According to Arturo Devesa, chief AI architect and head of innovation at EXL, industry-specific Large language models (LLMs) are expected to drive innovation as AI deployment continues.
EXL, a global analytics and digital solutions firm, recently launched its first in-house LLM and its first LLM specific tothe insurance industry.
"Field-specific or company-specific LLMs: I think that will continue to be the focus of innovation in LLMs. Of course, there will continue to be innovation among general-purpose LLMs. Maybe they will be commercialized, but it will still be a huge business. But we're a little bit different. We're building something more specific," says Devessa.
EXL's insurance law professionals are trained in Life insurance audits, Property & Casualty insurance claims, Homeowners insurance property appraisals and the review of medical records for health insurance.The company is also working to expand this expertise into other areas of insurance.
They recently launched EXLerate.AI – a platform using carefully trained language models.The goal is to help businesses rethink their work processes.The platform enables seamless integration of EXL with AI tools from other companies.All of this fits into customers' existing business processes.
Devesa says the demand for industry-specific LLM programs continues to grow as more insurance companies look to incorporate AI solutions that can offer better accuracy, speed and cost savings than generic AI tools.
What is an LLM?
A big language model is an artificial intelligence tool designed to process natural language and respond to natural language as well.Devesa began his career in natural language processing.He explains that these language models are pre-trained with "vast amounts of data" from the Internet.Hence their name "large language patterns".
"They pre-train a model to understand whole words on the Internet, and then train those models on specific tasks like reasoning, code conversion, code writing, summarization, question and answer, ...
According to Arturo Devesa, chief AI architect and head of innovation at EXL, industry-specific Large language models (LLMs) are expected to drive innovation as AI deployment continues.
EXL, a global analytics and digital solutions firm, recently launched its first in-house LLM and its first LLM specific tothe insurance industry.
"Field-specific or company-specific LLMs: I think that will continue to be the focus of innovation in LLMs. Of course, there will continue to be innovation among general-purpose LLMs. Maybe they will be commercialized, but it will still be a huge business. But we're a little bit different. We're building something more specific," says Devessa.
EXL's insurance law professionals are trained in Life insurance audits, Property & Casualty insurance claims, Homeowners insurance property appraisals and the review of medical records for health insurance.The company is also working to expand this expertise into other areas of insurance.
They recently launched EXLerate.AI – a platform using carefully trained language models.The goal is to help businesses rethink their work processes.The platform enables seamless integration of EXL with AI tools from other companies.All of this fits into customers' existing business processes.
Devesa says the demand for industry-specific LLM programs continues to grow as more insurance companies look to incorporate AI solutions that can offer better accuracy, speed and cost savings than generic AI tools.
What is an LLM?
A big language model is an artificial intelligence tool designed to process natural language and respond to natural language as well.Devesa began his career in natural language processing.He explains that these language models are pre-trained with "vast amounts of data" from the Internet.Hence their name "large language patterns".
"They pre-train a model to understand whole words on the Internet, and then train those models on specific tasks like reasoning, code conversion, code writing, summarization, question and answer, classification, word extraction, etc.," he says.
The process by which LLMs take information in human format and provide a response is known as Retrieval-augmented generation (RAG).The AI first digs into a document or database to extract information related to the user's question, then enters that data into a language model as context, and finally generates an answer based on that.
LLMs get better and "smarter" as they develop over the years.But it wasn't until ChatGPT "took hold of the mainstream" in 2022 that corporations began to look at LLMs as practical tools and how to use them with their own data, notes Devesa.
However, companies face three main challenges with general purpose LLMs:
- Inaccurate answers ("artificial intelligence hallucinations");
- Slow responses;
- High costs.
Field-specific LLMs can help address these challenges.
Why does insurance need its own LLM?
Devesa explains that an industry like insurance, which can be complex and span a wide range of disciplines, could benefit from an LLM specifically designed for its purposes.Such a model not only helps to improve accuracy and eliminate „hallucinations“, but is also the most cost-effective and efficient as it uses industry-specific data.
"This is useful for our customers using their specific LLM, or even for customers who want to hire us to help them build their own LLM with their own data," says Devesa.
Devesa tells how EXL started with a general-purpose LLM, then refined it using industry-specific data and training large teams of operators, data scientists and AI engineers in their offices around the world.
"The theory is that fine-tuning will always outperform RAG because you're taking the knowledge of your use case and building it into the LLM."
The LLM never saw the context.The LLM never saw the assignment.Therefore, it outperforms the basic LLM because it is no longer a pre-trained model,” says Devesa.
Field-specific LLM models were sought
According to Devesa, a growing number of clients are looking to AI solution providers like EXL for industry-specific LLMs.He noted that despite the proliferation of general-purpose LLMs, none of them "are going to start hiring armies of insurance experts and building potentially hundreds of LLMs for different fields."
"In our case, we're already building several of them. It's very challenging and requires domain expertise and specific infrastructure. For us, that's really important. It's not a consumer product. Customers are starting to say they're more interested in our LLMs. They want domain-specific LLMs—enterprise LLMs. That means they want to use LLMs with their own data. They don't want to do RAG with GPT-4," he says.
EXL Serviceis a global data analytics and digital solutions company based in New York.Founded in 1999, the company has a team of 1,500 data processing professionals and operates across a wide range of industries.
