Data Science Insider: January 15th, 2021

In this week’s Super Data Science newsletter: DASA Awards £3m to Fund New Wave of Military AI. South Korean Chatbot Suspended for Hate Speech. Neural Networks and Concept Whitening. Scientists Warn That Containment Algorithms Won’t Stop Super-Intelligent AI. Premiership Club Uses AI for New Signings. Cheers, - The SuperDataScience Team P.S. Have friends and colleagues who could benefit from these weekly updates? Send them to [this link]( to subscribe to the Data Science Insider. --------------------------------------------------------------- [DASA Awards £3m to Fund New Wave of Military AI]( brief: The UK’s Defence and Security Accelerator (DASA) has awarded £3m in funding for its Intelligent Ship competition to improve decision-making, mission planning, and automation using AI. The competition, led by DASA on behalf of the Ministry of Defence’s (MOD) Defence Science Technology Laboratory (Dstl), seeks to develop technologies for naval vessels from 2030 onward. The £3m funding was split between nine projects from companies including Rolls Royce, Nottingham Trent University, and Decision Lab. The other funded companies are CGI IT UK, DIEM Analytics, Frazer Nash Consultancy, Montvieux, and SeeByte. Funded proposals include a decision-making support system for engineering crews to manage power and propulsion systems and an ‘innovative mission AI prototype agent for decision-making to support decision-making during pre-mission preparation, mission execution, and post-mission analysis.’ DASA delivery manager Rachel Solomons said: “DASA is focussed on finding innovation to benefit the defence and security of the UK." Why this is important: In January, DASA announced the competition's first wave of funding with nine projects each receiving a share of £1m but this latest round of funding is from phase 2 of the competition, which is set to develop broader applications for AI across defence. [Click here to find out!]( [South Korean Chatbot Suspended for Hate Speech]( brief: A popular South Korean AI-driven chatbot with the persona of a 20-year-old female student was taken down this week after it was accused of bigotry towards sexual minorities, the #MeToo movement and the disabled. Lee Luda, developed by Seoul-based start-up Scatter Lab to operate within Facebook Messenger, became an instant sensation for her spontaneous and natural reactions, attracting more than 750,000 users after its launch late last month. Luda's AI algorithms learnt from data collected from 10 billion conversations on Kakao Talk, the country's top messenger app. But the chatbot was rapidly embroiled in a spate of allegations that it used hate speech towards minorities, triggering a controversy that eventually forced the developer to take it offline. In one of the captured chat shots, Luda said she "despised" gays and lesbians. In her remarks about people with disabilities, Luda said she would "rather die" than live as a handicapped person. Why this is important: The comments stem from the database of billions of conversations that the AI programme learned from, proving once again that the technology is only as good as the data it has to work with. The controversy is similar to Microsoft’s Tay, which was taken offline after manipulating users into posting racist tweets. [Click here to read on!]( [Neural Networks and Concept Whitening]( In brief: The inner workings of neural networks have confounded the AI community since the boom in popularity of deep neural networks in the early 2010s. Many techniques have been developed to try and explain but they’re frequently found to be lacking in guidance to fix issues embedded in deep learning models at the training stage. This article from Tech Talks argues attempts to see inside NNs hidden layers can be misleading, unusable, or rely on the latent space to possess properties that it may not have. Instead, it argues, introducing a mechanism called concept whitening to alter a given layer of the network allows us to better understand the computation leading up to that layer. It states that when a concept whitening module is added to a convolutional neural network, the latent space is whitened and the axes of the latent space are aligned with known concepts of interest. Why this is important: As data scientists, we strive to understand the inner workings of neural networks alongside their potential applications. The idea of concept whitening “proves that we can impose top-down design constraints on neural networks without causing any performance penalties,” a fascinating shift in traditional theories that could have huge implications on deep learning research. [Click here to discover more!]( [Containment Algorithms Won’t Stop Super-Intelligent AI]( In brief: A team of computer scientists has used theoretical calculations to argue that algorithms could not control a super-intelligent AI. Their study addresses the question: How do we ensure super-intelligence machines act in our interests? The researchers conceived of a theoretical containment algorithm that would resolve this problem by simulating the AI‘s behavior, and halting the program if its actions became harmful. Their analysis found that it would be fundamentally impossible to build an algorithm that could control such a machine. The study found that no single algorithm could calculate whether an AI would harm the world, due to the fundamental limits of computing, Iyad Rahwan, Director of the Center for Humans and Machines stated: “Assuming that a superintelligence will contain a program that includes all the programs that can be executed by a universal Turing machine […] strict containment requires simulations of such a program, something theoretically (and practically) impossible.” Why this is important: This type of AI remains confined to the realms of fantasy — for now. But the researchers note that tech is making strides towards the type of super-intelligent systems envisioned by science fiction writers. The question, therefore, arises whether this could at some point become uncontrollable and dangerous for humanity. [Click here to see the full picture!]( [Premiership Club Uses AI for New Signings]( In brief: Premiership football club Burnley have announced they are to use AI to help them identify young talent. The team known as the Clarets have partnered with talent identification platform AiSCOUT and launched an “open global talent search” in which they are looking to unearth players over the age of 14 that could join their academy. Players can use a free mobile app to upload footage of themselves performing specific individual drills, first physical and then technical. Scores will be produced from that via AiSCOUT’s AI capability, and the information can be used by scouts. The club say “exceptional” players identified through the app will be invited to attend a formal academy trials day later this year, with the chance to earn a place in the youth system. The Delaware-based ALK Capital investment group, which completed a takeover of Burnley on New Year’s Eve, invested in AiSCOUT last August. Why this is important: Although human scouts are still being used in conjunction with the app, the ability to upload footage that AI then ranks means that football is likely to become more accessible to a wider-range of young players, who have dreamed of being scouted. [Click here to find out more!]( [SuperDataScience podcast]( In this week's [SuperDataScience Podcast](, Erica Greene joins us to discuss her day-to-day and higher-level work as a machine learning manager at Etsy, including feature drift, prioritizing problems to solve, and scaling machine learning. --------------------------------------------------------------- What is the Data Science Insider? This email is a briefing of the week's most disruptive, interesting, and useful resources curated by the SuperDataScience team for Data Scientists who want to take their careers to the next level. Want more conversations like this? Earlier this year, we held our first-ever DSGO Virtual Conferences, where more than 3,500 data scientists gathered to learn, grow, and connect! If you missed them or want to repeat this fantastic experience, stay tuned to our upcoming virtual and in-person events that will take your DS career to the next level. DSGO is your go-to place to elevate your technical skills, gain life-long career lessons from industry experts, and build lasting connections with data-driven peers. If you want to learn more and register for our future events, [click here](. Know someone who would benefit from getting The Data Science Insider? Send them [this link to sign up.]( # # If you wish to stop receiving our emails or change your subscription options, please [Manage Your Subscription]( SuperDataScience Pty Ltd (ABN 91 617 928) 131, 63 Blamey, St., Kelvin Grove, QLD 4059, Australia