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 Great Race for Cloud AI Dominance and its Impact in the AI Startup Ecosystem.
Amazon, Microsoft and Google have embarked in a frantic race to dominate the artificial intelligence(AI) market. To some extent, AI has become one of the few areas of differentiation between the three cloud giants. Every relevant capability of AI applications from experimentation to optimization can be found in the form of cloud services in these platforms. Just this week, Amazon unveiled a new batch of AI services for the AWS platforms during its re:Invent conference.
The massive investments that the cloud giants are making in the AI ecosystem is certainly pushing innovation in the space to a new level but it’s also having tangible side effects in the AI startup ecosystem. In the quest for capturing the best AI talent in the market, Microsoft, Amazon and Google have begun rapidly acquiring AI startups that show any signs of promise. The result is that most AI technology startups haven’t had the opportunity to flourish as independent companies compacting the innovation in the space. This is certainly an atypical trend compared to previous technology markets and one that certain impacts the way investors and customers seem the AI startup ecosystem.
Now let’s take a look at the core developments in AI research and technology this week:
AI in Cooperative Games
Facebook AI Research(FAIR) published a paper describing an AI model that can achieve super human performance in Hanabi, a card game in which players must work together.
The Deep Double Descent Phenomenon
OpenAI published a paper describing a common phenomenon in deep neural networks in which performance first improves, then gets worse, and then improves again with increasing model size, data size, or training time.
Teaching Neural Networks to Remember
Microsoft Research published a paper proposing a method called Metalearned Neural Memory which uses a neural network itself as a memory mechanism.
Cool AI Tech Releases
A Human Centric Framework for Data Science
Netflix open sourced Metaflow, a new framework for managing the lifecycle of data science projects.
AutoML on AWS
AWS unveiled SageMarket AutoPilot, to bring interpretability to AutoML models.
A New IDE for Machine Learning
AWS announced SageMarker Studio, a new IDE for machine learning applications.
AI in the Real World
China vs. The US on AI Investments
China investments on AI seem a bit overblown based on new numbers published this week.
AI Assistant Returns to Space
CIMON-2, the second version of the famous floating robot for astrounant assistance, was sent to space as part of the SpaceX ISS resupply mission.
The Dilemma Facing Chinese US-Educated AI Engineers
Many Chinese experts have great opportunities in the US or can return home to be part of China’s grand AI strategy.