Download the app
Download the app
According to Arturo Devesa, chief AI architect and head of innovation at EXL, industry-specific Large language models (LLMs) are expected to drive innovation as AI deployment continues.
EXL, a global analytics and digital solutions firm, recently launched its first in-house LLM and its first LLM specific tothe insurance industry.
"Field-specific or company-specific LLMs: I think that will continue to be the focus of innovation in LLMs. Of course, there will continue to be innovation among general-purpose LLMs. Maybe they will be commercialized, but it will still be a huge business. But we're a little bit different. We're building something more specific," says Devessa.
EXL's insurance law professionals are trained in Life insurance audits, Property & Casualty insurance claims, Homeowners insurance property appraisals and the review of medical records for health insurance.The company is also working to expand this expertise into other areas of insurance.
They recently launched EXLerate.AI – a platform using carefully trained language models.The goal is to help businesses rethink their work processes.The platform enables seamless integration of EXL with AI tools from other companies.All of this fits into customers' existing business processes.
Devesa says the demand for industry-specific LLM programs continues to grow as more insurance companies look to incorporate AI solutions that can offer better accuracy, speed and cost savings than generic AI tools.
What is an LLM?
A big language model is an artificial intelligence tool designed to process natural language and respond to natural language as well.Devesa began his career in natural language processing.He explains that these language models are pre-trained with "vast amounts of data" from the Internet.Hence their name "large language patterns".
"They pre-train a model to understand whole words on the Internet, and then train those models on specific tasks like reasoning, code conversion, code writing, summarization, question and answer, ...
According to Arturo Devesa, chief AI architect and head of innovation at EXL, industry-specific Large language models (LLMs) are expected to drive innovation as AI deployment continues.
EXL, a global analytics and digital solutions firm, recently launched its first in-house LLM and its first LLM specific tothe insurance industry.
"Field-specific or company-specific LLMs: I think that will continue to be the focus of innovation in LLMs. Of course, there will continue to be innovation among general-purpose LLMs. Maybe they will be commercialized, but it will still be a huge business. But we're a little bit different. We're building something more specific," says Devessa.
EXL's insurance law professionals are trained in Life insurance audits, Property & Casualty insurance claims, Homeowners insurance property appraisals and the review of medical records for health insurance.The company is also working to expand this expertise into other areas of insurance.
They recently launched EXLerate.AI – a platform using carefully trained language models.The goal is to help businesses rethink their work processes.The platform enables seamless integration of EXL with AI tools from other companies.All of this fits into customers' existing business processes.
Devesa says the demand for industry-specific LLM programs continues to grow as more insurance companies look to incorporate AI solutions that can offer better accuracy, speed and cost savings than generic AI tools.
What is an LLM?
A big language model is an artificial intelligence tool designed to process natural language and respond to natural language as well.Devesa began his career in natural language processing.He explains that these language models are pre-trained with "vast amounts of data" from the Internet.Hence their name "large language patterns".
"They pre-train a model to understand whole words on the Internet, and then train those models on specific tasks like reasoning, code conversion, code writing, summarization, question and answer, classification, word extraction, etc.," he says.
The process by which LLMs take information in human format and provide a response is known as Retrieval-augmented generation (RAG).The AI first digs into a document or database to extract information related to the user's question, then enters that data into a language model as context, and finally generates an answer based on that.
LLMs get better and "smarter" as they develop over the years.But it wasn't until ChatGPT "took hold of the mainstream" in 2022 that corporations began to look at LLMs as practical tools and how to use them with their own data, notes Devesa.
However, companies face three main challenges with general purpose LLMs:
- Inaccurate answers ("artificial intelligence hallucinations");
- Slow responses;
- High costs.
Field-specific LLMs can help address these challenges.
Why does insurance need its own LLM?
Devesa explains that an industry like insurance, which can be complex and span a wide range of disciplines, could benefit from an LLM specifically designed for its purposes.Such a model not only helps to improve accuracy and eliminate „hallucinations“, but is also the most cost-effective and efficient as it uses industry-specific data.
"This is useful for our customers using their specific LLM, or even for customers who want to hire us to help them build their own LLM with their own data," says Devesa.
Devesa tells how EXL started with a general-purpose LLM, then refined it using industry-specific data and training large teams of operators, data scientists and AI engineers in their offices around the world.
"The theory is that fine-tuning will always outperform RAG because you're taking the knowledge of your use case and building it into the LLM."
The LLM never saw the context.The LLM never saw the assignment.Therefore, it outperforms the basic LLM because it is no longer a pre-trained model,” says Devesa.
Field-specific LLM models were sought
According to Devesa, a growing number of clients are looking to AI solution providers like EXL for industry-specific LLMs.He noted that despite the proliferation of general-purpose LLMs, none of them "are going to start hiring armies of insurance experts and building potentially hundreds of LLMs for different fields."
"In our case, we're already building several of them. It's very challenging and requires domain expertise and specific infrastructure. For us, that's really important. It's not a consumer product. Customers are starting to say they're more interested in our LLMs. They want domain-specific LLMs—enterprise LLMs. That means they want to use LLMs with their own data. They don't want to do RAG with GPT-4," he says.
EXL Serviceis a global data analytics and digital solutions company based in New York.Founded in 1999, the company has a team of 1,500 data processing professionals and operates across a wide range of industries.