In an effort to democratize artificial intelligence (AI) and natural language processing (NLP) technologies, a group of over 500 researchers organized the project “Big Science” to improve scientific understanding of AI and NLP capabilities and open it to the scientific community.

 

The project’s central objective is to demystify the mechanisms of NLP models - taking the power out of the hands of tech giants, and allowing for democratized decision making and more transparent ethical discussions. As the project continues to expand, including several Ghanian researchers, it has become the largest open science contribution in artificial intelligence ever.

 

An open sourced approach to the NLP industry would finally match its impact-driven work that puts people first. Below are three of the latest human-driven applications of NLP.

 

  1. XLM - RoBERTa

 

The machine-learning system is the most advanced of its kind, able to understand dozens of languages better than other systems trained to comprehend a single language. This software has enabled Facebook and Google to reform their automatic blocking systems for hate speech and online bullying. Facebook reported in May that its AI systems are capable of detecting 97 percent of the hate speech that’s eventually removed from the site before it’s reported.

 

Facebook credits AI technology such as XLM - RoBERTa as “humanity’s best chance” at confronting malicious and divisive online speech.

 

  1. Sony AI x PlayStation

 

Video game artificial intelligence is evolving… enough to create characters that can think? Sony AI is partnering with PlayStation developers to create intelligent computer-control characters that teach themselves how to act through machine learning trial and error. As AI expands, gaming is expanding along with it, building systems capable of generating reactive characters that can mirror the intelligence and creativity of their players.

 

  1. Erica

 

As Covid-19 lockdowns came into effect, people faced new financial challenges and had new unanswered questions. In response, Bank of America (BoA) used NLP to create Erica, a virtual financial assistant trained to understand over 60,000 Covid related questions. From questions on government payments and stipends to health insurance, machine learning technology enabled BoA to meet customers’ expectations and avoid financial panic.

 

From countering online hate speech to better gaming and banking, NLP-mobilized solutions place people first. A future of open-source NLP would not only mean more similar impact-driven innovations, but a more transparent understanding of how progress is made.