Morgan Stanley built its internal AI tool to address a difficult coding problem: reworking the old code inherited in more updated coding lengths.
Morgan Stanley presented the AI tool, which is based on the OpenAi GPT models, in January, per The Wall Street Journal. The tool, called Devgen.AI, translates the code into older languages, such as Perl (released in 1987), to simple English, which developers can be the basic use to rewrite the code into newer languages such as Python.
Related: The Amazon Cloud CEO predicts a future in which most software engineers do not encode, and AI does it in its place
Mike Pizzi, Morgan Stanley’s world chief of technology and operations, said WSJ That in the five months after its launch, Devgen.AI has worked in nine million lines of code, saving the company’s 15,000 developers approximately 280,000 hours of work.
Pizzi said Morgan Stanley chose to build the tool that technological companies had no solution that could adapt to the exact specifications of Morgan Stanley. Commercial tools lacked the oldest coding languages, especially the specific ones of a company.
“We discovered that building our sponsorship like certain capacities that we are not really seeing in some of the commercial products,” Pizzi told WSJ. “We saw the opportunity to make the jump early.”
Related: Morgan Stanley plans to fire 2,000 workers, replacing some with AI
Morgan Stanley trained Devgen.AI in languages within his own code base, including personalized languages for the company. However, the ai tool still has to do when it comes to complete translation. He thought that the tool can, in theory, rewrite the code of an older language to a newer one, is not not that it is not, it cannot do not do not the new code efficiently or as well as a human developer, Pizzi said.
That is why Morgan Stanley keeps human developers involved in the translation process of the old code or inherited into new languages. Pizzi revealed that the company will not reduce its software engineering workforce as a result of the AI tool, thought that the company made 2,000 or its workforce of people from 80,000 people in March.
Morgan Stanley has published several AI applications for employees, including one that helps them summarize video meetings and another that quickly find information for the company’s research body.
Morgan Stanley CEO, Ted Pick, told investors last year that artificial intelligence tools could save employees up to 15 hours per week and be “potentially really changing the game,” Reuters.
