Subscribe via RSS Feed Connect on Google Plus Connect on Flickr

AI

Welcome to the protocol era

Tuesday, 9 April, 2019

The “protocol era”, in case you were wondering, is “where rapidly surfacing ideological battles over the future of A.I. protocols, centralised and decentralised internet protocols, and personal and political protocols compel us to ask ourselves who are we, what are we, what do we stand for, and what are we heading towards?” So says Holly Herndon, an American singer, who “operates at the nexus of technological evolution and musical euphoria,” as she says herself. The song Eternal is from her third album PROTO, which will be released on 10 May.


A deep learning colouriser prototype

Wednesday, 13 February, 2019

“We used the popular fast.ai and PyTorch libraries to develop our model, with an architecture and training steps inspired by Jason Antic. We trained our model based on a new set of more than 500,000 old, publicly available Singapore based images that we compiled, using a local GPU cluster with NVIDIA V100 GPUs.”

So writes Preston Lim about a project of the Data Science and Artificial Intelligence Division of GovTech Singapore. If you want to “colourise your black and white photos,” this AI will do it. Rainy Day tried it and here are the results. Impressive, eh?

Mr and Mrs RD in BW

Mr and Mrs RD in colour


AI by Mother

Monday, 28 May, 2018

Mother is the UK’s largest independent creative agency and it has offices in London, New York and Buenos Aires. The company’s Shoreditch HQ was designed by Clive Wilkinson Architects and one of the major features is a 76-metre-long concrete desk which can seat all the Mother staff. “AI Therapy” was made by Mother for @TEDTalks.

Bonus: “The AI makes a lot of mistakes at first. But it learns from its mistakes and updates its model every time it incorrectly predicts an action the human will take. Its predictions start getting better and better until it becomes so good at predicting what a human would do that we don’t need the human to do it anymore. The AI can perform the action itself.” Source: McKinsey Quarterly April 2018, The economics of artificial intelligence by AI guru Ajay Agrawal.


Google: The duplicity of Duplex

Friday, 11 May, 2018 0 Comments

On Tuesday, Google announced an Artificial Intelligence product called Duplex, which is capable of having human-sounding conversations. “We hope that these technology advances will ultimately contribute to a meaningful improvement in people’s experience in day-to-day interactions with computers,” wrote Yaniv Leviathan, Principal Engineer, and Yossi Matias, Vice President, Engineering, Google. But that’s not good enough. They did not address the moral and ethical implications of Duplex. And these are enormous. For example: What will happen to the meaning of “trust” when the synthetic voice of synthetic intelligence is made to sound human? But before we go any further, let’s listen to Duplex phoning two different businesses to make appointments.

We’re racing towards a future where machines will be able to do anything humans can do. Duplex is an important signpost on the road but people should be thinking seriously about where we’re going. During Google I/O, which ended yesterday, tech journalist Bridget Carey posed some of the questions more of us should be asking:

I am genuinely bothered and disturbed at how morally wrong it is for the Google Assistant voice to act like a human and deceive other humans on the other line of a phone call, using upspeek and other quirks of language. “Hi um, do you have anything available on uh May 3?” #io18

If Google created a way for a machine to sound so much like a human that now we can’t tell what is real and what is fake, we need to have a talk about ethics and when it’s right for a human to know when they are speaking to a robot. #io18

In this age of disinformation, where people don’t know what’s fake news… how do you know what to believe if you can’t even trust your ears with now Google Assistant calling businesses and posing as a human? That means any dialogue can be spoofed by a machine and you can’t tell.

Speak now or forever hold your peace.


Deep learning for the kids with Lobe

Friday, 4 May, 2018 0 Comments

“Lobe is an easy-to-use visual tool that lets you build custom deep learning models, quickly train them, and ship them directly in your app without writing any code. Start by dragging in a folder of training examples from your desktop. Lobe automatically builds you a custom deep learning model and begins training. When you’re done, you can export a trained model and ship it directly in your app.”

Just a decade ago, the algorithmic complexity Lobe offers sounded like science fiction. Now, it is freely available for anyone with a computer and Lobe founder Mike Matas says the visual interface is so user-friendly that children can understand it. So, c’mon young and old, it’s time “to build custom deep learning models, quickly train them, and ship them directly in your app without writing code.”

Note: If you want to talk the deep learning talk and walk the deep learning walk, there’s no avoiding Python and cloud, calculus and linear algebra, however. “How to learn Deep Learning in 6 months” provides some useful pointers.


Clearing history at Facebook

Thursday, 3 May, 2018 0 Comments

Facebook’s F8 Developer Conference has ended and the announcements that captured most attention were a cheaper Oculus Go headset, enhanced Instagram Stories and dating. The latter gave rise to much mirth since Facebook is always vigilant when it comes to relationships, data and not doing harm. Right?

The really big announcement was underreported, though. It’s the upcoming “Clear history” functionality and Mark Zuckerberg posted about it himself:

“In your web browser, you have a simple way to clear your cookies and history. The idea is a lot of sites need cookies to work, but you should still be able to flush your history whenever you want. We’re building a version of this for Facebook too. It will be a simple control to clear your browsing history on Facebook — what you’ve clicked on, websites you’ve visited, and so on.”

Note 1: Facebook is using your Instagram photos to train its AIs: “Using Instagram images that are already labeled by way of hashtags, Facebook was able to collect relevant data and use it to train its computer vision and object recognition models.”

Note 2: WhatsApp CEO Jan Koum is exiting because of “privacy issues” and Cambridge Analytica is closing, citing “loss of business.”

Facebook has become synonymous with creative disruption, er, destruction.


London and Macedonia connected by Cognism

Sunday, 29 April, 2018 0 Comments

The London Co-Investment Fund is managed by Funding London and Capital Enterprise. It has raised £25 million from the Mayor of London’s Growing Places Fund to co-invest in so-called “seed rounds” (an offering in which an investor invests capital in exchange for an equity stake in the company) between £250,000 and £1,000,000. A recent beneficiary is the Macedonian AI startup Cognism, which has its development team in Skopje and its sales force in London, while the CTO, Stjepan Buljat, is based in Croatia.

Cognism develops AI tools for finding sales and recruitment leads, and the new funds will be invested in improving the company’s data research, upgrading its technology and expanding the teams in Skopje and London as well as opening an office in the US. By the way, the company says its sales intelligence is also GDPR compliant.

Note: The General Data Protection Regulation (GDPR) proposed by the European Commission will unify data protection for individuals within the European Union and address the export of personal data outside the EU.


Do You Trust This Computer?

Monday, 9 April, 2018 0 Comments

Courtesy of Elon Musk, new documentary about AI titled Do You Trust This Computer? was streamed for free over the weekend. The film explores the role of artificial intelligence in all aspects of modern society, and features commentary from educator Jerry Kaplan, scientist Rana el Kaliouby, entrepreneur Andrew Ng, investor Shivon Zilis, roboticist Hiroshi Ishiguro and screenwriter Jonathan Nolan.

“Director Chris Paine and his team have done an amazing job with this movie. It’s a very important subject that will affect our lives in ways we can’t even imagine — some scary, some good,” said Musk in an announcement. The founder of Tesla and SpaceX is known for his dark outlook on artificial intelligence and he warns that tyrants of the past were hindered by the fact they were human, a limitation not shared by supercomputers. “You would have an immortal dictator from which we can never escape,” he says in the documentary. Musk says that we need to assimilate machine learning before we are overtaken by it.

Trivia: At 1:02:00, Alexander Nix, then CEO of Cambridge Analytica, makes an appearance saying that US voters need “a persuasion message… and it needs to be nuanced.” The candidate he was selling? Ted Cruz.


Twitter thread on AI and FB

Friday, 6 April, 2018 0 Comments

Note: A thread on Twitter is a series of connected Tweets from one person. With a thread, you can add updates, context and background by connecting multiple Tweets together.

François Chollet constructs exemplary Twitter threads. A software engineer and artificial intelligence researcher at Google, he’s the creator of Keras, a leading deep learning framework for the Python programming language, and he has a new book out, Deep Learning with Python. In other words, he knows his AI, and he knows how Facebook uses AI to achieve its ends. Chollet’s Twitter thread from 21 March is informative and disturbing. Highlights:

The problem with Facebook is not *just* the loss of your privacy and the fact that it can be used as a totalitarian panopticon. The more worrying issue, in my opinion, is its use of digital information consumption as a psychological control vector.

We’re looking at a powerful entity that builds fine-grained psychological profiles of over two billion humans, that runs large-scale behavior manipulation experiments, and that aims at developing the best AI technology the world has ever seen. Personally, it really scares me

Twitter thread

And this is a powerful call to arms by Chollet: “If you work in AI, please don’t help them. Don’t play their game. Don’t participate in their research ecosystem. Please show some conscience”


IBM THINKS of AI for mental health

Thursday, 22 March, 2018 0 Comments

One of the highlights of the IBM inaugural THINK Conference, which ends today in Las Vegas, was the announcement of what Big Blue describes as the world’s smallest computer. It’s smaller than a grain of salt, will cost less than 10 cents to make and can monitor, evaluate and act on data. It also packs several hundred thousand transistors into a tiny footprint to act as a crypto anchor technology.

Along with the tiny computer, the other headline-grabber at THINK was the list of predictions covering five technologies that IBM researchers believe will transform the world over the next five years. Arvind Krishna, director of IBM Research, lists them in a blog post titled The Power of Thinking Big: IBM Research’s ‘5 in 5’. There’s hyperimaging, macroscopes, medical “labs on a chip” and smart sensors, but the most fascinating idea is the potential role of AI in language analysis. Not the usual “instant translation”, this time, but the idea that what we say and write could be assessed by cognitive systems as indicators of our mental health and physical well-being. In theory, the patterns in our speech and writing could enable healthcare professionals to track mental illness and detect degenerative neurological diseases.

As Zadie Smith said, “The past is always tense, the future perfect.”


Kersti Kaljulaid and Sophia talk weaponized AI

Friday, 16 February, 2018 0 Comments

The organizers of this year’s Munich Security Conference decided they’d try something novel for the pre-event titled “The Force Awakens: Artificial Intelligence & Modern Conflict”, so they put Sophia centre stage and had her do the introductions. Hanson Robotics, Sophia’s creator, describe her as their “most advanced robot” and for many last night this was their first opportunity to see a chatty bot in action.

The verdict? Unimpressive. The quality of Sophia’s audio output was sub-standard, but much worse was her language. The Munich Security Conference is an annual gathering of a global elite that’s comfortable with the global lingua franca but those in charge of Sophia’s speech rhythms ignored that fact that speed does not always equal progress. Her pace of delivery was way too fast for even most native speakers present. Earlier this week in the Financial Times, Michael Skapinker posited that “Europe speaks its own post-Brexit English” and he claimed that this so-called “Eurish” is a mix of “romance and Germanic influences — and no tricky metaphors”, but Sophia, clearly, does not read the FT and neither do those in charge of her interaction with the real world. Skapinker’s “Eurish” is mostly imaginary but chatbot programmers would do well to slow the pace of delivery, simplify the vocabulary and go easy with the metaphors.

That aside, the real star of the show was Kersti Kaljulaid, President of Estonia. Her English was perfectly attuned to the wavelength of the audience and her knowledge of both artificial intelligence and modern conflict was extraordinary. Then again, she would be familiar with both topics as Estonia is a leader in digital transformation and the 2007 Russian cyber-attack on Estonia was a sign of the dangerous new world we now share with the ruthless regimes in Moscow, Beijing and Teheran. Kersti Kaljulaid is on the front line and we are lucky that she understands the grave nature of the threats posed by AI in the hands of those who wish to destroy the civilization and the society she represents so eloquently and so knowledgeably.

Sophia