How Recent Advancements in AI may Affect Quantitative Finance.

If you’re interested in the field of artificial intelligence, the odds are that you’ve been keeping up with the revolutionary breakthroughs that have been rolling out over the past few weeks. From innovations in large language models to improvements in image generation, artificial intelligence is all the buzz right now and for the right reasons. Many people are already capitalizing on these innovations by spinning up new companies (ex: Jasper AI) and developing new products (AvatarAI). While many of these early adopters have already found product-market fit, there’s still an endless amount of potential that hasn’t yet been tapped into.

Given these exciting developments currently being released, the question arises as to what role they will play in influencing the quantitative finance industry. It’s no surprise that quants rely quite heavily on machine-learning-based techniques in their investment strategies, and it's likely that many will explore how these new AI models can be integrated with their existing workflows.

In this article, I’ll share three exciting breakthroughs that have been recently released by OpenAI - an AI research laboratory on a mission to create general intelligence. In doing so, I’ll highlight some potential applications of these models in the quant finance industry. I’ll also share some helpful links so that you can get up-to-speed on the inner workings of these models and how you can play with them yourself. Whether you’re already a working professional, or looking to land a job as a quant, keeping up with these technologies will help you throughout your career.

DALL·E 2

DALL·E 2 is a text-to-image artificial intelligence model. As the name suggests, it can be fed any description and will produce a photo-realistic image as the output. The model accomplishes this by forming connections between the relationships of images and the text used to describe them. For example, the prompt “An astronaut riding a horse in a photorealistic style” produces the following image:

The trick with this model is to learn how to write prompts that will generate the image that you desire. Since the model's inception, an entire community has erupted around the concept of “prompt engineering” - the art of constructing prompts to generate desired images. If you’re interested in giving this a try yourself, you can check out OpenAI.

Now you may be wondering how a text-to-image model, like DALLE, can be used in quantitative finance. Well, while there may be many applications that arise in the future, one clear use case today would be in the generation of financial visualizations. Quantitative researchers often rely on data visualization when conducting analyses or conveying their findings. However, the task of creating charts is often tedious and not very exciting in comparison to conducting advanced statistical research. In these situations, text-to-image models can be used to generate graphs and charts to help quant finance professionals analyze and understand complex data sets. Additionally, text-to-image models could be used to create more engaging/informative presentations faster. The combination of speed and communication is essential in any quant job.

Whisper

Whisper is a general purpose speech recognition model developed by OpenAI. In essence, this model takes as input an audio message and is able to generate a transcription of the speech, translate the speech to another language, generate images from the speech, and more. The model is open-sourced and can be accessed here.

There are numerous ways in which OpenAI’s Whisper can be used in quantitative finance. One significant one is in the transcription of financial news reports. Quant professionals often have to stay up to date with news from all over the world, and speech-to-text models can be used to transcribe reports generated in other languages for these quants to digest. Furthermore, Whisper could be used to transcribe conference calls, or recordings from the trading floor, into a textual format that can then be used for further financial analysis.

ChatGPT

ChatGPT is OpenAI’s latest model and has quickly taken the world by storm. Unlike Whisper and DALLE, ChatGPT is a large-language model that is built on top of GPT-3. With ChatGPT, you can pose any question to the model and it will output a relevant answer to your question. The results are quite stunning but you have to experience them yourself to get the full picture. You can play around with the model here.

ChatGPT may have the most potential in terms of its influence on quantitative finance. One example of where it could help quant researchers is its ability to analyze financial news articles and extract relevant information that can later be used as input in a trading strategy or model. Another potential application in which it could help quant developers is its code debugging ability. By just feeding the model a code snippet, ChatGPT can identify errors and vulnerabilities that may be present. This could help expedite a quant developers development process and also help ensure that the systems that are built are ultimately safe and secure. In the future, a majority of programming may be conducted in conjunction with an AI-assistant, so learning about these tools can help you in your quant job.

Final Note

It’s no hyperbole to say that the recent innovations in AI have the potential to radically transform the quantitative finance industry. With advancements in speech-to-text, text-to-image, and large-language models, the quant industry will soon start incorporating these models to fit various use-cases. If you currently have a quant job, or are looking for one, keeping up with the latest AI technologies will prove valuable in your career.

Thanks for reading this article. If you enjoyed it, feel free to check out more of our content on our quant blog. Also, if you’re currently searching for a job in quantitative finance look no further than OpenQuant for the best quant jobs.