The growth of large language models is taking the RegTech industry by storm – and it may have only just scratched the surface of what is possible.
With the huge potential becoming ever more apparent, lying adjacent to this is the clear realisation that there is still much to be learnt about NLPs and LLMs.
Large language models (LLM) are a type of machine learning model that can perform a variety of natural language processing (NLP) tasks such as generating and classifying text, answering questions in a conversational manner, and translating text from one language to another.
A very relevant example of this OpenAI’s ChatGPT platform. ChatGPT is an AI-powered language model developed by OpenAI, capable of generating human-like text based on context and past conversations.
Will Basnett, head of advisory at Novatus Global, remarked, “As LLMs and NLPs pave the way for efficiency gains in areas such as automated document analysis or personalised customer communication, it’s clear that more in-depth education is key. Only with a solid understanding can businesses fully leverage these technologies, enhancing their workflows and ultimately, their bottom lines.”
Basnett stressed that given the security and ethical challenges posed by LLMs and NLPs – such as protecting user data and ensuring unbiased algorithms – the upscaling in educational initiatives is vital. “Equipping businesses with knowledge on these topics can transform these challenges into manageable risks and fuel responsible tech adoption,” he said.
The Novatus head explained that currently, the digital landscape is filled with companies that have misunderstood or underestimated the capabilities of LLMs and NLPs, often leading to missed opportunities.
How can these opportunities not be missed? Basnett believes that a rigorous educational programme can transform this trend, turning the complex language tasks that these technologies handle into sources of strategic advantage.
According to Basnett, the assertion that education around LLMs and NLPs is a luxury is ‘fundamentally flawed’. “In reality, it’s an investment towards future-proofing business operations. The more businesses understand about identifying and leveraging optimal use-cases of these technologies, the better their decision-making and growth potential become,” he explained.
The fear of the unknown often acts as a barrier when it comes to LLMs and NLPs, Basnett remarked. “However, through comprehensive education, these technologies can be demystified. Imagine harnessing NLP to analyse customer sentiment on social media or using LLM to generate relevant marketing content. These are practical, transformative possibilities education can illuminate,” he concluded.
In the opinion of Cognitive View CEO Dilip Mohapatra, meanwhile, the ChatGPT technology has created ‘huge awareness’ about AI and its applicability.
However, when it comes to enterprise adoption, Mohapatra said that many businesses are still trying to learn more about it and figuring out where to start. “Businesses definitely require more practical ways to understand the applicability of LLM and NLP in the enterprise context where security, privacy, and accuracy matter,” he underlined.
As the Gen AI technology continues to evolve, Mohapatra said that businesses can visualise individual roles and how each role can move from activity driven to value driven.
He explained, “Businesses can ask the right questions, for example, if I am a compliance or customer complaint resolution professional, how can AI support me in my repetitive and mundane tasks so that I can spend more time on creating value for our customers, and shareholders? What does my co-pilot look like? These questions will not only help in understanding AI better but will lead to new ideas to support blue ocean strategies.”
One of the big potentials that could come from LLM is its potential ability to improve the experience of the customer – the magnum opus of any successful business. Vall Herard, CEO of RegTech firm Saifr, underlined this point.
He said, “More education on the opportunities of large language models (LLM) is needed to help further grow adoption. Large Language models are a subset of foundation models (FMs) that make up the Generative AI ecosystem. To fully capitalize on the business growth possibilities, firms will need to understand both risks and rewards afforded by the entire ecosystem.
“Because LLMs, and more broadly generative AI, can help deliver better customer experiences at scale and at lower cost, business leaders have a responsibility to ensure their workforces are educated to take advantage of the efficiency gains from this new technology. Generative AI will come embedded in many business applications and companies aren’t proactively getting their workforce educated will fall behind.”
Challenges and risks
While more general education around LLMs appears to be a popular next road to take for many in the industry, Carlo Latasa – VP of engineering at Sigma Ratings – believes the mission-critical aspect of this process is understanding the challenges and risks with the application of LLMs.
“A simple oversight or assumption could get you in trouble – fast. Seek to fully understand how you will be using your model. What kinds of output do you need? Can you host the model or do you need to leverage Open AI – where ALL the input is sent to them? Clearly defining the application, inputs and outputs, as well as information flow is a great place to start,” he concluded.
In the opinion of Martina Rejsjo, head of product strategy at Eventus, the financial industry has dealt with rapid technological change before and, in many ways, is typically on the cutting edge of tech adoption. According to Rejsjo, we should expect that pattern to apply to LLMs and NLP and any other new application of AI.
She said, “In a regulated environment like financial markets, compliance teams will ask how these tools can be supervised and understood. Regulators have emphasized that licensed firms that use AI tools are ultimately responsible for meeting compliance requirements.
“In July, the Chair of the SEC stressed how important explainability is–that is, being able to describe why an AI model yielded a given output. The UK’s FCA and Bank of England, in a report about AI, likewise have said that explainability and transparency must remain part of financial compliance.”
Rejsjo emphasised that one innovative method is to combine rules-based parameters with machine learning models and automation, as Eventus does. “This helps compliance by showing the work–with a documented audit trail for regulators–and also helps analysts with improved efficiency so they can give their most tedious work to machines,” she concluded.