PwC Generative AI Knowledge Modernization

PwC and Microsoft are using

Microsoft's OpenAI services to create the future of knowledge work by enabling

Generative AI (GenAI) for specific industry and business use cases built on a

foundation of trust.

Focused on responsible AI since 2017,

PwC’s Responsible AI Toolkit was ranked as a 2020 World Changing Idea by Fast

Company. PwC has also been ranked as an AI Consulting

Leader by Forrester for the past four years.

PwC provides a lifecycle of

offerings on Azure from ideation to execution:

• mapping client challenges
• assessments & collaborative proof-of-value workshops
• getting your data ready for GenAI
• preparing workforce & building skills for GenAI
• conducting discovery & model training
• executing delivery integration & deployment
• establishing operations that mitigate risks
• applying controls and governance to build trust

We are reimagining the way we work

with Azure OpenAI Service, internally and with our clients. GenAI is, today,

amplifying knowledge work with unprecedented efficiency, helping you scale

further, work faster, reduce costs and enable new business models.

What is GenAI and what does good

governance look like?

GenAI is a subset of deep learning

that involves training a model to generate new data. PwC has built trusted

solutions on Large Language Models (LLM) like GPT on Azure OpenAI that

are in live production.

Our governance approach includes AI

experts -

including data scientists and engineers - and technology practitioners -

alongside prompt engineers, business analysts and leaders for each use-case pod. In a controlled, tiered environment,

they build tools on Azure OpenAI

Services like CoPilot, to enable capabilities

for your broader workforce, making sure your data remains trusted and secure.

Real-world use

cases

PwC is actively transforming

knowledge work to help drive value realization for enterprise businesses and

their IT groups using deep industry use cases powered by Azure OpenAI Services.

Here are some of the results we are currently seeing with our clients:

• Regulatory Text Classification in financial

services Regulatory rules

are correctly identified ~80-95% of the time and policy review time can be

reduced by more than ~50%. Use cases include ESG, tax and audit reporting.

• Safety Automated Narrative

Generation in manufacturing ~90% touch-time savings for 250,000 cases of inventory a year.

• Inclusion Exclusion criteria

generation across industry Our approach performs at an accuracy level of ~78% where no automation existed

previously.

• Negation Detection in life sciences Generating templated clinical trial inclusion exclusion criteria to help drive

patient enrollment and trial success, improving the precision of one model

pipeline from ~73% to 94%

• Insurance pre-authorization in healthcare effectively create

pre-authorization letters with optimal approval rates for patient insurance

claims, targeting more than ~70% time savings

• Text Summarization and synthesis for Marketing

Content Reduction in human-hours

that goes into enhancing search engine metadata on a monthly basis, as the AI

generated outputs can serve as a starting point.

Delivering GenAI at scale

PwC’s Generative AI Knowledge

Modernization is a pioneering factory model. Multidisciplinary agile pods develop specific

business and industry-led uses cases built on a foundation of responsible AI to

help drive efficiency while enabling good governance. This model is being

applied within PwC as a first use case and replicated with our clients. We are

building out solutions on GPT that are live in production, built for our

regulated industry.

With PwC’s factory model, Azure

OpenAI has the potential to help drive 50% to 90% productivity improvements in

repetitive tasks such as knowledge capture, process automation, code creation

and content generation.