China's AI industry is tanking

In Q2 2018, Chinese investors sank $2.87b into AI startups; in Q2 2019, it was $140.7m. Read the rest

AI is like a magic trick: amazing until it goes wrong, then revealed as a cheap and brittle effect

I used to be on the program committee for the O'Reilly Emerging Technology conferences; one year we decided to make the theme "magic" -- all the ways that new technologies were doing things that baffled us and blew us away. Read the rest

Computerphile explains the fascinating AI storyteller, GPT-2

GPT-2 is a language model that was trained on 40GB of text scraped from websites that Reddit linked to and that had a Karma score of at least two.  As the developers at OpenAI describe it, GPT-2 is "a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization—all without task-specific training." Because the model is probabilistic, it returns a different response every time you enter the same input.

OpenAI decided not to release the 40GB-trained model, due to "concerns about malicious applications of the technology" but it released a 345MB-trained model which you can install as a Python program and run from a command line. (The installation instructions are in the DEVELOPERS.md file.) I installed it and was blown away by the human-quality outputs it gave to my text prompts. Here's an example - I prompted it with the first paragraph of Kafka's The Metamorphosis. And this is just with the tiny 345MB model. OpenAI published a story that the 40G GPT-2 wrote about unicorns, which shows how well the model performs.

In this Computerphile video, Rob Miles of the University of Nottingham explains how GPT-2 works. Read the rest

Make: a machine-learning toy on open-source hardware

In the latest Adafruit video (previously) the proprietors, Limor "ladyada" Friend and Phil Torrone, explain the basics of machine learning, with particular emphasis on the difference between computing a model (hard) and implementing the model (easy and simple enough to run on relatively low-powered hardware), and then they install and run Tensorflow Light on a small, open-source handheld and teach it to distinguish between someone saying "No" and someone saying "Yes," in just a few minutes. It's an interesting demonstration of the theory that machine learning may be most useful in tiny, embedded, offline processors. (via Beyond the Beyond) Read the rest

Using machine learning to pull Krazy Kat comics out of giant public domain newspaper archives

Joël Franusic became obsessed with Krazy Kat, but was frustrated by the limited availability and high cost of the books anthologizing the strip (some of which were going for $600 or more on Amazon); so he wrote a scraper that would pull down thumbnails from massive archives of pre-1923 newspapers and then identified 100 pages containing Krazy Kat strips to use as training data for a machine-learning model. Read the rest

This machine learning-assisted cat door keeps kitty from bringing furry surprises home

Using an Arduino, a bunch of code and a little machine learning, Benn Hamm created a cat door to keep his cat from bringing dead--and sometimes live--rats and birds into his home in the middle of the night. It's not often that I'm down with bringing surveillance technology into homes but, as a former cat owner who's had to clean bird shit off a flat-screen TV, I have nothing but love for this project.

Image via Wikipedia Commons Read the rest

Police cameras to be augmented with junk-science "microexpression" AI lie-detectors

The idea that you can detect lies by analyzing "microexpressions" has absorbed billions in spending by police forces and security services, despite the fact that it's junk science that performs worse than a coin-toss. Read the rest

Rage Inside the Machine: an insightful, brilliant critique of AI's computer science, sociology, philosophy and economics

Rob Smith is an eminent computer scientist and machine learning pioneer whose work on genetic algorithms has been influential in both industry and the academy; now, in his first book for a general audience, Rage Inside the Machine: The Prejudice of Algorithms, and How to Stop the Internet Making Bigots of Us All, Smith expertly draws connections between AI, neoliberalism, human bias, eugenics and far-right populism, and shows how the biases of computer science and the corporate paymasters have distorted our whole society. Read the rest

EU expert panel calls for a ban on AI-based risk-scoring and limits on mass surveillance

The EU Commission's High-Level Expert Group on AI (AI HLEG) has tabled its Policy and investment recommendations for trustworthy Artificial Intelligence, recommending a ban on the use of machine learning technologies to generate Chinese-style Citizen Scores and limits on the use of the technology in monitoring and analyzing mass surveillance data. Read the rest

Analog computers could bring massive efficiency gains to machine learning

In The Next Generation of Deep Learning Hardware: Analog Computing *Sci-Hub mirror), a trio of IBM researchers discuss how new materials could allow them to build analog computers that vastly improved the energy/computing efficiency in training machine-learning networks. Read the rest

Independent evaluation of "aggression detection" microphones used in schools and hospitals finds them to be worse than useless

One of the griftiest corners of late-stage capitalism is the "public safety" industry, in which military contractors realize they can expand their market by peddling overpriced garbage to schools, cities, public transit systems, hospitals, etc -- which is how the "aggression detection" industry emerged, selling microphones whose "machine learning" backends are supposed to be able to detect "aggressive voices" (as well as gunshots) and alert cops or security guards. Read the rest

Finally, a useful application for augmented reality: rendering virtual kitchen roaches

Laanlabs's showreel for 6d.ai meshing technology is an augmented reality demo in which virtual cockroaches crawl all over a very real kitchen. It's the best use of augmented reality I've ever seen. (via Beyond the Beyond) Read the rest

Machine learning classifiers are up to 20% less accurate when labeling photos from homes in poor countries

A new study from Facebook AI Research evaluates common machine-learning classifiers' ability to label photos of objects found in households in rich countries versus household objects from poor countries and finds that the models' performance lags significantly when being asked to classify the possessions of poor people. Read the rest

Training a modest machine-learning model uses more carbon than the manufacturing and lifetime use of five automobiles

In Energy and Policy Considerations for Deep Learning in NLP, three UMass Amherst computer science researchers investigate the carbon budget of training machine learning models for natural language processing, and come back with the eyepopping headline figure of 78,468lbs to do a basic training-and-refinement operation. Read the rest

DARPA's Spectrum Collaboration Challenge: finally some progress towards a "Cognitive Radio" future

For 17 years, I've been writing about the possibilities of "cognitive radio", in which radios sense which spectrum is available from moment to moment and collaborate to frequency-hop (and perform other tricks) to maximize the efficiency of wireless communications. Read the rest

Collecting user data is a competitive disadvantage

Warren Buffet is famous for identifying the need for businesses to have "moats" and "walls" around their profit-centers to keep competitors out, and data-centric companies often cite their massive collections of user-data as "moats" that benefit from "network effects" to make their businesses good investments. Read the rest

The Training Commission: an email newsletter from the future, after a civil war and digital blackout

"The Training Commission" is Ingrid Burrington and Brendan C Byrne's serialized science fiction tale, taking the form of an email newsletter that lets you eavesdrop on the correspondence between the story's principal characters: it's set after a civil war ("the Shitstorm"), sparked by misbehaving and easily abused machine-learning systems, and which was resolved after a protracted and catastrophic digital blackout. Read the rest

More posts