Archive for the ‘Crowdsourcing’ Category

Thank you to all Idibon contributors

Friday, June 3rd, 2016

This post is a profound thank you to everyone who has contributed to Idibon, whether as a staff member, advisor, investor, partner or client. When we founded Idibon four years ago, Silicon Valley wasn’t interested in artificial intelligence, thinking there was little place for it outside of academia. Our goal was to bring language independent technology to as much of the world as possible, and we achieved this for smart phones, gaming, the auto-industry, healthcare, and finance services, transforming them into industries that now use machine learning to automate core business tasks. We guided our clients into the next generation of computing, as a company working on cutting-edge technology with the belief in supporting the diversity of the world’s languages and the power of technology created by a diverse workforce.

Idibon was formed to bring language technologies to all the world’s languages. Only 5% of the world’s conversations each day are in English, but as the world comes online the majority of people are being excluded. Language technologies we take for granted in English, from search engines to spam filtering, work only in a handful of the world’s 7,000 languages. In these and more advanced language technologies, like smart personal assistants and social analytics at scale, we are falling far too short when enabling people to enter the digital age on their own terms.

We showed the world that it was possible to build and ship language technologies that used artificial intelligence to automatically adapt to any language. Organizations using Idibon’s technology supported users in more than 60 languages. Some of the language we’ve supported have less than 100,000 speakers, which is a community that wouldn’t be financially viable to support if the technology needed to be manually adapted to each language. In short, we proved the market for supporting the full diversity of the world’s languages.

idibon_world

Worldwide, our products impacted leading companies. As a result, the largest manufacturer of smartphones became smarter in more languages. We powered the most visited website for information about automobiles. The most popular game in history made decisions based on community feedback that we filtered. The world’s most successful auditing company selected us as their preferred vendor for social media analytics. We helped the leading financial news service expand into more languages. The leaders in these markets paid seven figures for Idibon’s products, and they found enough value in our technology that we never lost a single client to a competitor.

Our client’s happiness was built on technology that is ahead in many dimensions. Our AI was built to adapt to any language using only input from native speakers. It took us just days working in new languages to beat prior state-of-the-art performance for tasks like sentiment analysis, filtering, and information extraction. It was always exciting to see a non-technical user start interacting with our technology, and behind the scenes the machine learning seamlessly started adapting to their specific task, while ensuring the quality of the data coming from the human users. On the deployment side, it was a joy to see our team give an already well-engineered machine learning library within Spark a 40x speed up, open-sourcing the innovation so that engineers everywhere can take advantage of low latency scalable machine learning.

The impact was often more direct. MIT and Stanford students used our technology to research disaster response, FEMA used real-time damage analysis that we hosted when responding to hurricane Sandy in New York, and we have been used across sub-Saharan Africa for everything from election monitoring to maternal health. When they needed to process text messages sent by citizens globally, the United Nations came to us.

idibon_bike

With our mission to have a positive impact on the world, we said “no” to helping governments spy on citizens: in one case, on four different occasions to the same government. Allowing organizations to support social good globally has been core to Idibon’s mission, and we are proud that we brought computational linguistics to many languages for the very first time, while directing the best technology towards positive impact.

More than anything else, I am proud of the team that worked together. The team embraced diversity more than anywhere I have seen before and this was our strength. We always spoke more languages than we had people and we consistently had a gender-balanced staff, including in engineering. I’m proud that we had a company where single mothers thrived, at least one doubling her salary in jobs before/after Idibon. Most staff members had worked in both international development and Silicon Valley companies before joining Idibon, and most people joined Idibon with more lucrative competing offers. Despite this, over our entire life only two people declined an offer to join us, and only one engineer ever quit. I am profoundly grateful for everyone’s commitment to Idibon’s goals. In too many tech companies, staff derive comfort in coming to workspaces populated by people just like themselves. We found inspiration coming to an office where everybody was profoundly unique.

On a personal level, I am grateful that many staff members have called Idibon their favorite place to work, and that many of our investors have said that they would happily invest in our future ventures. Every staff member, advisor, investor, partner and client should be proud of the positive impact they had on the world, from advancing AI to advancing the strength that comes from diversity: both linguistic diversity and from among the engineers creating our future technologies.

Robert Munro
June, 2016