Infinity AI, the synthetic data generation startup, has recently raised $5M in a seed funding round led by Diana Kimball Berlin at Matrix to build faster AI models using synthetic data. The founders and operators from the companies like Tesla, Snorkel AI, and Google also participated in the round.
The company noticed that AI models are only as good as the data they have been trained on. So, data collection is one of the main challenges in making better AI models. According to Infinity AI’s studies, many data scientists spend 80% of their time gathering, organizing, and labeling AI training data. As a result, many AI projects do not lead to production.
According to Infinity, the training data collection problem can be solved using synthetic data. It allows users to upload a single real video on its platform and transform it into hundred of perfectly labeled synthetic videos.
Over the past two years, several companies have relied on synthetic data to solve the training data collection problem and enhance their AI and machine learning models, including Tesla, Amazon, and Microsoft.
Infinity AI stated that the accuracy of the AI model is directly correlated to its training data. The process of collecting real-world data is very time-consuming and expensive. After the data is collected, it has to be correctly classified and annotated before using it for training. Therefore, many organizations are moving towards synthetic data, especially if the data acquisition and training costs are limited.