1 – Google’s New DeepDream Algorithm Could Foreshadow a Creative Artificial Intelligence Google just open sourced the codes to its engineers’ algorithm DeepDream, which was designed to examine how neural networks work. The result is a program that, it turns out, creates some wacky and whimsical art, taking ordinary images and enhancing those with patterns of images that it has learned. Could an ability to ‘create perception’ one day evolve into an AI force that shapes its own image of reality? (Full article on BloombergView) 2 – Can Merging AI with the Internet of Things Result in a Peaceful Union? AI Pioneers John Underkoffler, the chief executive of Oblong Industries, Professor Sanjay Sarma, the director of digital learning at MIT, and others believe we’re not asking the right questions when it comes to the inevitable future of merging AI and the Internet of Things. Underkoffler admits to being an optimist in this sphere:
…millions of new objects all connected to the internet – wow, to make sense of that is going to require an incredible new interface, how do we talk to all these objects in a coherent way? That’s a really great design problem.
Instead of fearing the advance of more connected AI, we should be considering the many opportunities that could advance society for the good. In order to do so, we have to ask and answer tough questions about digital architecture, customer privacy, and “artificial stupidity.” (Full article on The Guardian) 3 – Ad Campaign Uses AI to Evolve Reactions Based on Viewer M&C Saatchi has developed a new ad campaign that evolves based on viewer responses. The ad, shown in one location in London, uses a genetic algorithm based on various ad components and analyzes strengths and weaknesses. Following a similar existing design-based approach for websites, the data collected by the algorithm will be able to reveal which ad “genes” were successful and which were not, and future ads will be modified based on such input. The intent is for such technology to be use such data to virtually create ads on its own. (Full article on TechTimes) 4 – Hitachi Ltd. Develops Artificial Intelligence that Reasons Japanese-based company Hitachi Ltd. revealed an artificial intelligence software that makes “reasoned responses” to sensitive topics, supported by evidence drawn from big data. The project was launched in collaboration with researchers at Tohoku University and draws on a database of millions of articles and an index capable of 250 million correlations. A portion of the system was tested at an international forum and deemed to be more accurate in diagnosing and finding a disease than existing systems. Developers foresee users one day having logical debates with the software. (Full article on The Japan Times) 5 – Studying the Human Genome with Artificial Intelligence University of Toronto has spearheaded the new entity Deep Genomics, which will use deep machine learning capabilities with artificial intelligence to study the human genome and the potential of disease. The initiative is a new mover in the area of precision medicine. CEO Brendan Frey likens its algorithm-based approach to a “…Google search engine for genomics.” The algorithms specifically look at mutations of cells in an individual that could potentially develop into a disease. (Full article on MedCityNews)