1. Google super AI tools, was used to do a face piece A;
Technology has always come from human nature, so with the help of the AI tools, there are now a group of people who take Emma Watson, Emilia Clarke, Natalie Bowman and other actresses' The actress's head, to produce a variety of weird video, in the Reddit forum caused a face-lift craze, feeding the crazy netizen fantasy. How did this happen?
In 2015, Google announced that it would release TensorFlow, an internal tool for developing AI algorithms, which will change the direction of AI research and development around the world. In Google's CEO's words, the influence of AI technology may be as profound as electricity, and Google provides The tools are open, free, and the entry-level AI threshold is also reduced from a doctorate to just one laptop.
But the power of TensorFlow is now beyond Google's control. For more than two years, academia and Silicon Valley have used TensorFlow to create a number of compelling results, but the anonymous Reddit forum user deepfakes is the latest focus.
Deepfakes builds on the AI software to almost seamlessly sew any facial image into a movie, and you can easily imagine what it would be doing. "Motherboard reports that the software was used to put any A person's face (such as a female celebrity or a face friend) puts on a porn actress's body.
After "Motherboard" first disclosed the news, Deepfakes created its own subreddit and accumulated more than 91,000 subscribers.Another Reddit user, deepfakeapp, also released a tool called FakeApp that anyone can download and use for their own AI software. Currently Reddit has banned the community on the grounds that it violates the website for non-voluntary sex norms.
Based on FakeApp's user guide, the software was built on TensorFlow, a Google employee who used TensorFlow to do something similar with a slightly different set-up and theme, and a training algorithm to create an image from graffiti, Deepfakes said Many fun uses, such as putting Nicolas Cage's face in a variety of different movies, but we know that the 91,000 players who subscribed to the forum are for A movies.
TensorFlow This open-source resource brings many useful human results, such as the possible detection of cancer algorithms, the presence of FakeApp represents the dark side of the open source.
Google (and Microsoft, Amazon, Facebook) are a lot of technology, open to the world, now anyone can download AI software, experiment, create, which means you can fake political speeches (using a sound like AI ), Can also produce false retaliatory sex videos.All digital media are a series of 1 and 0, AI can be cleverly arranged to produce something that did not exist.
Because the software works directly on a single computer, big tech companies lose control as the software leaves the server, and the dogma of open-source ideas also means that these companies need not feel guilty or blamed for the consequences.
A storm for face is not enough to change the status quo, because free software is a good business for these companies, so that more people to develop AI. Every large technology companies are plunged into the fight for AI talent as much as possible, the more The better the influx of people into the field, the free labs write code, inspire new products, help pinpoint bugs, fix programs, and universities train college and graduate students to use the software to provide a good pipeline to increase familiarity with in-house tools New talent
Mazin Gilbert, vice president of advanced technology at AT & T, said: "Everyone has been talking about a big breakthrough in machine learning in the past five years. But the real big breakthrough is not an algorithm. Actually, the algorithm is basically the same as the 1970s and 1980s. It is open source. "
"Open source has reduced the barriers to entry, so it's not IBM, but Google and Facebook at the moment."
As public opinion attaches importance to AI's development ethics, the issue of open source software is also complicated by the fact that Google's current tools are not intended to create the terrorist threat of the Devil Terminator's Skynet, but can still cause real harm. Businesses that offer open source AI frameworks, such as Google and Microsoft, have been shouting the importance of AI ethics and letting scientists work on them, but these companies do not offer any guidelines for downloading free software. There are no ethically relevant guides like the TensorFlow website Description.
Let AI no longer open source does not seem to be the ideal solution, ArXiv, Github and other sites to share information for free, on behalf of stakeholders can identify potential pitfalls and demands accountability. As the core technology developers will not be willing to face For example, Gfycat removes all GIFs from deepfakes, Reddit forbids this forum, and PornHub also says that if similar videos are found, it will be deleted, but deepfakes.club The site still exists outside of major social platforms.
