Uber AI Expands Model Training Resources
Source: pymnts.com
Uber AI Solutions Expansion
Uber AI Solutions, the artificial intelligence (AI) data services business of Uber Technologies, has introduced new solutions accessible to AI labs and enterprises across 30 countries.
According to a Friday (June 20) press release, Uber AI Solutions is providing other businesses with solutions developed over the past decade from Uber’s own use of data and AI.
Executive Insights
Megha Yethadka, general manager and head of Uber AI Solutions, stated that they are combining Uber’s platform, people, and AI systems to help organizations build smarter AI more quickly. She added that the platform is scaling globally to meet the increasing demand for reliable, real-world AI data.
Offerings
One solution from Uber AI Solutions is a platform connecting enterprises to global talent for annotation, translation, and editing of multilingual and multimodal content. This includes experts in coding, finance, law, science, and linguistics.
The company also provides datasets for training large AI models for generative AI, mapping, speech recognition, and other use cases. Additionally, they offer task flows, annotations, simulations, and multilingual support for training AI agents, as well as internal platforms for managing large-scale annotation projects and validating AI outputs.
The press release states that Uber AI Solutions aims to be the human intelligence layer for AI development, combining software, operational expertise, and global scale.
AI Industry Data Needs
PYMNTS reported in July that the AI industry has been experiencing a shortage of high-quality data for training AI models. Despite the large amounts of data generated daily, the required diverse, unbiased, and accurately labeled data is scarce.
In separate news, SandboxAQ launched a dataset on Wednesday (June 18) to help researchers advance AI models in drug discovery. This dataset, created using SandboxAQ’s AI large quantitative model capabilities and Nvidia’s development platform, allows researchers to train AI models to predict protein-ligand binding affinities faster than traditional methods.