Oct 01, 2024 | AI

Over the past 30+ years, Thomson Reuters has brought together a diverse group of technology experts who specialize in legal, tax & accounting, and risk & fraud domains to explore cutting-edge technology and determine how best to apply it to real-world business needs. As a distinguished engineer with Thomson Reuters Labs – the dedicated applied research division of Thomson Reuters – I’ve had a front-row seat to witness the evolution of AI in solving customer problems.

In the 1990s, statistical and rules-based methods were state of the art. Over time, these evolved into more advanced modeling and machine learning techniques. Today, we leverage transformer-based, generative AI models. Regardless of the era, Thomson Reuters has consistently been at the forefront, figuring out how to best use these tools to serve our customers.

Below is a look at the Thomson Reuters Labs projects, collaborations, and breakthroughs that are an integral part of our approach to innovation. From pioneering research to practical applications, Labs’ role in shaping the future of information technology offers a fascinating glimpse into the synthesis of advanced technology and expert knowledge.

Who is Thomson Reuters Labs?

Thomson Reuters Labs is focused on the research, development, and application of artificial intelligence (AI) and emerging trends in technologies. We generate ideas and solutions to determine the art of possible, and in partnership with product engineers and other stakeholders, we deliver smarter and more valuable capabilities for our customers.

AI and Thomson Reuters Labs: A multi-decade partnership

The groundwork for much of what we are doing now with generative AI (GenAI) was established in the early 1990s with our work on Westlaw Is Natural (WIN), the first commercially available search engine with probabilistic rank retrieval.

We have also long employed ModelOps, an extension and adaptation of the DevOps principles for the AI ecosystem. The objective is to shorten the AI delivery lifecycle and ensure long term sustainability and quality levels of AI solutions through a combination of automation, continuous delivery and monitoring best practices.

Dedicated to end-users from the beginning

Trust is one of our most important values. Thomson Reuters Labs exists to deliver the best technology and tools that professionals can trust to make their work lives better. That’s why we always begin with the end. The end-user, that is.

We take a multidisciplinary approach to developing AI we call human-centric AI. This process involves recognizing the current and future needs of the professional user, starting with formulating a hypothesis to clearly define the problems, and then determining what successful outcomes would entail. Only then do we begin to build by creating and testing prototypes for viability, then iterate that process all the way through to final production.

“All our effort is to impact and meaningfully improve our products for the end-users to make their lives easier and their work better,” said Zahra Shekarchi, senior research engineer, Thomson Reuters Labs. “As much as AI is fantastic for making our lives easier and providing smarter solutions, it can influence us in other ways with unintended consequences. That’s why I have been following ethics in AI – privacy, bias and fairness, diversity, and social impacts.”

Our cross-functional approach to developing solutions

We believe we have created a culture that taps into the best of all possible worlds. We involve research and data scientists, engineers and designers at an early stage. And by having embedded teams we can move quickly, identify obstacles and opportunities early, and ensure a diversity of perspectives on problem-solving.

We combine skills such as machine learning, search and recommendation, and natural language processing with engineering acumen, design capabilities, and a human-centric approach all in the service of creating the best possible user experience.

“I work with talented teams in London, Toronto, and Switzerland,” said John Hudzina, lead research scientist, Thomson Reuters Labs. “We get different perspectives from each other, different experiences, and insight into different legal systems. The reason we thrive in the area of AI is first identifying the need or the problem, then decomposing it, solving all the pieces, and then bringing them all together.”

Innovation that never ceases

There is no such thing as “downtime” at Thomson Reuters Labs. We’re never satisfied that everything is finished. Our continuous pursuit of improvement has enabled us to produce since 2020. The work is challenging and fast-paced but rewarding when we find the answers that help our customers thrive.

“I believe the key to our success at Thomson Reuters Labs is that we embrace change and never stop learning,” said Mokarrom Hossain, senior research engineer, Thomson Reuters Labs. “AI is reshaping our world, so staying curious and continuously learning is key to not just keeping up but excelling.”

Our priority is to deliver smarter, more valuable tools for professionals. By combining deep customer knowledge with cutting-edge technology, we not only meet but anticipate the evolving demands of our customers. As we continue to explore new frontiers in AI and other emerging technologies, we are dedicated to making a meaningful impact on the professional landscape, helping customers thrive in an ever-changing world.

Follow our AI @ TR timeline to discover our innovation journey and learn more about our core pillars of AI and technology solutions.

This is a guest post from John Duprey, distinguished engineer, Thomson Reuters.