By Tanvi Bajaj
With the massive amount of information streaming over the internet, radio and TV, many people can feel overwhelmed by their news feeds and at times, skeptical about whether what they are reading is “fake news.”
Dev Sharma, a UC San Diego alumnus, may have a solution. In early 2019, inspired to create a social service for his community, he began developing InfoPost, an app which uses AI to give news articles an accuracy rating, thereby providing news transparency to its users.
“News accuracy is so important, now more than ever,” Sharma said. The confusion surrounding the current COVID-19 crisis only goes to show that “it’s hard for people to know whether or not the news they’re reading is accurate and unbiased. InfoPost solves that problem”.
We chatted with Sharma, founder and CEO, about his InfoPost journey, along with his experience building a startup in San Diego’s growing tech ecosystem.
Year Founded: 2019
Key Players: Dev Sharma (CEO), William Vuong (Lead Backend Developer), Ravi Nagda (Lead AI Developer)
Headcount: Nine interns, along with the leadership team
Headquarters: Qualcomm Innovation Space (Atkinson Hall, UCSD)
Technology: InfoPost’s technology is based on three central categories that are used by the AI to filter every article. Finalizing the three categories was a long, tedious process Sharma explained: “We originally sent out surveys to the UCSD student body, then paid for online surveys to about 1,000 particpants. After that, we did research on ‘fake news’ and how researchers all over the world have quantified it, and looked for recurring themes.”
The InfoPost team finally settled on the three categories of political bias, headline accuracy, and source accuracy. Then came the development.
Since the system is a mix of different AI algorithms, it required a vast array of natural language processing techniques. The team took state-of-the-art models published by Google and Microsoft and tweaked them to adhere to the goal of their app.
The hardest part for the InfoPost team was developing the perfect tech
How it Works: InfoPost has differentiated its news aggregator app via its unique AI. “Google news, Apple News, and USA Today all use AI to personalize your news, and to show you the articles you care about. We go one step further by also bringing transparency to that. We are the only news aggregator that tells you how biased your news is,” Sharma said.
InfoPost’s app includes color-coded scores that provide users with ratings before they read an article. Using new breakthroughs in Natural Language Processing, InfoPost can accurately determine if an article is politically biased and to what degree. The startup’s proprietary technology informs the reader about the spectrum of political news.
While there are many fact-checking organizations (such as Politifact and FactCheck), each and every one is human-driven, meaning that each article check typically takes around two-to-three hours. InfoPost on the other hand, has managed to successfully automate that process and reduce it to a minute. On top of that, the startup’s information is available in the form of an app which anyone with a smartphone can download.
Sharma summed it up perfectly: “Politifact checks about 20-to-25 articles a day. Meanwhile, InfoPost can check 30-to-40 articles in around 40 minutes.”
Partners: InfoPost has already gathered a list of impressive partners, including Amazon Web Services, Google Cloud, Stripe Atlas, UC San Diego’s The Basement, and startR.
Inspiration: Sharma described his motivation behind his startup, saying, “The media has always been trusted with the news, but since 2015, a lot of research has shown that 78% of Americans have problems trusting what they read.”
Nowadays, many people read the news with suspicion, and so Sharma, along with the rest of his team, wanted to provide transparency – in essence, classifying the news by accuracy so that people know what they’re reading.
Challenges: Sharma discussed how initially, being a solo founder came with the challenge of learning how to run a startup and manage his own team. However, the biggest challenge came in the form of technical development. Since political bias is very subjective and hard to assess, there aren’t not a lot of data sets that exist for it. “We had to cross-check our work over and over again and constantly find new ways to validate our work,” he said.
Funding: While Sharma self-funded InfoPost with about $15,000, a later down payment based on funds from friends and family raised another $12,000. Currently, the team has a seed fund goal of $200,000 and is looking for investors.
The InfoPost team got started using resources from UCSD’s Basement program
Mentors: Along with The Basement at UCSD, InfoPost was also a part of the Rady Startup Accelerator Program, as well as Pitch Prep, UCSD’s pilot innovation program that gave Sharma and his team access to Rady professors and other startup-focused mentors. Sharma specifically named George Eiskamp, Hem Suri, Paul Shockley, and Brianna Weisinger as some of InfoPost’s most influential mentors.
On the Horizon: Over the next few years, Sharma hopes to expand the company, with more people using the InfoPost app. His long term goal is to verify the news that people read and create a platform known for being a trusted source of information. Moreover, InfoPost also plans on expanding to other sectors of news- past just US politics – such as the entertainment industry. Lastly, the team wants to build relationships and partnerships with local news and social media companies to strengthen their brand.
San Diego Tech Ecosystem: According to Sharma, San Diego is the perfect place for his startup because of all the opportunities he’s found here. Strong entrepreneurship programs through all three major universities as well as city-sponsored events (such as the San Diego Angel Conference and the San Diego Tech Hire Program) have been major factors in InfoPost’s growth. Additionally, incubators around the city as well as large tech companies such as Illumina and Qualcomm contribute to San Diego’s supportive tech community.
Let’s Taco ‘Bout Tacos: When asked for his favorite San Diego tacos, Sharma named a classic: The Taco Stand in La Jolla. Taco Stand was one of the first taco places he tried after his move to San Diego and since then, he says, “it’s been my favorite. Makes me feel like home.”
Keep up with InfoPost through Instagram, Facebook, and Linkedin.
Editor’s Note: TritonTech is an original series on UC San Diego created by Fresh Brewed Tech that showcases the innovative ideas born in the halls of academia that are making a great impact on our ecosystem and beyond.
