Last Week in AI

Source: Deep Learning on Medium

Every week, my team at Invector Labs publishes a newsletter that covers 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: The Most Controversial AI of the Year

Earlier this year, artificial intelligence(AI) research lab OpenAI made headlines when it unveiled an AI model so great at creating fake news that many people thought it was too dangerous to release it. Named GPT-2, the model uses hundreds of millions of attributes for generating fake text that is indistinguishable from real news. is Despite the criticism, OpenAI proceeded to release several updates to the model including its latest one that leverages 1.5 billion parameters.

The controversy surrounding GPT-2 is far from trivial. While it is scary what bad actors can do manipulating the GPT-2 code, we should also think that the model can help with the creation of algorithms for detecting fake news. After all, those type of algorithms need massive labeled datasets for training and what better way to create those datasets than by using the best model at generating fake news out there. Certainly controversial.

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

AI Research

GPT-2 is Here Again

OpenAI released the latest version of its controversial GPT-2 model that can generate incredibly convincing fake news.

>Read more in this blog post from OpenAI

Fake News vs. Satire

Amazon George Washington University AI researchers published a paper proposing a method that can differentiate between fake news and satire.

>Read more in this article from ZDNet

Redefining Fairness in the Time of Machine Learning

Harvard Business Review published an interesting analysis of how machine learning is pushing us to redefine our notions of fairness.

>Read more in this article from Harvard Business Review

A Benchmark for Evaluating Visual Representations

Google AI researchers published a paper proposing an evaluation protocol for visual representation tasks.

>Read more in this blog post from Google AI Research

Cool AI Tech Releases

Azure Cognitive Services Gets an Upgrade

Microsoft unveiled new search, language and speech capabilities for its Cognitive Services suite.

>Read more in this article from VentureBeat

A New Performance Inference Benchmark

MLPerf released a new performance benchmark for inference tasks in natural language processing and vision models.

>Read more in this blog post from MLPerf

A Knowledge Network for your Company

Microsoft introduced Project Cortex, an Office365 platform that leverages machine learning to organize the content across your organization.

>Read more in this blog post from the Cortex product team

AI in the Real World

AI Ethics for the Pentagon

The Defense Innovation Bard which is composed by prominent technologists, unveiled a group of AI ethics principles as a recommendation to the US Department of Defense to be used in combat and non-combat systems.

>Read more in this coverage from VentureBeat

AI at Nasdaq

Nasdaq is using deep learning systems to detect suspicious behavior.

>Read more in this article from MIT Technology Review

Finding New Ways to Inject Creativity into AI

Quanta Magazine published a fascinating article about the different ways in which AI is pushing the boundaries of human intelligence.

>Read the complete article from Quanta Magazine