A face-changing film not only harmed the face-changing actress, but also belittled the professional work of actress A, treating them as "replaceable", disseminating them without their knowledge, disagreement, or control Manipulation; This technology can be applied to celebrities, of course, some people will inevitably be used in the general population, such as classmates, colleagues, children, classmates, ex-girlfriend, as a means of retaliatory sex, sexual harassment will be Into the golden age, as long as any woman was shot too many photos have suspicion, not to mention the photos uploaded to the community website more time thinking about permissions issues.
Cathy O'Neill, a mathematician who has always called accountability for big data, is not surprised by the fact that she said it has existed for years, and she suggests that in the era of fake news, Anything you see, what you hear, what you read, what you hear is skeptical, because even the sound is editable.
2. Also AI + IoT! Google tried to acquire Internet of Things company Xively;
AI + IoT, recent marriage continuously.
Today, Google has announced a high-profile announcement that it plans to sell 50 million US dollars from the software company LogMeIn hands to buy the Internet of Things platform Xively.
When it comes to the Xively platform, a lot of people may feel unfamiliar, but maybe you hear about Pachube, right, the same platform, a different name.
As the first independently exposed platform for sensor data, Xively opens up web APIs for developers to connect their devices to Xively's web platform and then release real-time data on applications, storage and presentation devices.
Google said it hopes the deal will be a springboard to tap into the growing IoT market and give Google Cloud a well-established IoT platform to increase its portfolio of products.
Google believes that 20 billion smart objects will be connected to the Internet of Things market by 2020. Xively will paint Google Cloud as a tool that allows device designers to build connectivity directly and provide cloud mobility between end-user applications and connected devices connection.
"This acquisition will provide a fully managed IoT service that allows easy, secure connection, management and processing of data from globally dispersed devices," said Antony Passemard, Google Cloud Platform Product Manager, in a blog post.
Not only Google is accustomed to the trend of AI + IoT, Baidu also frequently moves recently.
In February of this year, Baidu invested 394 million yuan in financing of D & D, a smart home company. Behind this financing, there is Baidu's new plan for the development of DuerOS system: DuerOS will enable IoT industry to make smart devices more Enter more families soon.
At present, BroadLink products such as smart sockets, patch panels, smart remote controls, switches, curtains, sensors and other products have been docked with the DuerOS system, allowing users to control these devices with voice commands and realize 'smart home'.
What do you think about AI + IoT? 36 Krypton
3. Who can become a leader in artificial intelligence? China is determined to make the United States silent;
US media reported on February 18 that the United States said that China released a plan in July 2017 in the hope of becoming a global leader in artificial intelligence and will create a value of 150 billion U.S. dollars by the year of 2030 RMB) industry.
According to the New York Times website on February 13, this 28-page English version of the English-language technical expert on artificial intelligence is the most important technical study that the United States may be in the coming decades A direct challenge to leadership in the field.It outlines a Chinese government proactive and spared no effort to regard the development of artificial intelligence as the Apollo 11 mission to China - a program that will also stimulate national pride and, Trigger a technological breakthrough that determines the overall situation.
Elsa Cania, an associate research associate who is involved in translating this declaration and focused on the development of artificial intelligence in China, said: 'It is noteworthy that artificial intelligence has become the top priority of China's leadership and that all The business has started quickly.
According to reports, when China embraced artificial intelligence, at a crucial moment of technological development, the leading position the United States has been occupying has begun to shrink.
For decades, artificial intelligence is not so much a science as fantasy, but over the past few years tremendous progress has prompted billionaires in Silicon Valley, Detroit and some of China's giants to invest in dialogue with people Everything from auto-driving cars to home appliances.
Reported that as ethics discuss how much autonomy should be given to weapons that can be thought independently, artificial intelligence has also become an important part of national defense policy.
American companies such as Amazon and Google do more than any other company in converting artificial intelligence concepts into real products.But due to factors such as the fear that the Trump administration will limit the number of engineers immigrating to the United States, many of the key AI The research is being transferred to other countries, including hot tech cities such as Beijing, Toronto and London.
