AI Perspectives is devoted to a better understanding of AI and its impact.
all posts in series
Reflections on the progress, promise, and impact of AI.
- Sun 29 Jul 20181 Bias in AI Happens When We Optimize the Wrong Thing
Bias is a pervasive problem in AI. Only by discouraging machine learning systems from exploiting a certain bias can we expect such a system to avoid doing so.
- Sun 2 Sep 20182 For AI, translation is about more than language
Translation is about expressing the same underlying information in different ways, and modern machine learning is making incredibly rapid progress in this space.
- Wed 5 Sep 20183 Practical Guidelines for Getting Started with Machine Learning
The potential advantages of AI are many, and using machine learning to accelerate your business may outweigh potential pitfalls. If you are looking to use machine learning tools, here are a few guidelines you should keep in mind.
- Sun 16 Dec 20184 DeepMind's AlphaZero and The Real World
Using DeepMind's AlphaZero AI to solve real problems will require a change in the way computers represent and think about the world. In this post, we discuss how abstract models of the world can be used for better AI decision making and discuss recent work of ours that proposes such a model for the task of navigation.
- Wed 9 Jan 20195 Massive Datasets and Generalization in ML
Big, publically available datasets are great. Yet many practitioners who seek to use models pretrained on this data need to ask themselves how informative the data is likely to be for their purposes. Dataset bias and task specificity are important factors to keep in mind.
- Fri 25 Jan 20196 Proxy metrics are everywhere in Machine Learning
Many machine learning systems are optimized using metrics that don't perfectly match the stated goals of the system. These so-called "proxy metrics" are incredibly useful, but must be used with caution.
- Fri 24 May 20197 No Free Lunch and Neural Network Architecture
Machine learning must always balance flexibility and prior assumptions about the data. In neural networks, the network architecture codifies these prior assumptions, yet the precise relationship between them is opaque. Deep learning solutions are therefore difficult to build without a lot of trial and error, and neural nets are far from an out-of-the-box solution for most applications.
- Wed 7 Aug 20198 On the efficiency of Artificial Neural Networks versus the Brain
Recent ire from the media has focused on the high-power consumption of artificial neural nets (ANNs), yet popular discussion frequently conflates training and testing. Here, I aim to clarify the ways in which conversations involving the relative efficiency of ANNs and the human brain often miss the mark.
- Mon 30 Dec 20199 Machine Learning & Robotics: My (biased) 2019 State of the Field
My thoughts on the past year of progress in Robotics and Machine Learning.
- Fri 13 Mar 202010 The Valley of AI Trust
Particularly for safety-critical applications or the automation of tasks that can directly impact quality of life, we must be careful to avoid the valley of AI trust—the dip in overall safety caused by premature adoption of automation.