We're raising funding/want to know you! Contact us: [email protected]
These articles were made purely through bigbird.dev. We're aiming to supercharge journalists with our AI so they can spend more time fact-checking while also being able to cover more stories.
We're staging our release to users at the moment. Just email us at [email protected] and we'll get back to you ASAP. We're aiming to supercharge sports journalists first since sports articles are time sensitive and on average we can create content within seconds rather than minutes.
Who's behind Big Bird and what model are you using?
I'm a HS student in the Bay Area. I don't have any compute (just my dad's old desktop) but "hacked" my way with GCP/DO credits enough to get my basic model to train. With all the rich content on language models, sequence to sequence, and the sort I've made a novel model that I am yet to write a paper on (to be clear it does work off of pre-existing models like GPT-2 1.5B because I don't have millions in the bank :) ). I love infra and worked with infra at a company in the past so I decided it was time to see how scalable and HA I could make this thing. The best part? I rely only on Colab for GPU inference (for 1.5B)... also sorry for abusing Colab :(.
What's our goal with Not Real News?
Goal:
We want to show what we at Big Bird have been working on for a while. Ultimately here's how we want to supercharge journalists: We want to allow journalists only spend their time gathering source information and we want to fill in the rest and make it coherent. This way journalists can cover more substantive and legitimate stories.
What's happening with news now?
TIL - again - people don’t read the actual articles.
— Lora Kolodny (@lorakolodny) December 31, 2019
You’re wrong. NYT does pay attention to subscriber cancellations. It’s one of the metrics for “outrage” that they take to distinguish between “real” outrage and superficial outrage. What subscribers say can back up dissenting views inside the paper about what it should do and be.
— sarah jeong (@sarahjeong) September 27, 2019
When it comes to chasing clicks, journalists say one thing but feel pressure to do another
At many websites, writers are directly encouraged to think about traffic.
Columbia Journalism Review
The way newspapers report news…you produce more errors if you try to produce things fast. That’s just a fact.
As journalism is ramping up in speed and digital media at an all time high, clickbait and widespread misinformation are getting more and more prevalent in our society. You, yourself reading this probably knows about this, but a massive number of people from all corners of the world don't. They're either under China's totalitarian control with strong censorship or Singapore's "fake news" laws that give the government right to prosecute anyone for anything they write, regardless if it is real or not. Many just believe what they see since it supports their views. These are small, widespread sites with an agenda. By utilizing Big Bird's AI, we want to empower journalists and these small news by providing them with news... real, fact-checked, and empowering news that doesn't spread lies or misinform the masses. We give them unique, fact-checked articles in return of a cleaner web.
What does "supercharge" mean?
We don't want to completely takeover journalism. Rather we want to help generate a "smart template" so journalists can spend time just editing these templates with fact-checked and reliable information. In an age of fast-media and where more articles are incentivized by publications, we hope to help support that while also giving journalists more time to fact-check and ensure their content is substantive rather than very clickbait-y.