Reported that for China's increasingly prosperous science and technology sector, to promote the realization of the next big event in the industry ('next big thing' is the mantra in Silicon Valley) is becoming an attractive possibility.
Reported that, with its huge market size and rapid experimentation, China will become one of the artificial intelligence powers, and even the most powerful artificial intelligence power.
Artificial intelligence has been the focus of Chinese technology experts for some time now.China published more research papers in 'deep learning' than in the United States, and deep learning allowed the machine to learn by analyzing large amounts of data, driving the rise of artificial intelligence One of the key technologies.
It is unclear what the overall Chinese spending in this area is now, but a provincial government in China promised to invest $ 5 billion in artificial intelligence and the Beijing government has pledged $ 2 billion to an artificial intelligence development park in the city. Nearly $ 1 billion has been allocated to this and Canada is home to many of the top researchers in the field, pledging to invest 125 million U.S. dollars, partly to attract new talent from other countries.
Reported that it is difficult to assess the current amount of investment in the United States government is exactly how much.USAIA, the National Institute of Standards and Technology and the United States National Science Foundation and other government agencies continue to fund new research universities and the private sector according to The Office of Technology Policy reports that the federal government spent about $ 1 billion on this area in 2015. The Trump administration said spending soared to $ 3 billion in 2017. But the administration said it could not match the 2015 statistics Make a viable comparison because it is not certain how the Obama administration calculates it.
'We may have some small initiatives that work well within the government, but we do not have a central national strategy,' 'Jack Clarke was a journalist and is now responsible for overseeing the creation of Elon Musk, the chief executive of Tesla. Artificial Intelligence Laboratory OpenAI policy work, he said. 'Confusingly, the technology we have so obvious power and value, we do not get full support, including financial support.'
According to a report of the American Association for the Advancement of Science, the Trump administration's budget for 2018 will cut 15% of the research funding for science and technology within government departments.
'They happen to be in the wrong direction,' said Thomas Khalil, a former Obama administration for technology and innovation at the White House Office of Technology Policy (OSTP). "Since China has placed it as a strategic priority, this is especially true Worrying.
4. Millet and South Korea Naver reached Internet of Things AI cooperation agreement;
Recently, millet and South Korean search giant Naver signed a cooperation agreement in the artificial intelligence business, Naver's artificial intelligence platform Clova will be applied to Xiaomi's IoT device.
Naver's Clova will work with Millet's own AI to improve their technology together. Naver announced last November that it has partnered with LG to include Clova in LG's Smart Speakers, which is also working with Qualcomm and last year acquired Xerox The European AI Center.
In the meantime, Xiaomi plans to launch 80 models of IoT products in South Korea this year, including smartphones and vacuum cleaners, etc. As China's smart phone market matures, domestic manufacturers have gone out to sea. At present, millet has surpassed Samsung to become India's No.1 mobile phone manufacturer Android China
5. Experts: AI is too high IQ must be merged with human beings to survive
Recently, some experts predict that with the evolution of artificial intelligence, its wisdom can be billions of times higher than that of human beings, and in order to prevent artificial intelligence from surpassing human beings beyond their control, human beings may need to connect the brain with them in order to survive.
Future scientist Ian Pearson, who works for Futurizon Consulting in Switzerland, commented that humankind needs to be integrated with machines in order to survive.1 AI may be able to go further than humans, We really need to take steps to keep up.
He said that the human brain can be linked to artificial intelligence to obtain the same IQ, which can resist the machine beyond. "In fact, I think the development of these artificial intelligence is not safe unless we can connect these devices directly with the human brain So that they will not go beyond us too much. '
In August last year, Tesla founder Elon Musk warned that artificial intelligence poses a great threat to humankind and Musk has been advocating government and private sector regulation of artificial intelligence technology. "No one likes to be Regulation, but anything threatening the public is regulated (such as cars, airplanes, food, medicine, etc.) as well as artificial intelligence. "He believes that control measures are necessary to prevent the machine from being controlled by humans . World Wide Web