Artificial Intelligence and TV Dramas
Previous Wired article that said that there are an increasing number of cases where science fiction writers are commissioned to write scenarios to determine the direction that companies and policymakers should take, and that this is starting to become a viable business. In order to predict what technologies will be needed in the future, I think it is important to get inspiration not only from the accumulation of current technologies but also from the world of fiction, including actual usage scenes.
In both contemporary literature and modern television — particularly Western dramas — the involvement of technology specialists in script development is increasingly prevalent. This collaboration often results in scenarios that, while imaginative, are grounded in realistic potential. Noteworthy examples from the realms of artificial intelligence and information technology include “Person of Interest” (2011–2016) and “Mr. Robot” (2015–2019), both of which showcased narratives that blend fiction with feasible technological advancements.
“Person of Interest” (POI) is a drama featuring two artificial intelligences, “Machine” and its adversary, “Samaritan.” Initially created to prevent terrorism, the “Machine” begins to identify ordinary criminal activities, prompting its creator, a billionaire, to leave the government and combat everyday crimes.
In reality, systems like PISTA, utilizing Semantic Web technology to integrate databases such as email and travel information for terrorism identification, have been developed. Additionally, Palantir, a US company, markets a system called GOTHAM, which uses ontology (also part of Semantic Web technology) for counter-terrorism and criminal investigations, achieving approximately $500 million in sales in 2020. This convergence of fiction and reality highlights the growing overlap between the two worlds.
The advanced technologies depicted in POI include:
- Data Analysis and Network Reconstruction: Utilizing machine learning to analyze voice, image, and video data within the infrastructure, converting them into network data that changes over time.
- Predictive Simulation: Employing the analyzed data to forecast future events through simulations.
- Voice-Based Human Interaction
- High-Capacity, High-Speed Processing: Achieved through distributed computing.
- Data Ultra-Compression
- Powerful Computer Viruses
The extraction of patterns from non-text information (1) is currently being implemented using deep neural networks (DNNs). Transforming this data into network data is within the realm of the Semantic Web. A significant challenge lies in asynchronous information processing when incorporating a time axis into the network. Processing network data that changes across different time axes and granularities cannot be done synchronously, raising the question of how to manage this effectively. Presently, there is no definitive solution.
I will discuss voice-based human interaction, high-capacity processing, data ultra-compression, and powerful computer viruses separately if the opportunity arises. However, the most intriguing challenge currently lies in predictive simulation (2).
Current machine learning technology requires vast amounts of training data to achieve highly accurate results. This presents a major challenge when applying artificial intelligence to real-world scenarios. Simulation technology offers a promising solution to this issue.
Simulation in machine learning encompasses various approaches, such as Generative Adversarial Networks (GANs), which adopt a DNN-like approach, and Bayesian inference, which follows a probabilistic method. I plan to explore these technologies in the future.
