Bloomberg is building an AI chatbot for its Terminal
What's the story
Bloomberg is planning to upgrade its Terminal with an AI chatbot-style interface called ASKB (pronounced ask-bee). The move comes as the company grapples with an ever-increasing amount of data on the platform, from earnings and asset prices to weather forecasts and consumer spending patterns. Shawn Edwards, Bloomberg's Chief Technology Officer, admitted that finding relevant information in this data deluge has become increasingly difficult.
Tool introduction
ASKB in beta testing
The ASKB interface, which is currently in testing for about 125,000 Terminal users, is designed to help finance professionals with their work. The main goal of this new tool is to let users ask high-level questions or test abstract investment theses against the data through natural language prompts. Edwards explained that the primary issue they are trying to solve with AI is helping users find key insights and synthesize a view of the world around a particular idea.
User experience
Changing the game for finance professionals
The ASKB interface is expected to change how finance professionals interact with the Terminal. Instead of going through a lot of data points, they can now ask broad questions like "How is the war in Iran and a change in oil prices going to affect my portfolio?" Edwards emphasized that while these tools won't make an average employee great, they will allow experts to do better analysis and deeper research.
Workflow automation
Potential to automate junior analysts' work
ASKB also has the potential to automate some of the legwork that junior analysts usually do. Edwards said he could create workflow templates with ASKB, write a long query, and ask for a synopsis of the bull and bear cases, what the Street is saying, and what guidance is. He added this would let him schedule these workflows or trigger them when certain conditions are met in the world.
AI concerns
Impact on jobs and the importance of training
When asked about the potential impact of tools like ASKB on the labor market, Edwards emphasized that there is a big question of how to educate, train, and mentor junior analysts. He said for at least the next few years, one can't just accept what an AI system says without having a rooted understanding of their craft. This is especially important in finance where decisions based on faulty information can have serious consequences.
AI reliability
Addressing hallucinations in ASKB
Edwards also spoke about Bloomberg's efforts to minimize hallucinations in ASKB. He said they take a very conservative approach and build in validations at every single step. "We have a validation check that looks at the information content of a summary and checks whether those facts are fully contained in the paragraphs we synthesized from," he explained.