Our latest data snapshot series has explored the current features and available data in CSET’s Private-sector AI-Related Activity Tracker (PARAT). We are excited to preview some new features and data that will be released in an expanded version of PARAT coming soon. PARAT was designed to enable users to identify artificial intelligence-related companies through different output metrics (e.g., research publications and patents). In the coming updates, we aim to improve a user’s ability to identify, measure, and compare AI-related companies with new and updated output metrics across a larger number of companies.
While PARAT includes a company’s counts of AI-related research publications and patents, we will add features to count these publications and patents by various subfields of AI and machine learning (ML). These AI/ML subfields will give users more insight into the specific areas a company might be focused on, or highlight how general its AI/ML output is. For example, does a company produce mainly Natural Language Processing research papers, or do they have a wide distribution of research papers across multiple fields, such as Robotics, Computer Vision, and High Performance Computing? Similarly, for AI-related patents we will include the count breakdown across AI/ML subfields, such as Language Processing, Speech Processing, and Computer Vision.
An updated version will also include companies’ AI publication citation counts. Citation counts can be a useful feature to sort companies, as well as to compare companies with similar research publication counts. The updated tool will also identify which PARAT companies are highly cited in other PARAT companies’ research publications to highlight connections between private sector AI-related research.
…the addition of workforce metrics will allow for different types of company identification and filtering in PARAT, extending the way in which we can measure a company’s AI-related activity.
Another addition will provide metrics on a company’s AI-related jobs. The tool will provide a count of AI-related jobs, building on some of the AI workforce categories outlined in recent CSET research, and a count of positions at the company focused specifically on AI development (e.g., ML engineers, research scientists). Since not all AI-related companies publish research or patents, the addition of workforce metrics will allow for different types of company identification and filtering in PARAT, extending the way in which we can measure a company’s AI-related activity.
As we continue our work toward a PARAT 2.0 we would love to hear from you! We welcome feedback on the current user experience with PARAT, and are also eager for any suggestions and comments on what is to come! Drop us a line at cset@georgetown.edu.