Published on January 31, 2024, 5:25 am

Building Well-Defined Language Models For Ai: Accenture Emphasizes Responsible Practices

Technology developers are being urged to build well-defined large language models (LLMs) for artificial intelligence (AI) in order to optimize business processes responsibly. This advice comes from global professional services company, Accenture. During a recent briefing, Arvin Yason, innovation lead at Accenture Technology in the Philippines, emphasized the importance of setting up LLMs with narrow parameters. This approach helps prevent “hallucinations” caused by knowledge conflicts and gaps that can arise in rapidly evolving technology.

However, before seeking AI solutions, businesses must first identify the dimensions and functions within their operations that can benefit from such capabilities. According to Mike Lao, data and AI lead at Accenture Technology in the Philippines, organizations need to examine how AI can align with their business use cases and target outcomes. He also highlighted the significance of human-centric design when implementing tech solutions within a business context.

Accenture’s Technology Vision 2024 report predicts that AI will play a key role in curating responses through advice and result summation. The report suggests that information search will evolve into synthesis. Business leaders who reimagine how information is managed within their organizations and equip their employees with AI-enabled enterprise knowledge tools stand to gain exponential performance gains and competitive advantages.

In line with these predictions, the Accenture Pulse of Change 2024 index reveals that companies in the Asia Pacific region are already customizing foundation models using proprietary data from proof of concepts. The index also highlights an increasing trend among C-suite executives planning to invest more in AI-related initiatives this year.

To support the growing demand for generative AI solutions, Accenture plans to expand its network of generative AI studios in the Philippines as part of its $3 billion global investment announced last year. The focus will be on developing industry-specific solutions while doubling the company’s AI talent pool to 80,000.

During the briefing, several platforms were introduced as examples of Accenture’s latest AI developments. Companion, for instance, optimizes routes and handles chatbot inquiries to expedite utility and field service operations, reducing manual inspection time. It also allows field personnel to quickly assess damages via image input and generate corresponding work orders with bill of materials.

Another platform called CareCoach serves as a training tool for tech support representatives. It can simulate various customer personas based on specified criteria such as churn or level of difficulty. CareCoach is further customizable according to industry requirements and provides personalized feedback and scores to improve training outcomes.

Accenture has also introduced Gen-E, an internal AI assistant designed to simulate employee viewpoints. Gen-E is prompt-engineered using the company’s own knowledge sources and operates via text, image, and speech inputs. Among its capabilities is the ability to conduct office tours.

Looking ahead, Mike Lao stressed the importance of close collaboration between companies that prioritize value creation, building a digital core, transforming work and talent, and fostering responsibility. He emphasized that successful implementation of AI solutions is not solely driven by technology but also by ensuring that everyone in the organization is empowered for productivity and creativity in the workplace.

As businesses increasingly adopt AI technologies, it is crucial for developers and organizations alike to prioritize responsible practices while reaping the benefits offered by these powerful tools. With careful planning and consideration, AI has the potential to revolutionize business processes across industries while maximizing performance gains and delivering competitive advantages.


Comments are closed.