Element AI, a company cofounded more than two years ago by deep learning expert Yoshua Bengio, is rolling out its first publicly available products for enterprise customers today.
The AI Enablement Tools and Insight Libraries help automate document reading and processing for various industries using computer vision, optical character recognition (OCR), and other techniques. Element AI products will be used to do things like enable port workers to make predictions about operations, help workers answer questions about internal operations using natural language, and assist cybersecurity analysts at the National Bank of Canada in the assessment and prioritization of potential threats to its online systems.
Element intends to create both systems that can be tailored to assist with decision-making processes in specific industries but also generic AI that works out of the box for people across several industries. AI for transportation and logistics workers are in the pipeline.
High level learnings from use of Element AI’s tools are shared across tools and industries, so that systems made for an insurance underwriter to automate document processing can be used for banks processing loans or in other industries.
June 5th: The AI Audit in NYC
Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
“You’ve got a lot of toolboxes out there that are generic. Our nut to crack is how to build an agent that understands claims in a generic way and then can be used across the board and learn how to interact with the underwriter and the fraud analyst etc,” Element AI CEO Jean-François Gagné told VentureBeat in a phone interview.
Element argues its AI is unique because it controls the entire software stack and is supported by a team of about 80 research scientists who focus on key machine learning challenges like causality learning, few-shot learning, and transfer learning.
“Owning every piece of the stack really gives us a huge advantage in terms of performance, but it’s still a long journey ahead. Let’s be honest: We’re early on in terms of delivering on all the promises of the vision of making true adaptive systems, but we’re pretty proud of the progress we’ve made in just a few years,” Gagné said.
Like Andrew Ng’s Landing AI, one of Element’s key pitches to businesses is its ability to deploy advanced AI using small amounts of data.
This is made possible in part through the use of synthetic data. For example, computer vision deployed by its document-processing AI uses transfer learning so a synthetic receipt more closely matches the characteristics of a receipt that’s been crumpled up and shoved in a person’s pocket in the real world.
Style transfer can be used to mimic artistic styles or create deepfakes, but it can also make businesses more robust and agile, Gagné argues.
“We hear a lot about fake news, which is like the downside, but there’s a humongous value in fake data. The ability to create high fidelity events and simulate them with lots of context is strengthening the ability to use advanced systems in a very small data environment,” he said.
Later generations of Element AI’s enterprise automation tools will incorporate features like how to detect and address bias found in AI models.
“Right now we don’t have necessarily all the reports that make the analysis of bias and all the different important things that we need to have in place to properly manage this, so it relies on data scientists for the moment,” he said.
Element AI was founded in October 2016 and raised a $102 million funding round in June 2017. The company is based in Montreal with offices in Singapore, London, South Korea, and Toronto.
