close
BLOCKCHAIN

Media Technology Trends Report 2020: 13 major areas, 89 changes in full output

37f56c12-0b9e-11ea-a025-aaaa00151f43.jpg

Editor’s Note: This article is from the WeChat public account “Tencent Media Research Institute” (ID: TencentMRI), author FTI, 36 is authorized to publish.In the past few years, the media industry has witnessed too many technological changes.While enjoying the convenience brought by the development of technology, we are also silently bearing the consequences of technical abuse.Smart devices and voice assistants are becoming more and more meticulous. Synthetic media can provide personalized services for our tastes and preferences. We are exposed to a diversified, multi-interface smart ecosystem.However, in the past year, large-scale data leakage, fake news flooding, when the machine began to copy our voices and predict our behavior, where should humans go?As winter has arrived, the information age of technology companies has tightened, and the information age where opportunities and risks coexist may have quietly faced an inflection point.Future Today Institute (FTI) recently released the 2020 Entertainment, Media and Technology Trends Report, which introduces 16 frontier trends including artificial intelligence, synthetic media, blockchain, and 157 specific innovations, TencentThe Media Research Institute has selected a selection of reports for readers.As Neo said at the end of “The Matrix”: “I don’t know what the future looks like. I am not here to tell you how everything ends, but to tell you how everything is done.”Let us jointly touch the future of technology and media formats.Trend 1: Artificial intelligence AI is no longer a trend, but a third era in the computer age.This report summarizes the trends in entertainment, media, and technology from different aspects of AI.In the journalism industry, AI has become the darling of major news organizations.Reuters’ Lynx Insight program helps journalists mine big data.The Washington Post’s Heliograf was used to cover elections and sporting events.The CinemaBuster of the Duke University Journalist Lab can help journalists with fact checks.The Proliferation of Franken-algorithms algorithm is just a rule for defining and automatically processing data.They use the “if…then” logic that the computer can understand.For example, if the IP address of the website viewer is in Chicago, the algorithm allows them to enter directly; if the IP address is in London, the algorithm will first display the privacy and cookie policy as required by GDPR.Although one can design an algorithm directly according to his or her own ideas, all of the algorithm systems working together can cause problems.Developers don’t always know in advance how algorithms will work together.Sometimes, several developer teams deal with different algorithms and data sets independently, but only when the algorithm is designed can you see how it works.This is why the stock market and e-commerce websites have recently collapsed.For a large company like Facebook, this is a formidable challenge, because at any given time, billions of algorithms work at the same time, and it is impossible to predict the results.Proprietary, Open and Homegrown AI Languages ​​Python is an advanced programming language with many pre-built libraries and frameworks.MIT has developed an open source language called Julia that focuses on numerical computing, in addition to the Lisp language that AI’s author John McCarthy created in 1958.Major companies are starting to build and distribute their own software packages and unique programming languages ​​for AI applications.Uber wrote his own probabilistic programming language Pyro in Python.Unlike the opposition between OSX and Android or the early Mac and PC camps, this move means that the AI ​​ecosystem will be split in the future.The market will find that conversion between different AI frameworks and languages ​​is costly.Problematic Data Sets The AI ​​developed by the company’s self-training is not representative and therefore cannot be widely adopted.Researchers at MIT have released an AI called “Norman” to capture recognition photos.They did a set of comparative experiments, one group of trained standard data, the other group was untrained, and the results were astounding: the system using standard data showed “a black and white picture of a baseball glove”.The other system shows “a man who was shot abroad during the day.”Some new systems for generating natural language were released in 2019.Although these systems were previously trained, Reddit, which was used to learn natural language, and comments on Amazon were removed.The reason is that Reddit and Amazon commentators are biased towards whites and men, so the words of these people do not represent everyone.This shows that developers still face challenges.It has become difficult to get real data from real people to train the system, and because of the new privacy policy, developers can only rely more on public data sets and problematic data sets.Deep Linking of Norman AI and Standardized AI Photo Capture Contrast Data Since the advent of smartphones, deep mobile connectivity has enabled users to find and share data across all software in their phones.But now deep connections make it harder and harder for users to find the information they want.In 2019, the Yelp restaurant marked the exact contact information in its software, but when the customer clicked, they were jumped to the Grubhub software to order the meal.Even if the customer turns off the software and wants to make a direct order, the software still converts it to the number on Grubhub, because Grubhub can classify it as a “marketing” event and charge the restaurant a high commission.There are three types of deep links today: traditional deep links, delayed deep links, and contextualized deep links.Traditional deep links redirect you from a piece of software or a website: If you click on a Baltimore Sun link that someone posted on Twitter, then in theory, as long as the user installs Baltimore Sun software, it automatically opens Baltimore Sun.Delayed deep links are also linked directly to the software (if installed) or directly to the software store for users to download the software first.Contextualized deep-linking services are more powerful, allowing users to go directly from the site to the software, from the software to the site or from the software to the software, and can also provide personalized information, despite deliberately concealing the entire process from the consumer.AI in the Cloud In the past year, the leaders of the artificial intelligence ecosystem have been vying for “artificial intelligence cloud sharing” in order to become a trusted AI remote service provider.In the West, the field is led by Amazon, Microsoft and Google, followed by Apple, IBM, Salesforce, SAP and Oracle.In Asia, AI Cloud is dominated by giants such as Alibaba.This is a $250 billion industry and is still growing rapidly.Arun Sundararajan, a professor at New York University’s Stern School of Business, said: “The (competitive) benefits are the operating system of the next technological era.” Entertainment and media companies will find more ways to use AI clouds in the coming years.AI Chipsets For us, the performance of CPUs on laptops and mobile phones is constantly improving, but it can’t meet the requirements of machine learning.The problem with them is that there are not enough processing units to complete the connections and calculations needed for the next computer era.At this time, a new set of processors is needed. Enterprises such as Huawei, Apple, and IBM are testing the construction of new systems and SoCs.In short, this means that the chip can already play a role in the AI ​​project, with faster speeds and more accurate data – and it’s not hard to predict that several companies will compete in the future.Tesla’s new custom AI chip, although not as compelling as originally described, was released in April 2019.Google’s Tensor Processing Unit (or TPU) is built specifically for deep learning of AI and is designed to work with the company’s TensorFlow system.Figure 2 Google Tensor Processing Unit Ubiquitous Digital Assistants Digital assistants such as Siri, Alexa, and Tmall use semantic and natural language to process our data, sometimes even if we don’t know what to askForecast ahead of time what we want or need to do next.The FTI model predicts in 2017 that nearly half of Americans will own and use digital assistants by 2020, and the FTI model will continue to track trends in this direction.Amazon and Google dominate the smart voice market, but digital assistants are ubiquitous.There are now thousands of digital assistant software and applets that track responses.News agencies, entertainment companies, marketers, credit card companies, banks, local government agencies (police, road management), political campaigns, and many other activities are also communicating important information through digital assistants.Nvidia, a chip designer using Generating Virtual Environments From Short Video, is teaching AI to build realistic 3D environments with short video clips, leveraging previous research on generating GANs.Graphics generated from the open source dataset by the Nvidia system will be used in the automated driving arena.The designer trained the GAN using short clips that were divided into different categories (buildings, sky, vehicles, signs, trees, people) to generate new versions of these objects.The promising application of automatically generating virtual environments is logistics (warehouses, factories, transportation centers), urban planning simulations, and even testing of traffic scenarios in amusement parks and shopping centers.The actual video content and AI generated Machine Recognition (Machines Performing Cognitive Work) companies no longer rely solely on the AI ​​system to perform cumbersome repetitive tasks.More advanced systems are helping companies optimize their workflows and proactively generate policies.This means that people are not completely replaced by robots; instead, robots are created in accordance with human capabilities.From the warehouse to the auditing firm, the artificial intelligence system begins to perform cognitive tasks—in this process, humans only need to perform basic operations.Amazon’s automation system helps improve warehouse efficiency and guide employees through their work.Wal-Mart uses computer vision to find cooked agricultural products, and its AI system can inspect people who only remove bad apples from the heap.In a newsroom, a similar system can help journalists filter very large data sets to find anomalies or identify people.Robotic Process Automation Machine Process Automation (RPA) enables companies to automate tasks and processes in the office, allowing employees to spend more time on more valuable work.Google’s Duplex is a type of RPA that is used to make regular phone calls to others.Amazon uses RPA to screen resumes and then rank the best candidates.In banking, Blue Prism and Automation Anyware can help employees handle repetitive tasks and increase employee productivity.This technology will enable media and entertainment companies to save money in many different areas, such as customer service, to make better real-time forecasts.Robots (Bots) Basically, a robot