As a class of technology centered on automating information, language AI is one of the most widely deployed forms of artificial intelligence across all industries and sectors, led in large part by the globalization potential of machine translation and solidified in the widespread adoption of tools like chatbots, document text recognition, and linguistic data-driven analytics. Regardless of the fundamental product or service a company provides, or the scope of its online market, automating exchanges with customers needs little explanation as a chance to scale and save costs, not to mention document processing and other linguistically intensive tasks. Nevertheless, some industries with an abundance of opportunity to enhance their communications continue to resist AI, for instance when a traditional service area with an established presence leaves companies with a lot to lose from the inherent risks of a major shift in their service model. Among these, the banking sector is one of the most stalwart, being generally hesitant to take its eyes off the money, so to speak. Now, thanks to the natural language processing algorithm GPT that features so often in this series, all of that may be starting to change for the financial world, with one group applying its deep learning capabilities in an enterprise language AI product called GPT Banking that helps financial groups gain insights and make quicker work of huge volumes of language-based data.
As TechTarget reports, SambaNova Systems rolled out GPT Banking in late February as a standalone platform that continues its earlier innovation with a language model component for its “Dataflow as a Service” business analytics tool. In other words, what was at first a general machine learning product for sectors including finance has taken a hard turn toward GPT’s potential to furnish an end-to-end solution for financial enterprises, in particular. GPT (and most famously GPT-3) is an incredibly powerful NLP/NLG mechanism that can answer unstructured prompts (i.e., random prompts) with cohesive statements that generally reflect its training on vast quantities of internet language date – the things people say in forums, and more, as custom-trained variations on the underlying technology emerge from licensing opportunities. With its adaptation of the model to finance-specific content, SambaNova continues a trend we noted earlier in the relatively innovative life sciences industry, with Microsoft’s work developing the algorithm for uses around linguistic medical data such as patient records. In doing so, it has won ground in an industry that while not innovation-averse, poses a less direct mandate for constant change and development.
TechTarget notes that one of the major challenges to using AI in finance is ensuring it is explainable – a clear mandate for anyone overseeing decisions over money, but one of the bigger challenges with AI in general. With outputs that don’t necessarily reflect the processes behind them, accountability for decision making can become complicated for institutions relying on algorithms. The report thus casts skepticism on whether this solution is readily deployable across the many areas it can potentially alleviate, such as client-facing language generation, claims transcription, and market intelligence reporting. Fortunately, LSPs like CSOFT already offer one of the more distant value proposals: translation for companies’ entire financial service and product ecosystems, or financial enterprise language management, in short. With custom translation solutions that adapt to the changing needs of technology-driven businesses, there may not be any one algorithm that banks can own to entrust with their multilingual communications today, but instead a considerably better solution in human-machine translation services that ensure banks and financial institutions never have to turn a blind eye on their global communications. From MTPE (machine translation post-editing) to software localization, learn more about CSOFT’s language AI, translation, and localization solutions at csoftintl.com!