This article was written by a human. Here’s how to tell.

Accenture has won a broad AI services contract from S&P Global, the company behind the eponymous stock market index.

S&P is active in quite a few other areas as well. One of its subsidiaries currently ranks as the world’s top credit ratings agency. Through a number of lesser-known business units, the company provides market research to executives in industries such as the auto sector.

S&P has high hopes for its deal with Accenture, the world’s largest consultancy by headcount. Its chief AI officer Bhavesh Dayalji put it this way in a prepared statement: “we are partnering with Accenture to ensure that our people and our customers are not only prepared for the gen AI transformation, but are ready to lead it.”

The first component of the Accenture deal is an employee training program. The program, which is set to launch in August, is meant to equip 35,000 of S&P’s staffers with new AI skills. The companies detailed that the program will harness offerings from Accenture’s LearnVantage employee training business, including a customized repository of AI learning content.

The second component of the collaboration focuses on S&P’s client base. The goal will be to help the firm’s customers, which include many large financial institutions, implement AI in their business operations. S&P and Accenture plan to achieve that using technology from Kensho, an IT firm the former company bought in 2018.

Kensho maintains a scoreboard that ranks popular LLMs based on how effective they are at tasks relevant to financial professionals. For example, the tool evaluates the accuracy with which a model extracts key business metrics from an earnings report. Banks and other financial sector players can use the scorecard to find LLMs applicable to their internal AI projects.

Through its partnership with Accenture, S&P will enable companies to combine Kensho’s scoreboard with AI optimization services from the consulting giant. Those services are meant to ease tasks such as model fine-tuning and prompt engineering. The former term refers to the practice of modifying LLMs in a way that makes them better at the tasks for which an organization wishes to use them. Prompt engineering, meanwhile, is the process of finding the prompts that maximize an LL’s model’s response quality.

text

text