Investment Banking
Artificial intelligence (AI) is expansive, complex and very much in demand. It’s also in the early innings of transforming global business, with many companies eager to understand its taxonomies and processes.
Baird took a deeper dive into this topic earlier this year to investigate AI’s impact on business today. We partnered with Dr. Yves-Alexandre de Montjoye, from the Department of Computing at Imperial College London, commissioned via Imperial Consultants, for in-depth discussions with nine companies across the globe that are using AI in various sectors and use cases. Its recent paper, An AI State of Mind, outlines Baird’s findings from these conversations.
To continue the AI discussion, Baird hosted a panel with industry leaders who are applying AI both in their businesses and with customers. Hosted by Simon Pearson, Managing Director in Baird’s Global Technology & Services Investment Banking group, the panel featured Eric Brabaender, Chief Product Officer, Empolis; Siobhan Hanna, Managing Director, AI data solutions, TELUS International; and Dr. Muhammad Shoaib, team leader AI/data science, Visable.
Quality is King“All AI needs high-quality training data to function… and that need is essentially infinite,” said Hanna. “The process of generating and enriching training data is really a ‘garbage in, garbage out’ proposition... you are the data you eat, and better data means better AI.” |
|
Leveraging All Data Sources“When we train the data and when we look at the data, [we] really have to see which information we can use to get the right training model,” said Braebender. Empolis leverages machine learning as well as knowledge-related data to build solutions. “We then analyze the relationships between all the information to find the right solution… [analyzing] the machine data that we can get out of the IoT systems, but also to connect it with systems or with semantic search.” |
|
Keeping Humans in the LoopDr. Shoaib highlighted the power of a human-in-the-loop approach to creating taxonomies and ontologies for very specific industries. “You have to use the same words, which are B2B-specific, and which are industry-specific... You can hire lots of ontologists or taxonomists, but it would be really challenging.” Visable leverage natural language processing pretrained models to cluster keywords and phrases in target languages to clean up the data before the internal team took over. |
|
Adding Value for Customers“You want to prove that you can find a needle in the haystack,” said Brabaender. Empolis generates special industry and machine-related data to ultimately create a knowledge graph model that reflects customers in a given industry – in turn empowering its customer success managers to effectively demonstrate the company’s value proposition onboard new customers sooner. |
|
The Accessibility OpportunitySpeaking about the AI application in making product experiences accessible to all, Hanna said, “I think we're really just at the tip of the iceberg, in terms of what technology and AI can do to improve product reach, but also improve and democratize product access. I think accessibility is an emerging space.” |
|
Off-the-Shelf on the Rise“I think that’s one of the amazing advancements in the recent 2-3 years,” said Dr. Shoaib, speaking about the rise of preconfigured and off-the-shelf models for training data, and affirming users may need to fine tune the data inputs according to their use case. Hanna built on this observation, sharing, “I think there is room for all of it, and that the demand is so extensively insatiable that we will need to continue to see growth in that off-the-shelf availability.” |
