Remove Conservancy Remove Policy Remove Proposal
article thumbnail

Facebook reportedly ignored its own research showing algorithms divided users

The Verge

Changes might disproportionately affect conservatives and might hurt engagement, the report says. Leading the effort to downplay these concerns and shift Facebook’s focus away from polarization has been Joel Kaplan, Facebook’s vice president of global public policy and former chief of staff under President George W. The reason?

Facebook 122
article thumbnail

Science Technology and Society Forum in Kyoto

Beneblog: Technology Meets Society

I'm in Kyoto, Japan for the STS forum, an incredible gathering of top scientists and policy leaders from around the world. It seemed like a larger scale version of what we are already doing for the conservation movement with Miradi, and what we've been kicking around as the "Carbon P&L" concept for a possible new social enterprise.

professionals

Sign Up for our Newsletter

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

article thumbnail

Pre-training generalist agents using offline reinforcement learning

Google Research AI blog

Scaled Q-learning: Multi-task pre-training with conservative Q-learning To provide a general-purpose pre-training approach, offline RL needs to be scalable, allowing us to pre-train on data across different tasks and utilize expressive neural network models to acquire powerful pre-trained backbones, specialized to individual downstream tasks.

Offline 97
article thumbnail

CASE STUDY: Marine Conservation Group Wins Victory with Care2 - Online Fundraising, Advocacy, and Social Media - frogloop

Care2

The Marine Fish Conservation Network (MFCN), a coalition of over 190 organizations dedicated to ocean conservation, wanted to educate members of Congress and ask them to reject proposals that might do irreparable harm to several fish species. All Care2 terms of service and privacy policies apply. All rights reserved.

article thumbnail

Top 2021 Fundraising Strategies: Mastering Relevant Content Marketing

Bloomerang

To give you an example of how to do this, a recent client was having problems telling the story of the marine conservation program they were building. Fundraisers must understand they no longer just write hard copy fundraising letters and grant proposals. They must write for the website, blog, email, social media, and more. .

Content 126
article thumbnail

How to Contribute to Theoretical Reward Learning Research

The AI Alignment Forum

However, even in the general RL setting, it is relatively straightforward to prove that there always exists a deterministic optimal policy. STARC metrics are essentially a continuous measurement of how much the policy orderings of two reward functions differ relative to a particular transition function and discount factor.

article thumbnail

Google Research, 2022 & beyond: ML & computer systems

Google Research AI blog

An illustration of AllGather-Einsum with 2-way intra-layer model parallelism, proposed in CollectiveEinsum strategy. PRIME uses logged accelerator data, consisting of both feasible and infeasible accelerators, to train a conservative model, which is used to design accelerators while meeting design constraints.