Aharon Wasserman

Head of R&D @nytimes.

[Tuhin Chakrabarty @TuhinChakr]

🚨New paper on AI & Copyright

👨‍⚖️Courts have credited LLM companies' claims that safety alignment prevents reproduction of copyrighted expression.

But what if fine-tuning on a simple writing task ruins it all?

Worse : Fine-tuning on a single author's books (e.g., Murakami) unlocks verbatim recall of copyrighted books from 30+ unrelated authors, sometimes as high as 90%.

Joint work with @niloofar_mire (@LTIatCMU), Jane Ginsburg ( @ColumbiaLaw) and my amazing PhD student @irisiris_l (@sbucompsc )

(1/n)🧵

[NYTimes R&D @NYTimesRD]

We’re also looking for help on short term projects across a variety of fields: LLMs, AI, 3D design and modeling, full stack development, and prototyping across emerging tech. Let us know what you bring to the table: https://t.co/tOnV2otqVM

[NYTimes R&D @NYTimesRD]

How *exactly* did Kadarius Toney slip past Philadelphia Eagles receiver Zach Pascal to help clinch #SuperBowlLVII?

We used AI to help freeze that moment in 3D and analyze the play. Read more about our collaboration with @nytgraphics: https://t.co/NsAYjoH7d5

[Michael Black @Michael_J_Black]

ICON was now used by the @nytimes in their Super Bowl coverage to reconstruct a play in 3D from a single video frame. We like to give European and American "football" equal time :) https://t.co/5YhilTIZJ6

[glTF 3D @glTF3D]

During NYT R&D’s work developing 3D storytelling tools with the newsroom, they built plugins that fill the gap for Vite and Rollup. These plugins enable build tools to parse and optimize 3D scenes using the popular and web-friendly glTF file format.

https://t.co/QJbaqQ9Bz1

[Chloé Desaulles @cdslls]

@NYTimesRD just published an article on our process for #worldcup2022qatar immersive articles -- using ICON to generate 3D captures which are then stylized and edited in the web, with an internal platform based off of @theatre_js and @threejs. Check out the piece below :

[NYTimes R&D @NYTimesRD]

Applications for our summer 2023 engineering internship are open! R&D interns embed on a team of engineers, designers, journalists and creative technologists to explore how emerging technologies can be applied in the service of journalism.

Apply here: https://t.co/7cMt5rz038

[Alexandre Devaux @AlexandreDevaux]

Some more prod using #AI for 3D clothed human reconstruction using the brilliant work from @Michael_J_Black and his team.🙂
Lots of fun experiments @NYTimesRD @nytgraphics
https://t.co/KSApa82SeJ

[NYTimes R&D @NYTimesRD]

How exactly does @mikiebarb sound in spatial audio? We worked with the NYT Audio team to find out. Listen to episodes of the Daily in our immersive spatial audio playlist: https://t.co/1LYB3onPRe

[NYTimes R&D @NYTimesRD]

Questions and answers are central to Times journalism. Last year we launched a research effort to explore generating open-ended questions from news articles. Read more about our approach: https://t.co/7zPfNWHJrQ

[NYT Graphics @nytgraphics]

We’ve added a 3D segment to this analysis to show how much of the goal was open to Weston McKennie. No doubt an opportunity McKennie would want back. See our full story here: https://t.co/XMYrIg2Uwv

[Michael Black @Michael_J_Black]

Neural avatars come to the World Cup. @nytimes uses our ICON method to create 3D football players from video. Congratulations @yuliangxiu, Jinlong Yang, and @dimtzionas Check out ICON at https://t.co/I0Ji9qKHwq

[Chloé Desaulles @cdslls]

First article in a #WorldcupQatar2022 series using photographs and ML to generate 3D models of the players and their environment to break down important plays from different angles

[NYTimes R&D @NYTimesRD]

In 2021, the Dixie fire burned nearly one million acres in Northern California. Creating 3D animations of the fire and its atmospheric impact required us to use new tools and design new techniques. Learn how the visualizations were created on Times Open: https://t.co/fjDt76fGFH

[NYTimes R&D @NYTimesRD]

R&D is experimenting with storytelling applications of Neural Radiance fields (NeRF), a machine learning technique for achieving photorealistic detail in 3D. We recently published a series of tests using this method: https://t.co/LNjkPSIKfm

[Marc Lavallee @lavallee]

The news nerd survey is vital for our community to understand where we are and what we need.

If you work at the intersection of tech and journalism and haven't yet filled it out, please consider doing so. It's quick and painless!

https://t.co/XSwHYs6SDy

[NYTimes R&D @NYTimesRD]

Exploring new ways we can bring 3D captures to readers, R&D engineer Peter Whidden built a @threejs library allowing web browsers to render interactive NeRF scenes. Learn more about all the ways we’re moving 3D web technology forward: https://t.co/w617O7GZ1o

[NYTimes Communications @NYTimesPR]

We’re excited to welcome @jashkenas back to The Times. He returns as head of graphics for @nytopinion, leading a team at the forefront of visual journalism. https://t.co/H7jcl0s4Mx

[Aharon Wasserman @AharonWasserman]

Experiments like this help us understand the range of storytelling tools that will be available for journalists in the near future. Here, we explore the possibility of altering a reader's physical environment.

[NYTimes R&D @NYTimesRD]

R&D Engineer @AlexandreDevaux explains what’s behind a reality-warping experiment in mixed-reality. Read more about what we’re doing to push spatial journalism forward: https://t.co/SzbFsMXmud

[Raiz - Volumetric Arts /Immersive artworks @RaizNewMedia]

Son Fornés archaeological site (10th century BC)
The biggest talaiot of the Balearic islands. #Photogrammetry model in #3d , #vr or #ar through @Sketchfab at: https://t.co/dOx6717OQx
processed @RealityCapture_
#sonfornes #talaiot #bronzeage #scan #archaeology #webgl #vfx