Artificial Intelligence Technology Scouting

  • By: JAIC Public Affairs
Artificial Intelligence Technology Scouting

MARCH 2021

This edition will address the below topics:

  • AI Assistants for COVID Research
  • AI Chip Supply Chains
  • AI Can “Imagine” New Images and Art


Topics

AI Assistants for COVID Research

AI has the potential to save human researchers and analysts enormous amounts of time. For a task that requires viewing many documents, data, or footage, AI can provide a human collaborator an assist by highlighting specific pieces of information for human review.

During the COVID-19 pandemic researchers have been using AI assistants to review large amounts of literature about the virus SARS-CoV-2 using data sources established by the White House OSTP.  These systems, researchers and their AI assistants can speed up the research process, highlighting key findings from large troves of academic research about the virus, to develop new therapies and treatment options more quickly.

Link: AI Tools Aim to Tame the Coronavirus Literature | Nature

Securing the AI Chip Supply Chain

Access to state-of-the-art semiconductors, specifically the chips that make up GPUs, TPUs, and ASICs, is a critical requirement for transformative AI. Amidst a global chip shortage, the DoD is providing incentives for secure chip supply chains in the U.S.

The National Security Technology Accelerator (NSTXL), is spinning up a new program called RAMP-C, or Rapid Assured Microelectronics Prototypes Commercial, to assure continued access to leading chips for AI and other applications at the DoD.

Link: DoD Looks to Bolster Domestic Chip Manufacture | Reuters  
Link: RAMP-C | NSTXL  

AI Advances in Image Generation

Several months ago a new language model, a type of AI that can interpret and generate human-like text, named “GPT-3” received lots of attention for its ability to perform a wide-range of human tasks without human supervision or task-specific training.

Researchers are now applying the same architecture used to build GPT-3 to images, building models that can reason about the content of images and videos, including generating new images, based on textual descriptions. For example, given the description “An armchair in the shape of an avocado”, the network “imagines” what the object described would look like.

Link: Using Artificial Intelligence to Counter Cybersecurity Threats for Drones | OpenAI

This blog is for AI/ML learning and education purposes only. The programs and initiatives mentioned do not represent an endorsement by the U.S. government, the DoD, or the JAIC.