The Future of Work Summit will feature CIOs, CTOs, and other C-level and senior execs on data and artificial intelligence strategies. You can learn more.
An Eastern European bank is teaming up with the government of Hungary to create a large language model of the Hungarian language with the help of an artificial intelligence supercomputer.
The government of Hungary gave half of the funding for the supercomputer developed under contract with SambaNova Systems, thanks to an agreement between the bank and the government. Péter Csnyi, deputy CEO and head of the digital division at OTP, said that the government will have access to the system for public and academic research.
The Generative Pretrained Transformer is a tool that SambaNova announced in October as dataflow-as-a-service. Marshall Choy, VP of product at SambaNova, said that building a system like this to run a GPT model is not something any bank has done before.
The bank decided to create applications for the Hungarian language in order to benefit from the digital economy. One of the core capabilities that we need to bring in-house is our ability to adapt to a changing world.
The Hungarian language is a hard language to learn and one that few people speak, so it is unlikely to get its own artificial intelligence model anytime soon. The experience of working on this project will give the bank the knowledge it needs to create language models for other countries in the region.
The artificial intelligence supercomputer is scheduled to go live in 2022.
There are nation-sized needs.
We have been observing the trends for a couple of years now, where larger and larger models require larger and larger resources to build. Choy said that they have productized this. The product from SambaNova allows enterprises to create their own GPT models, instead of relying on shared resources.
Péter and his team are going to be able to hire hundreds of data science and machine learning professionals thanks to us. He is going to be able to do this with a few people.
The GPT approach to creating large language models uses deep learning techniques to discover patterns in a language without being trained on them. The generative part allows the software to write its own rules based on analysis of large volumes of text content from that language. Large language models can speed up development and reduce labor required, but they also have pitfalls, such as a tendency to incorporate biases and falsehoods from the text they consume.
Many nations are nationalizing research into large language models for fear of being left behind, because a recent State of Artificial Intelligence report found large language models to be so significant.
Choy said businesses are concerned with keeping pace. The internet did similar things to what the artificial intelligence is going to do.
When members of his IT team first came to him with the idea, Csnyi thought the budget and talent required would be beyond the bank's reach. He said that SambaNova made this accessible at a reasonable cost. He thinks the bigger challenge will be preventing his people from devoting all their time to it and neglecting the bank's routine operational needs.
Csnyi said that natural language processing applications will be put to work for things like customer service, fraud prevention, loan origination, and cybersecurity, as well as purposes that may not become obvious until the technology is in production. There are no shortage of use cases.
The mission of VentureBeat is to be a digital town square for technical decision-makers to gain knowledge about technology and transact. Our site has information on data technologies and strategies to help you lead your organizations. To access, we invite you to become a member of our community.
Information on the subjects of interest to you is up-to-date.
Our newsletters.
discounted access to our prized events, such as Transform 2021, and gated thought-leader content.
Networking features are included.
Become a member.
