(cover image from Powerful Medical)
- Traditional recruiting isn't enough: How AI is changing the rules in the human capital market: The HR industry was an early adopter of machine learning technologies to improve and automate talent acquisition. One of the toughest challenges is to find and engage passive candidates but now AI bots crawl the web, looking for the best match or candidates that are likely to switch jobs.
- Algorithm better at diagnosing pneumonia than radiologists: More and more, machine learning models become better than humans at specific tasks and those involving image analysis are usually very good candidates. Stanford researchers have trained a model that detects medical conditions from X-rays and can diagnose pneumonia with more accuracy than radiologists.
- As China marches forward on AI, the White House is silent: China bets high on AI and challenges America's lead on the topic, so many worry that there seems to be little reaction or concern from the US government.
- How To Become A Centaur: A very engaging paper making the case for Intelligence Augmentation, or how AI works best as a partner complementing human abilities and not a technology that can and will eventually replace us.
- How I shipped a neural network on iOS with CoreML, PyTorch, and React Native: The design and implementation of an iOS app that tracks the accuracy of mechanical watches using machine learning, from ideation to early experiments to full execution. A lot to learn from this very detailed walkthrough with many examples and explanations.
- Cloud TPU machine learning accelerators now available in beta: Tensor Processing Units, the specialized hardware built to speed up ML workloads, are now available on the Google Cloud Platform.
- Enterprise Bot Architecture: A great article that discusses the typical issues that bot deployments face in an enterprise environment and suggests a sample architecture to mitigate them.
- Analysing bot to human hand-off interactions: Chatbots have their own limits and can't always fulfill every user request, so the best implementations provide a hand-off mechanism to have a human seamlessly join the conversation when necessary. This extension to the Ibex dashboard enables bot administrators to analyse and optimize these particular situations.
- QnAMaker sync library v1 and QnAMaker dialog v3: Gary Pretty has updated his custom dialog that faciliates the integration of the QnAMaker service with a chatbot based on the Microsoft Bot Framework. He also introduced a new QnaMaker sync library that automates the update of your QnA knowledge base when the source data changes.