Instead, the onus is on us to determine how we can produce better outcomes while dramatically reducing the computing power we’re using in our daily work. Although that sounds counterintuitive, more sustainable algorithms and models are actually more impactful, driving stronger output more quickly, and all while reducing the resources required to run the technology.
Think of it in terms of building a rocket ship — if the engineers can figure out how to make the ship more sustainable in terms of energy used, it can fly much further for a longer amount of time. The same principle applies to AI: If we can create models that run more efficiently, it alleviates the strain on resources and, in turn, brings down the economic costs of AI for smaller companies.
Product Placement, Influencer Marketing and the Creator Economy: The Future Frontier
Creating more sustainable AI is arguably the most important consideration for the rapidly growing influencer marketing, product placement and burgeoning creator economy.
In our experience, AI has become a mainstay in helping brands understand the content ecosystem and identify the marketing opportunities that are guaranteed to result in impressions, click-throughs, sales and brand loyalty. It’s virtually impossible to navigate entertainment without it.
At the same time, social media and entertainment generate an unfathomable amount of data each and every day. Evaluating all of the audio, video, post aesthetics, captions, likes, views, comments and more have historically required hoards of energy and computing power. We’ve reached a breaking point — we can’t just keep building bigger and bigger models through which to funnel data. It’s our obligation to innovate a new way of doing things.
Influencer marketing and product placement companies can lead the way in this new frontier, thanks to the ever-expanding data pool we have to work with. If we can restructure our AI models to more efficiently analyze the entertainment ecosystem while reducing our computational power, the same structures can be applied across industries.
At BEN Group, our first step toward green AI is to build small neural networks inspired by state-of-the-art architectures. We’ve focused on creating environmentally-friendly algorithms, and we’re already seeing breakthroughs in reducing the amount of energy and money it costs to run our technology. With an increased focus on creating more sustainable, green AI, we expect to develop even more innovative models that cut down on the costs of AI tools — both environmental and monetary.
We have an opportunity to do right by the planet and restructure our technology to both perform at a high level and conserve energy. In fact, it’s imperative, and I urge other technology companies to follow suit.