Synthetaic’s AI Unravels Mysteries of Chinese “Spy” Balloon Incident
In a reminiscent nod to the Chinese “spy” balloon incident of 2023, where an airborne object of unknown origin traversed American airspace undetected before being intercepted by the U.S. Air Force, Synthetaic emerged as a pivotal player in unraveling its origins using satellite imagery.
This unexpected saga served as a potent demonstration of Synthetaic’s capabilities, catching the eye of investors, including defense giant Booz Allen Hamilton.
This week, Synthetaic secured $15 million in a Series B funding round, co-led by Lupa Systems and TitletownTech, a venture capital entity formed through collaboration between the Green Bay Packers and Microsoft. Notable participants in the round include IBM Ventures and Booz Allen Hamilton, propelling Synthetaic’s total funding to $32.5 million.
According to CEO Corey Jaskolski, the infusion of capital will fuel the rapid commercialization of Synthetaic’s computer vision technology, with plans to expand its workforce to 80 employees by year-end.
“The exponential growth of image data underscores the pressing need for advanced AI solutions to sift through and analyze this vast reservoir of information,” Jaskolski stated in an email interview with TechCrunch. “Synthetaic’s AI offerings in unsupervised learning and data analysis are strategically positioned to address the evolving tech landscape and cater to diverse industries, including defense, geospatial, video security, and drone-based monitoring.”
Jaskolski, a former technology director at National Geographic and an MIT graduate, epitomizes adventure, having explored Antarctica’s icy depths, plumbed the Titanic wreckage, orchestrated helicopter missions to map Mount Everest’s Nepalese face, and delved into flooded caves to document Maya sacrificial sites and Ice Age relics.
Image Credits: Synthetaic
What propelled an intrepid adventurer like Jaskolski to establish Synthetaic? According to him, it boiled down to a simple realization: AI, with its vast potential to categorize the world’s information, was hindered by the laborious need for manual data annotation.
“For a long time, the industry solution to this problem has been to literally have millions of people draw boxes on stuff and train AI, but what if we didn’t need human labeled data?” Jaskolski pondered.
Synthetaic, founded in 2019, introduced a solution called Rapid Automatic Image Categorization (RAIC), tailored to automate the analysis of large datasets, particularly satellite imagery and videos devoid of labels.
While conventional AI models rely on human annotators to label data, RAIC takes a distinct approach. Instead of relying on pre-annotated data, users feed RAIC a single image, and the tool identifies similar instances within a dataset.
For instance, during the Chinese balloon incident, Synthetaic’s platform effectively pinpointed the balloon’s location using nothing more than a sketch of its appearance and recent satellite images from the area where it was downed.
“RAIC means being able to handle scarce or complex datasets, accelerating AI development and improving predictive modeling without the constraints of data quantity or quality,” Jaskolski emphasized. “This positions RAIC as a strategic asset for driving innovation, operational efficiency, and competitive advantage, particularly in use cases where data is a bottleneck to AI adoption and implementation.”
Synthetaic is not alone in exploring synthetic data for model training. Companies like Synthesis AI, which secured $17 million in venture funding in April 2022, are pioneering platforms that generate synthetic data to train diverse AI systems.
Additionally, Scale AI offers a program enabling machine learning engineers to enrich real-world datasets with synthetic samples, while firms like Parallel Domain focus on crafting synthetic data for specific applications like autonomous driving.
Gartner’s forecast indicates that by 2024, 60% of the data utilized for AI and analytics projects will be synthetically generated. However, despite the industry’s progress, some experts express concerns about synthetic data’s potential drawbacks and dangers, urging careful consideration and evaluation of its implications.
Image Credits: Synthetaic
In a January 2020 study, researchers at Arizona State University revealed a troubling aspect of AI technology: an AI system trained on a dataset of professor images produced highly realistic faces, yet predominantly white and male. This outcome underscored the inherent biases present in the original dataset, which predominantly featured male and white professors.
Despite these risks, Synthetaic’s clientele has not shied away.
The startup asserts collaborations with diverse entities, including the U.S. Air Force for AI-powered object detection in geospatial data and The Nature Conservancy for identifying bird species previously believed to be extinct. Additionally, Synthetaic holds a contract with AFWERX, the Air Force research lab, to develop technology for object labeling, AI modeling, and object detection in satellite-captured images.
Jaskolski envisions RAIC’s applicability across various domains, ranging from AI prototyping to drone-based monitoring and content moderation. Citing Synthetaic’s endeavors with CNN to analyze war images from Gaza and its partnership with Planet Labs to provide analytics atop Earth imaging data, he emphasizes the company’s resilience amid tech industry fluctuations and broader economic challenges.
“Synthetaic’s technology offers a transformative approach to AI model training and creation, addressing the critical needs of technical decision-makers,” Jaskolski remarked. “For C-suite managers, Synthetaic’s RAIC means being able to handle scarce or complex datasets, accelerating AI development, and improving predictive modeling without the constraints of data quantity or quality. This positions RAIC as a strategic asset for driving innovation, operational efficiency, and competitive advantage, particularly in use cases where data is a bottleneck to AI adoption and implementation.”