Bloomberg plans to integrate GPT-style A.I. into its terminal

Bloomberg LP has developed an AI model using underlying technology similar to OpenAI’s GPT, and plans to integrate it into features delivered through its terminal software, a company official said in an interview with CNBC. Is.

– Advertisement –

Bloomberg says Bloomberg GPT, an internal AI model“CEO of Citigroup Inc?” Can more accurately answer questions like, assess whether headlines are bearish or bullish for investors, and even write headlines based on short blurbs.

Large language models trained on terabytes of text data are the hottest corner of the tech industry. giants like Microsoft And Google are rushing to integrate the technology into their products, and artificial intelligence startups regularly raise funding at valuations in excess of $1 billion.

– Advertisement –

Bloomberg’s move shows that software developers in many industries beyond Silicon Valley see cutting-edge AI like GPT as a technological advancement that allows them to automate tasks that used to require humans.

“The capabilities of GPT-3 and the way it achieved its performance through language modeling was not something I expected,” said Gideon Mann, Head of ML Product and Research at Bloomberg. “So when that came out, we were like, ‘Okay, this is going to change the way that we do NLP here.’

– Advertisement –

NLP stands for Natural Language Processing, the part of machine learning that focuses on extracting meaning from words.

The move also shows how the AI ​​market cannot be dominated by giants with massively normalized data.

Building large language models is expensive, requiring access to supercomputers and millions of dollars to pay for them, and some have wondered whether OpenAI and Big Tech companies will develop an insurmountable lead. In this scenario, they would be the winners, and would simply sell access to their AI to everyone else.

But Bloomberg’s GPT does not use OpenAI. The company was able to use freely available, off-the-shelf AI methods and apply them to its vast store of proprietary — if niche — data.

So far, Bloomberg says its GPT shows promising results such as whether a headline is good or bad for a company’s financial outlook, converting a company’s name to a stock ticker, detecting important names in a document, and even That answering basic business questions. Like who is the CEO of a company.

It can also do some “Generative AI” applications, such as suggesting a new headline based on a short paragraph.

An example in the paper:

OpenAI’s GPT is often called the “baseline” model because it was not intended for any specific task.

Bloomberg has a different perspective. It was specifically trained on a large number of financial documents collected by the firm to create a model that is particularly fluent in money and trading.

In contrast, OpenAI’s GPT was trained on terabytes of text, most of which had nothing to do with finance.

Nearly half of the data used to build Bloomberg’s model comes from non-financial sources scraped from the web, including GitHub, YouTube subtitles and Wikipedia.

But Bloomberg also added more than 100 billion words from a proprietary dataset called FinPile, which includes financial data accumulated over the past 20 years, including securities filings, press releases, Bloomberg news stories, stories from other publications and a web crawl. The focus is on. Financial webpage.

It turns out that adding specific training materials has led to sufficient accuracy and performance increases on financial tasks that Bloomberg plans to integrate its GPT into features and services accessed through the company’s Terminal product, Bloomberg isn’t planning a ChatGPT-style chatbot, though.

One early application would be to convert human language into the specific database language that the Bloomberg software uses.

For example, it converts “Tesla price” to “(get(px_last) for([‘TSLA US Equity’],

Another possibility would be to have the model do the behind the scenes cleaning up of the data and do other work on the back end of the application.

But Bloomberg is also using artificial intelligence to power features that can help financial professionals save time and stay on top of the news.

“There’s a lot of work that we’re doing to help clients deal with the data deluge of news stories, whether it’s through summarization, or through monitoring, or being able to query those news stories or transcripts. Yes. There are a lot of applications out there,” Mann said.