My son Arthur has just been awarded a prize for story-telling at his primary school. So when I watched that short movie whose script was generated by artificial intelligence, based on thousands of sci-fi books and films, I could certainly see a lot of similarity between both outputs. For me this epitomizes the current state of AI… It is raw, forming, full of potential but still with a long way to go towards maturity. Today we will be talking about the heart in artificial intelligence. After all, if artificial intelligence is, by definition, artificial, how can it have a heart, how can it have emotions?

Google created an AI that can easily remove the digital watermarks photographers put on their images to prevent unauthorized use. In a paper published online Thursday, research scientists at the firm describe how a computer algorithm can get past this protection and remove watermarks automatically when working with collections instead of single images. This gives users unobstructed access to the clean images the watermarks are intended to protect, and Google said the purpose of the research was to disclose the issue and find solutions. Google created an AI that can easily and automatically remove the types of digital watermarks photographers put on their images to prevent unauthorized use working with collections instead of single images. While removing a watermark from a single image is extremely challenging, the researchers found a'loophole' that makes it simple when working with an entire collection.

And manually removing them requires Photoshop skills, time and being ok with the image not looking its best post-removal. But Google has found a way around watermarks -- work it recently presented at the Computer Vision and Pattern Recognition Conference. The trick is to take lots of images -- we're talking hundreds or thousands of photos -- with the same watermark and use software to detect repeating structures. With enough examples, the watermark becomes the signal and all of the photos become noise. The watermark pattern can then be removed in totality from the image without reducing the quality of the image itself.

Google's partnerships with media companies for Home add-ons goes beyond advertisements for Beauty and the Beast. For next month's 125th anniversary issue of Vogue, readers can ask Google Assistant for more information on a quintet of articles. Once they do, the journalists who wrote them will share bits of interviews with Megyn Kelly, Nicole Kidman, Jennifer Lawrence, Serena Williams and Oprah Winfrey that didn't make it to print.

In her final months, she finds solace in an artificially intelligent holographic recreation of her late husband Walter, called "Walter Prime." They talk every day, recounting special moments of their life together. But her memory isn't perfect, and Walter Prime can only rely on retellings to piece together what happened. He also talks to other people, including Marjorie's daughter Tess and son-in-law Jon, who move in to take care of her. From all his exchanges, Walter Prime gathers various information on how to play his part, and pieces together a history shared between all the characters that he refers to in his conversations.

Hate crimes have sadly existed long before last weekend's tragic events in Charlottesville, Virginia but tracking them has been difficult. To help fix that, the Google News Lab has partnered with ProPublica, the New York Times, BuzzFeed News, the Southern Poverty Law Center and the University of Miami's School of Communications on the Documenting Hate News Index. Machine learning is used to pull locations, names and events from some 3.000 news stories published since this February into an easy-to-navigate feed of articles. "The feed is generated from news articles that cover events suggestive of hate crime, bias or abuse -- such as anti-semitic graffiti or local court reports about incidents," Google writes. "We are monitoring it to look our for errant stories that slip in, i.e. searches for phrases that just include the word'hate' -- it hasn't happened yet, but we will be paying close attention."

Do I have a bug, is it a matter of tweaking hyperparameters, or is the algorithm simply not that suited for these problems? The best performing submissions for cartpole you see on openai gym are discrete q learning with a q table represented as a lookup table. Because cartpole is simple you can discretize the 1x4 real valued state vector into a 1x4 integer vector with 8 or so bins per channel and then store a 84 element q table explicitly in memory. A neural net, even a small one is overkill as an approximation of this fixed sized q table so naturally the former performs better. Hyper parameters are a bitch to tune, and q learning isn't remotely stable 99% of the time until you balance them well enough so it can easily seem like nothing is working.

If you're a Google Home user without a paid Spotify account, your use of the service on Home has been limited. Up until now, Spotify has only integrated with Home for paid accounts. But at I/O, Google announced that would change; free Spotify users would be able to stream their library to Home. And now, it appears that the integration is now live for US users: free-tier Spotify users can now stream their music to Google Home. We've reached out to Google to confirm whether the service is available now or in the process of rolling out.

When we launched the Alternate Realities grant program in May we had no idea what to expect. We saw a need for funding in the arts happening at just the time when new media like AR and VR were starting to go mainstream. So, with support from our parent company, Oath, we set out to fund five immersive art projects that push the limits of storytelling through emerging technologies. Proposals came from as far away as Iran and Australia and ranged in discipline from theater to fashion, documentary to animation. There were multi-million dollar VR productions, animated shorts and escape rooms.

They rely on machine learning that purports to detect unique authorship with 96 percent accuracy.