The noticeable, the useful, the quirky: the pieces we noted as AI BRIEFS for 2019. In reverse chronology (January is at the bottom). Also available by month:  Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec December Art meets artificial intelligence: Austin…

‘#1. Scientists from the University of California, Berkeley, and Google Brain developed an AI system that taught a robot to walk.’ The newsletter INSIDE AI names its picks as the most significant AI developments of the past year. The ones…

‘Emotions and affect are outside of the domain of reason, and predictive analytics and machine learning operate based on the logic of mathematics — even if they’re asked to generate emotions, they need rational methods to do so.’ The future…

BERT is Google’s model for natural language understanding (NLU). ERNIE is Baidu’s. Their effectiveness is measured with a test known as General Language Understanding Evaluation, or GLUE for short. Language understanding is at the centre of many journalism uses of…

“Our goal was to use the most advanced artificial intelligence techniques available today to create the most believable result possible—and then point to it and say, ‘This is fake; here’s how we did it; and here’s why we did it”…

“Debiasing humans is harder than debiasing AI systems.” Olga Russakovsky

‘The paper proposes each use of AI by government should be accompanied by a cost-benefit analysis and public consultation before it is brought in.’  Every person is entitled to know how an algorithm has made a decision involving them says…

‘It’s meant to work quickly out of fear that a viral deepfake will need almost immediate identification and response.’ ‘Soft biometrics’ may help distinguish real video from imitations, says a research team at University of California, Berkeley. They’ve studied hundreds…

‘Algorithm design is an important new way for thinking about media diversity in society.’ News algorithms can be powerful expressions of journalistic values says journalism AI specialist Nick Diakopoulos, in COLUMBIA JOURNALISM REVIEW. He suggests curation algorithms align with individual…

‘The organization undertook its biggest test of machine learning-generated journalism to date, publishing nearly 700 individual news reports (649 in English, 40 in Welsh) throughout the night.’ An autoscribe pumped out BBC News election stories tailored to each UK constituency, TechHQ reports….

‘Prof Noel Sharkey, who is also a leading figure in a global campaign against “killer robots”, said algorithms were so “infected with biases” that their decision-making processes could not be fair or trusted.’ Decision-making algorithms should be treated like prescription…

‘…to benefit from the potential offered by AI technologies, news organizations will have to change. Again. They need to change their workflows, systems, and recruitment.’ This practical guide to implementing AI in newsrooms is by POLIS, the LSE journalism think-tank,…

Runs 55:02 Turing Award recipient Yoshua Bengio connects deep learning to concepts of cognition described in ‘Thinking fast and Slow,’ the book by Daniel Kahneman. Bengio says ‘System 1’ thinking is like deep learning’s current state: He suggests future deep…

Facebook AI is releasing 100,000+ videos as a training set for its $10 million challenge to help detect deep fakes. Each video presents real people with a fake counterpart. The challenge is to create automated tools that can identify deceitful…

“Agents are increasingly one of the only human connections people have with brands in a digital world, so we want to coach them to improve their performance with AI and augment them on live calls with insights that can be…

A read-to-me news app uses multiple sources and multiple voices to deliver news in a different way. AudiBrow uses AI text-to-voice software to present news from publications such as The Washington Post, The Economist, The Guardian, Reuters, BBC, The Verge,…

Finland will offer its free AI training to all EU languages starting in 2020. “It’s about one of the most pressing challenges facing Europe and Finland today: how to develop our digital literacy,” says Finnish MInister of Employment Timo Harakka…

‘Should a work produced by an AI algorithm or process, without the involvement of a natural person contributing expression to the resulting work, qualify as a work of authorship protectable under U.S. copyright law? Why or why not?’ The US…

‘…the system tracks players in real-time and can identify passes, crosses and shots on goal. For pre-selected video highlights, it can also incorporate commentary based on statistics and standings drawn from a database’ A proof-of-concept for AI-driven sports commentary was…

‘Virtual newsreaders, stock reports and election results written up by algorithms, and news organisations dishing up articles based on what you’ve clicked on are all examples (of) artificial intelligence in action.’ Journalists are ‘only scratching the surface of what’s possible’ with AI…

Art meets artificial intelligence: Austin artists explore new meaning in data | Stephenville Empire-Tribune | DEcember 31, 2019 | by Hojun Choi | ‘Morton said the new brand of AI-powered generative art forces people to look inward for meaning rather…

‘Today, reporters, as well as editors and the production staff, now must work faster than ever but getting the story out shouldn’t come at the expense of the facts.’ EDITOR & PUBLISHER surveys the field of fact-checking and provides several…

Runs 33:22 Remarks by Professor Charlie Beckett at the launch of ‘New Powers, New Responsibilities,‘ a global study of AI in journalism conducted by POLIS, a journalism think-tank at the London School of Economics. Beckett heads POLIS and wrote the…

Runs 29:41 A panel of news leaders at the launch of the POLIS report on journalism and AI, News Power, New Responsibilities. Moderated by the report author, Charlie Beckett, a journalists and professor at LSE who heads POLIS, the school’s…

‘What if you could manipulate the facial features of a historical figure, a politician, or a CEO realistically and convincingly using nothing but a webcam and an illustrated or photographic still image?’  A new technique known as facial reenactment enables…

‘Identifying potential loyal readers is beneficial not only for maximising marketing budgets, but also creating personalised reader experiences.’ Reader engagement is up at the South China Morning Post with a machine learning algorithm that tracks user loyalty, reports SCMP in…

‘…journalists need to continue to develop their technical skills to become better tool makers rather than tool users. It is in this way that we control, we shape, our relationship to AI and can perhaps create an entirely new technical…

‘Whether or not opinion columnists’ direst predictions come to fruition, it seems almost inevitable that newsrooms operating in a world of deepfakes will be forced to shoulder a heavy new burden.’ Balancing verification with timely reporting is getting tougher because…

‘The upcoming shift may well spell the end of PSM as we know it. But it could also be the start of a new era in which PSM organizations are more than ever called upon to act as curators of…

‘If we value journalism as a social good, provided by humans for humans, then we have a window of perhaps 2-5 years, when news organizations must get across this technology.’ AI systems could have ‘a wide-ranging and profound influence on…