Issue link: https://digital.copcomm.com/i/1521562
68 S U M M E R 2 0 2 4 I C A S Q U A R T E R L Y Not long ago, a friend shared a thought-provoking post. It told of how he had optimistically envisioned a future where technology liberated us from mundane manual labor, allowing us to focus on uniquely human creative functions and complex philosophical problem-solving. However, if the news was to be believed, reality was steering in another direction due to AI. AI: Is the Rise of the Machines Finally Here? b y K a r o l U r b a n C A S M P S E AI is simply technology that emulates human decision-making with machines and computers. Combining tech such as sensors, geolocation, robotics, etcetera, AI performs tasks that would otherwise require human intervention. Digital assistants and schedulers, GPS guidance, and self-driving modes in newer vehicles are examples of AI in our daily lives. AI is made possible by combining machine learning and deep learning, which allow "decisions" after "learning" from available data. The search data, decision-making skills, and success of those decisions ultimately allow these tools to make more accurate and useful decisions over time. AI, like humans, has the potential to improve itself over time with repetitive use. To explain it even further, UC Berkley defines machine learning, a type of computing science focused on data and algorithms that imitate how humans learn, into three main parts: decision process, error function, and model optimization. Machine learning views data with weighted preferences, compares its generated answer to an example determined as satisfactory, and then modifies its answer or 'model' until a threshold of satisfaction parameters is achieved. Deep learning, a subset of machine learning, uses neural networks to refine this process, elevating the complexity of the result. Similar to the human brain, neural networks contain nodes. Nodes contain an input layer, several secondary or hidden layers, and an output layer. If the output of an individual node is found to be acceptable, it is passed to the next layer of the network. If the output is not deemed acceptable, it goes no further, allowing for the classification of results in high velocity. This is what gives AI its impressive speed. This is what is currently used to make high-speed speech and image-recognition determinations in Google's famous search algorithm. But quick and accurate search results and GPS tools are not what put our creative, warm-blooded hearts aflutter. What is of specific concern to our work as mixers is generative AI. As a re-recording mixer, we are not generally annoyed to find there are better noise-reduction tools or even a quality background generator with the ability to add natural noise fluctuations to help patch some ADR. Our work is the craft of filmmaking. Ideally, we are principally tasked to carry out the unique voice and timbre of the director's vision. Spending more time tweaking the balance of elements and their subjective treatments and placement during sessions would be a wonderful service to this goal. No love is lost in gaining time back otherwise spent on removing hash and distortion if we still get to keep the time to experiment with the best timing for a jump scare or the entrance of a romantic music swell under the protagonist's kiss. The real issue is whether generative AI or derivative amalgamated art will be good enough for the consumer. Tools that claim to search and creatively modify sounds, create AI-generated music, create scripts, generate vocal perform- ances, and/or rebalance mixes are alarming. More than once in every mixer's career, we are forced to leave something we know we could ostensively correct or make more effective and impactful simply because time and funds are up and the powers that be have deemed the work done and satisfactory. Herein lies the danger, and its impact could be quite insidious. Generative AI does not just give us "answers" and "decisions"; it is a content generator or, more specifically, a derivative content generator. Arguably, so is humanity in many ways, but our data set, i.e., our life, is continually changing in infinite