In the News
Learning from videos to understand the world
Continuously learning from the world around us is one of the hallmarks of human intelligence. Just as we quickly learn to recognize people, places, things, and actions through observation, AI systems will be smarter and more useful if they can mimic the way humans learn.
Sponsor
Make your own AI models, no code required.
It’s now easier than ever to bring machine learning into your business. With Primer Automate, you can train NLP models to execute your organization’s repetitive text-based tasks, such as classifying, identifying, or summarizing, with human-level precision. Use Automate to scale up your information processing capacity, so you can focus on higher-level tasks.
In The News
Can Artificial Intelligence Replace Human Therapists?
Could artificial intelligence reduce the need for human therapists? Websites, smartphone apps and social-media sites are dispensing mental-health advice, often using artificial intelligence.
Artificial Intelligence: Blazing A Trail For The Future
However, amid its continuously evolving nature and the need for ensuring ongoing management, organizations still confront challenges when implementing AI. When combining human and machine work, companies may embrace such benefits as an accelerated working process, increased employee effectiveness...
Why AI can’t solve unknown problems
One set of numbers represents the inputs that the system receives, one set of numbers represents the outputs that the system produces, and the third set of numbers represents the machine learning model.” Therefore, while supervised machine learning is not tightly bound to rules like symbolic AI, it...
Applied use cases
Opinion: How AI can protect users in the online world
Step forward, Artificial Intelligence (AI) – advanced machine learning technology that plays an important role in modern life and is fundamental in how today’s social networks function. Advanced safety features AI has proven to be an effective weapon in the fight against online harassment and the...
Machine learning meets creative content: How new tools are taking the drag out of creativity and inventing new genres
Not only do they make content production easier, they take the drag out of the creative process itself. Machine Learning tools are reinventing the creative process There are two types of creative tools. Synthesia started out attempting to make video production easier and faster, but soon realised it...
Alignment of Language Agents
We categorise the ways a machine learning task can be misspecified, based on whether the problem arises from the training data, the training process itself, or from a distributional shift (i.e. Training process misspecification can occur when a learning algorithm designed for solving one kind of...
Ethics
Facebook gets a C; startup rates the ‘ethics’ of social media platforms, targets asset managers
We’ve built a model to analyze technology governance. We’ve covered 20 industries. So most of what we’ve published so far has been non-tech companies because these are risks that are inherent in many other industries, other than simply social media or big tech. But over the next couple of weeks, we’re going live with our data on things which are directly related to tech, starting with social media.
AI: Ghost workers demand to be seen and heard
The most well-established of these crowdsourcing platforms is Amazon Mechanical Turk, owned by the online retail giant and run by its Amazon Web Services division. These tasks could be anything from labelling images to help computer vision algorithms improve, providing help for natural language...
How easy would it be to snuff out humanity?
Without it the world can’t provide sustainable energy, nourishing food and good health for an expanding and more demanding population. And what about another transformative technology: robotics and artificial intelligence (AI)?
Robotics
Boston Dynamics’ New Robot Doesn’t Dance. It Has a Warehouse Job
Boston Dynamics’ New Robot Doesn’t Dance. It Has a Warehouse Job Called Stretch, the machine may look wildly different than its famous cousins Spot and Atlas, but it shares a ton of their DNA.
BMW’s IdealWorks Validates (and Challenges) AMR Sector
With the availability of excellent 3D cameras, increasingly affordable LiDAR, more processing hardware and an open-source ecosystem anchored by Robot Operating System (ROS) and SDKs like NVIDIA ISAAC, can any team of half-decent roboticists enter the market?
Boston Dynamic's new 'Stretch' robot is designed for boxes, not backflips
Boston Dynamics has revealed its latest robot, Stretch, designed for far less glamorous purposes than its other creations. Where Spot and Atlas grabbed headlines with backflips and dancing, Stretch is designed to move boxes in warehouses, TechCrunch has reported.
Research
Introducing the NeurIPS 2021 Paper Checklist
In 2019, NeurIPS introduced a reproducibility program, consisting of a code submission policy, a community-wide reproducibility challenge, and the inclusion of a reproducibility checklist as part of the paper submission process.
Graph Transformer: A Generalization of Transformers to Graphs
This blog is based on the paper A Generalization of Transformer Networks to Graphs with Xavier Bresson at 2021 AAAI Workshop on Deep Learning on Graphs: Methods and Applications (DLG-AAAI’21).
We present Graph Transformer, a transformer neural network that can operate on arbitrary graphs.
When are Neural Networks more powerful than Neural Tangent Kernels?
The empirical success of deep learning has posed significant challenges to machine learning theory: Why can we efficiently train neural networks with gradient descent despite its highly non-convex optimization landscape? Why do over-parametrized networks generalize well?