< Back
Slide2   Approved
Insights

ChatGPT and We – Exploring generative AI use cases for MillTechFX

Fintech
New technologies
ChatGPT
Sam Hunt Mini

Posted by Sam Hunt at MilltechFX

'2 min

22 January 2023

22 January 2023

When OpenAI launched public access to ChatGPT we got excited. Access to a powerful emerging technology . . . check . . . a company and a product with capital letters at the start and end . . . check. It felt like a perfect match for MillTechFX.

What is ChatGPT?

For those that haven’t had a chance to play with it yet. ChatGPT is an AI that uses natural language processing (NLP) to generate content and conversations. Chatbots have been around for years and years, but recent advances in NLP, with models drawing on a much larger number of factors, has changed the game. As a result, the conversations and communications that are outputted are much richer with more context, tone and style.

Applying ChatGPT to MillTechFX

We got stuck in and started exploring use cases.

The first out of the blocks was as a code-writing support tool. Andy, our cloud architect, was working on a failover script for our DR environment. He threw the challenge at ChatGPT. Whilst the solution it came up with was not perfect, it was a nearly workable solution with a few elegant elements that fed into our eventual solution. The feature of clicking a button to switch between languages was interesting and something that could play a useful role in refactoring code from one language to another. With Microsoft heavily involved in ChatGPT and also owning GitHub, which has Copilot its own code writing AI support tool, there's an exciting possibility for collaboration to build a powerful product to augment human engineers.

The next use case we explored was a policy generator. One of the challenges for an early-stage regulated organisation is to quickly put in place the range of policies required to define standards and protect your organisation. Whilst this will always be personal to the organisation, quickly generating boilerplate policies can really help accelerate this process.

We’ve not implemented anything here yet but see great potential for NLP to support our marketing efforts. I can see us improving the chat experience on our site, replacing the FAQs with something more dynamic and quickly generating social content.

Perhaps my favourite use so far, was Eric our CEO turning to ChatGPT to advise him on how to dress for the smart casual theme for the company Christmas party.

Looking to the future

From an industry perspective, a word of caution. Global regulators are working to define safeguards around NLP models. In the meantime we should assume that responsibility sits with the company using the model. For that reason we will be staying away from using unsupervised NLP models for regulated activities such as financial advice or prospecting clients.

From a tech perspective, ChatGPT has an exciting future. The current model is based on 175 billion parameters, the next major version, version 4.0 is expected to use 100 trillion parameters, which will lead to another major leap in its capability.

It’s certainly exciting and something we will continue to explore at MillTechFX.

 

Source: Wired, A New Chip Cluster Will Make Massive AI Models Possible

Sam Hunt Mini

Sam Hunt, CTO

With 13 years of technology experience, Sam's previous roles included establishing the Emerging Technology practice of Capgemini UK and technical leadership of the FinTech Practice of Star Global. As a true technologist at heart, he continues to explore and champion new technologies, frameworks, approaches and cultures.

Posts by tag

Other posts you may be interested in

Norbert1
Insights

Playing Lego with our front-end

Modularisation is a confusing word. It’s used in lots of different industries, like automotive, construction and manufacturing. But what does modularisation mean in software development?

Fintech
FX infrastructure
New technologies