While at MIT Media Lab in 2010, Karthik Dinakar and Birago Jones developed a machine learning tool destined to help content moderation teams at tech companies like Twitter and YouTube. The project excited many, leading to a demonstration at a White House cyberbullying summit. However, the system tripped over unconventional wording in teenage vernacular, revealing its limitations and the need for a better understanding of the data being used for machine learning.
Inspired by this, Dinakar and Jones evolved the program into Pienso, a point-and-click tool enabling non-experts to develop machine-learning models without writing any code. It has since been utilized for creating expansive language models that help detect issues such as misinformation, human trafficking, and weapons sales.
The founders discovered that spam filters trained by students at local schools were surpassingly good, causing them to recognize that empowering professionals in their field with AI, not just democratizing it, was a fruitful way forward.
A partnership with SkyUK allowed Dinakar’s and Jones’ work to understand common customer problems. Today, the models generated by the partnership process about 500,000 customer calls per day and have reportedly saved SkyUK over £7 million by reducing call times.
During the onset of Covid-19 in 2020, government officials utilized Pienso to mine thousands of research articles concerning coronaviruses. This study assisted officials in identifying and bolstering crucial supply chains for drugs, such as the antiviral drug remdesivir.
Operating on internal servers as well as cloud infrastructure, Pienso offers companies an alternate way to use AI services without giving up their data. Compared to Adobe Photoshop but for big language models on the web, Pienso permits users to import, refine, analyze, and manage data for deep learning.
In a bid to reduce latency and enhance accessibility to AI, Pienso recently announced a partnership with GraphCore to provide an improved computing platform for machine learning. The developers envisage a future where more effective AI models are created by the professionals most familiar with the problem at hand.
Dinakar concludes, “It’s about bringing a garden of models together and allowing them to collaborate with each other and orchestrating them in a way that makes sense — and the people doing that orchestration should be the people who understand the data best.”