ChatGPT is a new frontier of Generative AI, raising questions on how its many uses can benefit businesses and drive value. Experts Harald Gunia, Alec Boere, and Jin Tao outline the six steps to enhance ChatGPT centers of excellence and, consequently, best leverage GPT technologies.
Since its November launch last year, ChatGPT – the newest member of the Generative Pre-training Transformer (GPT) tools from OpenAI – has taken the world by storm. In just under two months, it’s managed to accumulate 100 million active users, a milestone that took TikTok nine months and Google Translate over six years to reach.
This success can largely be attributed to human curiosity, with people and businesses logging on to test the bounds of its capabilities. Many have found the opportunities it (and its parent GPT-3.5) presents astounding – from improving customer experience and optimizing operations to driving sales, marketing, and innovative R&D.
How best to leverage these opportunities, however, isn’t as simple. Some of the most important elements can easily be missed or under-utilized. Organizations will have to enhance their centers of excellence (CoE) to gain the most from ChatGPT and GPT-3.5, ensuring futureproof operations.
Six steps to enhance your ChatGPT center of excellence
1. Develop a GPT-enabled capability framework
GPT-3.5 and ChatGPT mark a first step towards more advanced Generative AI. As such, businesses must develop a framework that supports adoption, builds confidence, and reaps the benefits of these, and subsequent, technologies. To do so successfully, organizations must take time to understand these new capabilities to allow for innovation at scale.
A good way to implement such a framework rests in the application of business-led IT – e.g., running enterprise architecture and product centric delivery approaches. Such an approach will help companies avoid AI proof-of-concept (POC) islands, a common hurdle businesses face when dealing with new advanced technologies.
2. Implement a training program
Training is crucial. Employees need to understand the existing capabilities, limitations, ways of working, and change impacts to effectively contribute to successful GPT-implementation projects. In addition, having use-case libraries and examples in your CoE is important for the business to understand relevant applications.
3. Establish comprehensive GPT governance
Due to current hype, many departments, teams, and individuals will use ChatGPT on their own, in diverse and partially conflicting ways. Governance would define best practices on how to use GPT, where to challenge it, and how to share the results.
Today, we’re still playing catch up on setting the right governance framework for Narrow AI, including privacy, explainability, and discrimination. With the rise of enterprise-grade Generative AI solutions, it’s paramount for companies to set the initial conditions correctly for its broad application throughout the value chain.
4. Perform a GPT fit-gap analysis to help improve processes
Develop a workaround for each critical limitation. For example, if ChatGPT doesn’t have access to enterprise-specific data then you can equip it with a retrieval tool, which injects enterprise data into its prompts.
What’s crucial here, is for organization’s to understand how best to use GPT technologies (either ChatGPT with GPT-3.5, or the latter on its own) within their infrastructure while keeping their data secure.
5. Enhance design through personalization
Don’t forget the value that context provides to the responses and uses gained from ChatGPT. Standardized, high-quality interactions with ChatGPT can stem from either customer vectors or design persona-based prompt catalogues.
Github.com provides some good examples here, showcasing varied interactions from “Act as JavaScript Console” to “Act as an Advertiser”.
6. Create a text data warehouse to optimize technology
Most enterprises have their data scattered over several systems, databases, document stores, and web applications.
As GPT-3.5 and ChatGPT can only process text documents that are 1,500 words (maximum) at a time, large text data repositories are needed in order to provide the data backbones for scalable GPT applications.
Like data lakes, it’s recommended to leverage knowledge graph technologies, as these will support scaling better.
ChatGPT beyond its center of excellence
It’s clear that, if leveraged properly, ChatGPT and GPT-3.5 can help enterprises drive value. The uses for these technologies are so wide-reaching that they can be applied throughout the business.
For example, HR functions can be extended to cover Intelligent Assistants. This would open a plethora of possibilities for current and future employees to jointly accomplish tasks with GPT-powered virtual assistants. And that’s just one example, many functions outside of HR can also be improved through GPT technologies to keep employees happy, operations running smoothly, and business growing steadily.
Reach out to our experts for more information on ChatGPT and GPT technologies.
Dr Harald Gunia
Associate Partner, Director of AI & Data Science
Alec Boere
Associate Partner, Infosys Consulting
Alec is a seasoned digital and AI expert with over 16 years of experience within the digital space, working with leading agencies and top global brands. He advises firms on the strategic development and delivery of revenue generating or category differentiating digital products and services. He has worked across the digital ecosystem, including, platform builds, apps, social as well as proposition development. His areas of expertise include AI, innovation management, product management, delivery (certified Scrum master), digital platforms, mobile, customer experience and strategy.
Jin Tao
Senior Principal
