Welcome to Nural's newsletter where you will find a compilation of articles, news and cool companies, all focusing on how AI is being used to tackle global grand challenges.
Our aim is to make sure that you are always up to date with the most important developments in this fast-moving field.
Packed inside we have
- Tech startups in the spotlight at COP26
- Nvidia announces a "digital twin" of the Earth to fight climate change
- plus, Meta, Microsoft, Nvidia and the metaverse(s)
If you would like to support our continued work from £1 then click here!
Graham Lane & Marcel Hedman
Key Recent Developments
Ten tech startups featured at COP26
What: The COP26 Tech For Our Planet challenge programme features ten tech startups using digital and data solutions to address global climate change. Successful innovators will work with the UK government on bringing solutions to market and creating a legacy towards COP27 and beyond.
Key Takeaway: Many countries are relying partly on – as yet unspecified – “new technologies” to help reduce their carbon emissions. Hopefully, these tech companies will be part of that solution.
Nvidia "digital twin" of the Earth plus climate initiatives by Microsoft and Amazon
David F. Carr
What: At COP26, Nvidia committed to build a high resolution “digital twin” of the earth, called E2. A digital twin is a simulation that can stand in for the real thing, enabling innovations and large-scale changes to be tested before they are implemented. Digital twins have been used successfully to reduce carbon emissions in industrial chemical processes. At a larger scale, E2 is intended to accurately predict the climate decades into the future and guide efforts to mitigate global heating. The article also references the Microsoft Cloud for Sustainability and various climate and sustainability initiatives from Amazon.
Key Takeaway: Besides everything else, Amazon plans to finally tackle the annoying habit of sending small items in large boxes stuffed with plastic packing.
Nvidia wants to fill the virtual and physical worlds with AI avatars
James Vincent
What: Nvidia has announced a new platform for virtual agents called Omniverse Avatar. It combines speech recognition, synthetic speech, facial tracking, and 3D avatar animation. The virtual agents can interact with human users and humans can present themselves as avatars, for example on a conference call.
Key Takeaways: You wait 20 years for a metaverse and then three come along at once. Besides Nvidia, Meta/Facebook would like immersive virtual reality to supersede the mobile phone. Others promote augmented reality (like Pokemon Go) while Microsoft is incrementally developing the Teams collaboration platform within the enterprise. There are sure to be lots of developments, but will they work together?
AI Ethics
🚀 Why we forgive humans more readily than machines
In summary, “People judge humans by their intentions and machines by their outcomes.”
🚀 Privacy preserving machine learning: Maintaining confidentiality and preserving trust
A substantive blog post from Microsoft assesses how to understand, measure and mitigate ML privacy risks.
🚀 Current AI practices could be enabling a new generation of copyright trolls
New research suggests current AI research could be exposed to litigation if using datasets distributed with invalid licenses.
Other interesting reads
🚀 Google is taking sign-ups for Relate, a voice assistant that recognizes impaired speech
Google has launched a research project so that people with speech impairments can better use speech recognition software.
🚀 Meet M6 — 10 trillion parameters at 1% GPT-3’s energy cost
Alibaba R&D announced a huge AI model built with a relatively small training effort. If released, this could democratise ML research.
🚀 AI beats elite pros at 6-player Texas Hold’em poker
Unlike chess or go, winning at poker requires the AI to operate with incomplete information and to deal with bluffing.
Cool companies found this week
Manufacturing
Landing AI - Uses computer vision to identify quality defects in manufacturing even when little training data is available. Founded by Andrew Ng, it has received $57 million in round A funding
Transport
Kodiak Robotics - specialises in self-driving trucks and has raised $125 million in round B funding
Robotics
Soft Robotics - provides an AI-based picking solution for variable, irregular, and delicate products in the food industry. Received a $10 million round B extension in June.
And finally ...
AI/ML must knows
Foundation Models - any model trained on broad data at scale that can be fine-tuned to a wide range of downstream tasks. Examples include BERT and GPT-3. (See also Transfer Learning)
Few shot learning - Supervised learning using only a small dataset to master the task.
Transfer Learning - Reusing parts or all of a model designed for one task on a new task with the aim of reducing training time and improving performance.
Generative adversarial network - Generative models that create new data instances that resemble your training data. They can be used to generate fake images.
Deep Learning - Deep learning is a form of machine learning based on artificial neural networks.
Best,
Marcel Hedman
Nural Research Founder
www.nural.cc
If this has been interesting, share it with a friend who will find it equally valuable. If you are not already a subscriber, then subscribe here.
If you are enjoying this content and would like to support the work financially then you can amend your plan here from £1/month!