Go green

Inspiration: Using Socratic thinking, we wanted to plant a seed of thought into the minds of those present. Hopefully, this new school of thought will spread with the ultimate vision being the general population being more aware of how human consumerism plays a huge role in the destruction of our planet, and whilst the population does not care, companies are willing to turn a blind eye.

What it does:

In relation to the theme of net zero, we wanted to highlight the common practice of greenwashing in many sectors. We set about designing a fictional company that purports to “greenify” any company that uses our service. This is achieved solely by producing a bogus sustainability statement built with an NLP algorithm trained on the sustainability statements of real companies. A prospective client will enter their company name and sector into our web form and the algorithm will be consulted to generate a statement. This is meant to show how these statements can be pieced together simply by an AI and produce something that sounds similar to real-life examples because they contain very little actual substance.

How we built it: We used the data from the "sustainability pages" of the top greenwashing companies. We classified the companies into 4 different categories: Energy, Transportation, Food and Fashion. We wanted to use that data to train our model, based on transformers, to then create new sustainability statements for a company of one of those categories.

Challenges we ran into: We tried training our model with "Cohere". Unfortunately, the output was not specifically on sustainability but was trained with a lot of external data from the internet.

Accomplishments that we're proud of: Facing and tackling the challenges together as a team.

What we learned: We learned about the huge extent of greenwashing. We learned about the difference between actually doing something about climate change and only talking about it. We learned about the great manipulative power of language.

What's next for Go green: Bring all parts together; link the code to the domain names and the implementation of the website.

Built With

• css
• dalle
• html5
• javascript
• natural-language-processing
• photoshop
• python