We have been inspired by the possibilities of GPT3 and OpenAI, as well as by Regtech 3.0 and the data based decision-making of epidemiologists and data scientists during the Covid pandemic. There is an urgent need and possibility that will allow creating new regulations by AI directly from the data.

What it does

Our tool allows putting in a new issue that needs to be regulated and gives several appropriate rules and regulations as suggestions to put in place, even for issues unknown beforehand.

How we built it

We built it using GPT3 and for the frontend Django as a progressive web app, where the end users, data scientists or policymakers can put in a new issue in the world in a sentence, and having trained it on many and possible solution, the tool matches it to possible solution suggestions.

Challenges we ran into

Some challenges we ran into include the problem of semantic search which is somewhat less covered by the generative tool, as well as the tool falling into a loop of unknown or same answers even when varying the freedom parameters, finding the right data sets for regulations all over the world.

Accomplishments that we're proud of

Our tool comes up with good suggestions and some robust regulations for very general use cases and many domains. It is easy to use and can be an actual business use case. We are also happy with the new team we formed and forged.

What we learned

We have learnt how to use GPT3, how to train on new data sets, how to come up with the right prompts, and how to fine tune it for an actual potential business use case.

What's next for

We want to build out our solution as a full software for legal tech and the global market.

Built With

Share this project: