Google discusses new reinforcement learning model in new “off-policy classification” paper

A team of AI researchers at Google has recently published a paper titled “Off-Policy Evaluation via Off-Policy Classification” on its blog. The paper talks about “off-policy classification” or OPC -- as the researchers call it -- which assesses the performance of AI-driven agents by treating evaluation as a classification problem.

The team says that their approach, which involves a variant of reinforcement learning that uses rewards to drive software policies toward goals,...

21 June 2019 | Google

Google’s Duplex booking AI often relies on humans for backup

Google Duplex often calls on humans for backup when making reservations on behalf of users, and that should be welcomed.

Duplex caused a stir when it debuted at Google’s I/O developer conference last year. The AI was shown calling a hair salon to make a booking and did so complete with human-like “ums” and “ahs”.

https://www.youtube.com/watch?v=-qCanuYrR0g

The use of such human mannerisms goes to show Google’s intention was for the human to be unaware...

23 May 2019 | Bots

Google condenses AI to just 0.5GB for speedy on-device Assistant

During this year's I/O Keynote, Google announced it's managed to condense 100GB of AI to just 0.5GB for a drastically sped-up Assistant.

Google's Assistant has always been among the most impressive, but it's not exactly been known for its speed. Often it felt quicker to just do the task manually rather than wait for Assistant to respond.

The main reason for Assistant’s sluggishness, it transpires, is that it required around 100GB of storage in addition to an internet...

7 May 2019 | Google

Google axes its AI ethics board after less than a week

Google has axed its troubled AI ethics board less than a week after it had been created.

The board appeared fairly representative of society from the outset, although perhaps too much so. Some views shouldn't be represented.

In a statement, Google wrote:

"It's become clear that in the current environment, ATEAC can't function as we wanted. So we're ending the council and going back to the drawing board.

We'll continue to be responsible in our work on the...

5 April 2019 | Ethics

Google’s AI ethics panel is already receiving backlash

Google is already facing criticism of its AI ethics panel over featuring a conservative figure with some controversial views.

Kay Editor is the president of conservative think tank the Heritage Foundation. The foundation has a history of opposing LGBTQ and immigrant rights which many Googlers feel should not be promoted.

A group of Google employees which calls itself ‘Googlers Against Transphobia’ wrote a post criticising Mountain View’s appointment of the Heritage...

2 April 2019 | Ethics

DeepMind’s first commercial product diagnoses eye diseases

DeepMind is preparing to launch its first commercial product and it aims to aid the diagnosis of complex eye diseases.

The renowned Cambridge-based artificial intelligence experts hope their product will become parent company Alphabet’s first medical device.

In a live demonstration, a patient had a retinal scan performed on her eye. DeepMind’s system was able to provide a diagnosis and ‘urgency score’ in just 30 seconds.

The ability to perform such a...

1 April 2019 | DeepMind

Don’t be evil: Google is creating a dedicated AI ethics panel

Google is aiming to prevent societal disasters caused by its AI technology with the creation of a dedicated ethics panel.

The panel is called the Advanced Technology External Advisory Council (ATEAC) and features a range of academics and experts from around the world.

Eight people are currently on the panel, with some from as far as Hong Kong and South Africa. Among the roster is former US deputy secretary of state William Joseph Burns and University of Bath associate...

27 March 2019 | Ethics

Become a modern-day Bach with Google’s AI-powered doodle

Google is honouring the 334th birthday of famous German composer Johann Sebastian Bach with an AI-powered ‘doodle’ that mimics his musical style.

Users can input their own melody and the AI will create a harmony in the Baroque style of Bach.

“Bach was a humble man who attributed his success to divine inspiration and a strict work ethic,” wrote Google in a post. “He lived to see only a handful of his works published, but more than 1,000 that survived in...

21 March 2019 | Google

Google is using AI to predict wind farm power output

Google is using AI to solve the problem of predicting the power output of a wind farm ahead of time.

Wind farms are a fantastic source of renewable energy but their output can be hard to predict due to the nature of wind.

By improving the accuracy of predictions and providing more data, wind farm operators are better prepared to meet upcoming power demands.

In a blog post, Google wrote:

“Based on these predictions, our model recommends how to make...

27 February 2019 | Google

UK investment in AI startups nears the rest of Europe combined

The UK is splashing the cash on artificial intelligence startups so much that it almost totals the rest of Europe combined.

Venture capital firms invested a record $1.3bn (£998m) in UK-based AI companies last year.

The figures are provided by Dealroom and show a fourfold increase in five years. A couple of notable rounds includes $200 million for Graphcore in your humble editor’s hometown of Bristol, and $50 million for what feels like my second home of...

21 February 2019 | Europe