Recently Interesting Links – September 2016

by FT Labs,

A varied and inconsistent list of recently interesting items (for some definition of recent, and some definition of interesting):

chatbots are coming your way

AI / Machine Learning

“machine intelligence will sit between our actions and the world”
[http://motherboard.vice.com/en_uk/read/the-arrival-of-artificially-intelligent-beer]

“Enhance that image (just like in the movies)”
[https://twitter.com/_iest/status/771615918141034496]

DILBERT on text summarisation – “I don’t have time to read your long email. Tell me what it said.”
[http://dilbert.com/strip/2016-09-01] and, related, [http://dilbert.com/strip/2016-08-31]

Tensor Flow

“The Google Brain Team is open-sourcing #TensorFlow model code for summarization research”
[https://twitter.com/deeplearningldn/status/768557870203936771]

Udacity’s TensorFlow 101: [https://www.udacity.com/course/deep-learning–ud730]

Stanford’s TensorFlow 101: [https://twitter.com/math_rachel/status/775764096427765760]

“We show that WaveNets are able to generate speech which mimics any human voice and which sounds more natural than the best existing Text-to-Speech systems, reducing the gap with human performance by over 50%.”
[https://deepmind.com/blog/wavenet-generative-model-raw-audio/]

Chatbots

“chatbots have to compete with Google and, so far, it does not seem as if they have the same great potential”
[https://www.ft.com/content/35ce58be-6920-11e6-a0b1-d87a9fea034f – FT article]

“Their reactions showed that people may not be ready for bots that are too human-like.”
[http://digiday.com/publishers/guardian-learned-chatbots/]

The lesson here appears to be more that if you get the niche right, people will be happy to sign up to very specific topics via the chat channel.
[http://digiday.com/publishers/suns-football-chat-bot-drove-nearly-half-users-back-site/]

Learning ‘too’ well, or not well enough

“If AI learns language sufficiently well, it will also learn cultural associations that are offensive, objectionable, or harmful. At a high level, bias is meaning. “Debiasing” these machine models, while intriguing and technically interesting, necessarily harms meaning.”
[https://freedom-to-tinker.com/blog/randomwalker/language-necessarily-contains-human-biases-and-so-will-machines-trained-on-language-corpora/]

First there was HITL (human in the loop), then there was SITL (society in the loop): “we still lack mechanisms for articulating societal expectations (e.g. ethics, norms, legal principles) in ways that machines can understand. We also lack a comprehensive set of mechanisms for scrutinizing the behavior of governing algorithms against precise expectations”
[https://medium.com/mit-media-lab/society-in-the-loop-54ffd71cd802#.o2fyuptpo]

The Black Sheep Problem: analysing english texts, “black sheep” outnumbers “white sheep” by 25:1.
[http://nlpers.blogspot.co.uk/2016/06/language-bias-and-black-sheep.html?m=1]

Is already a winner, simply based on its name, but is a warning for automated classification projects…
[https://en.m.wikipedia.org/wiki/Ugly_duckling_theorem]

And not forgetting: the ‘uncanny valley’, when increasingly life-like robots go from cute to creepy
[https://en.wikipedia.org/wiki/Uncanny_valley]

Wardley Mapping

A nice and simple approach for creating a snapshot view of your current political and system architecture, and getting a sense of where it is (or is not) going.

“So it’s not just agile we are rebelling against, it’s the tyranny of misapplied doctrine.”
[https://medium.com/code-for-america/the-tyranny-of-agile-4e406c1da7fa#.2n5bob1up]

And a detailed look at the history and practice of Wardley Mapping by… Wardley himself.
[https://medium.com/wardleymaps/on-being-lost-2ef5f05eb1ec#.lnxbl9czk]

Assorted

Blockchain is not about identity
[http://thefinanser.com/2016/08/applying-blockchain-identity.html/]

Spinning LED displays: [http://s-m-l.org/]
(but probably wasted on birds: [http://io9.gizmodo.com/5881551/why-birds-dont-like-to-watch-movies])

“Combining asset, environmental & now volumetric VR, content (the capture of 4D human performances), we are well on our way to a very new unknown, a complete reinvention of story telling, education experiences & other unique media consumption opportunists”
[https://vrperception.com/2016/08/16/the-making-of-photogrammetry-to-vr-teaser-of-saint-matthew-in-the-city-aucklandnz/]

Andrew Betts calling for anyone interested in contributing some thought about paywalls
[https://twitter.com/triblondon/status/780056954550915072]