Last Week in AI

Jesus Rodriguez
Jun 2 · 3 min read

Every week, my team at Invector Labs publishes a newsletter to track the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:

From the Editor: Collaborative and Competitive AI

Collaboration and competition are characteristics that differentiate humans from other species. Similarly, aspects such as collaboration and competition are essential components of human intelligence that evolve organically based on the characteristics of an environments and the parties involved. Not surprisingly, collaborative and competitive behaviors have remain a high roadblock for AI solutions. If you think about your favorite examples of AI, they are likely to be constrained to a single agent operating in an isolated environment. The moment we transition into environments with multiple agents, then elements such as collaboration and competition become essential to achieve robust levels of intelligence.

When we evaluate collaboration and competition in the context of multi-agent AI environments, there are a few existential questions that fascinate AI researchers: can AI agents learn to collaborate and/or compete among themselves? How about with humans? This week, Google’s subsidiary DeepMind published a new study illustrating the emergence of competitive and collaborative behavior of AI agents trained on the Quake III Capture the Flag multi-player game. The results are totally fascinating and show us that collaboration and competition might not be as human as we think.

Now let’s take a look at the core developments in AI research and technology this week:

Research

Google AI researchers published a paper proposing a method to improve the accuracy of convolutional neural networks using AutoML.

>Read more in this blog post from the Google AI team

AI powerhouse DeepMind published a fascinating paper showcasing how AI agents playing Quake III Capture the Flag learned to collaborate and compete with humans and other AI agents.

>Read more in this blog post from DeepMind

Microsoft AI researchers published a paper proposing an innovating technique to use bias to fight bias in AI models.

>Read more in this blog post from Microsoft Research

Cool Tech Releases

Amazon Web Services(AWS) released Textract, a new service to extract accurate data from virtually any time of document.

>Learn more about it in the AWS press release

The famous TVM compiler for deep learning applications is now integrated into PyTorch.

>Read more in this blog post from the TVM community

Udacity just published a new free course about secured and private AI

>Find the details here

AI in the Real World

British former Prime Minister David Cameron joined the board of AI solutions firm Afiniti.

>Read more in this coverage from The Guardian

Russian raises $2 billion for a state fund to focus on domestic AI companies.

>Read more in this coverage for the Moscow Times

Google is doing an aggressive push in Africa towards the adoption of AI.

>Read more in this coverage from BBC

Jesus Rodriguez

Written by

Chief Scientist, Managing Partner at Invector Labs, Angel Investor, Columnist at CIO.com, Board Member of Several Software Companies