is a software application designed to automate a specific task.In the media world, robots can be divided into two categories: news bots and productivity bots.The former can help gather news information and automatically push specific news events for readers; productivity robots can help news organizations automate their daily processes.The next major advancement in robotics is not in terms of technology, but regulation.In the 2018 campaign, we saw the recovery of “botnets”, which are computer networks that send misleading content.As people are worried about more and more robotic fraud, California has enacted a new law that came into force on July 1, 2019, requiring robots to clearly, conspicuously and reasonably demonstrate their interactions with humans.I am not a human being.Tencent Media + Summit Live Writer is writing Real-Time Machine Learning. Machine learning refers to a system that applies algorithms to analyze data so that it can better perform various tasks, and over time, itWill be better at these tasks.But this system also faces efficiency problems: the system needs to stop and parse the data.Recent research shows that real-time machine learning can adjust models in real time as data is acquired.This marks a dramatic shift in the way data is moved and how we retrieve it.For example, even in the case of mixed languages, the technology can automatically interpret simultaneously; it can also adjust the content distribution at any time to provide readers with more personalized content.Compared to the use of historical data in a rigid way (readers XX only like sports reports), real-time preferences can incorporate content into the recommendation mechanism (readers XX may have more demand for election news in the next few days).Natural Language Understanding (NLU) For conversational AI systems such as Siri and Alexa, it is more difficult for the machine to accurately understand someone’s meaning.These systems are trained to understand the pronouns in the sentence at most.If the consumer asks “What time does the Lion King release at the Cinemark Theatre? Then park near that”, the system automatically infers that “that” means “Cinemark Theatre.”Technically, this process is called “slot carryover.”It can use the French language to understand the meaning of pronouns unless we say complex sentences with many different pronouns.The truth is that in everyday communication, our speech is very confusing, misusing words, and even using modal particles to convey meaning.In 2019, Amazon researchers made remarkable progress in NLU, and they introduced a new architecture that would help Alexa understand humans well without humans’ complete sentences.Amazon Alexa Home Machine Reading Comprehension (MRC) MRC makes it possible for the system to read big data, infer meaning, and immediately get answers.For example, when you search, do you want the system to give a definitive answer directly, or provide you with a bunch of URL collections, “How do you want to see more hyperlinks?”Let the machine figure out the problem itself, this is the MRC.In the future, MRC is one of the key steps to achieve strong artificial intelligence, and in the near future, it can help us to convert various materials such as technical manuals, historical maps and medical records into a collection of easy-to-search information.Natural Language Generation (NLG) Natural Language Generation (NLG) is now used by many media and marketing organizations to produce automated content based on large-scale data sets.NLG’s achievable features include integrating keywords, improving SEO (Search Engine Optimization, which uses search engine rules to improve the search ranking of websites), and providing personalized content to users in bulk.Arria NLG, IBM Watson voice-to-text technology, Amazon Polly, Google Cloud Voice-to-text technology, narrative science company Narrative Science and Automated Insights Automated Insights use large data sets to build narratives to help non-data sciences better understand their organizationWhat is happening in the middle.NLG has countless use cases in various areas of expertise to assist lawyers, politicians, doctors, consultants, financial analysts, marketers and others.Real-Time Context in Machine Learning The Project Debater developed by IBM can find out logical loopholes and fake messages from the context by digesting a large amount of text.Although it is currently in the testing phase, it has been able to use the real environment to distinguish true and false information through real-time learning.How Project Debater Works General Reinforcement Learning Algorithms AlphaZero’s team develops new algorithms to learn multiple tasks.For example, AlphaZero is not only unbeatable in Go, but also excels in chess and Japanese chess.Much Faster Deep Learning Deep Learning (DL) is a relatively new branch of machine learning and will soon be invisibly integrated into organizations.Designers use more specialized deep learning algorithms in conjunction with various databases including text, images, videos, presentations, and the like.Conceptually, it is not new, and the most recent update is the amount of computing power and the amount of data available.Putting it into practice means that more human affairs can be done automatically by the computer, such as designing software to write code.DL is constrained by the speed of computer networks: A few years ago, using the datasets in the ImageNet website to train image recognition could take a month or more; now, Facebook can achieve the same effect in an hour.As computer speed and hardware technology improve, the system will complete the task at a more “superman” speed.ImageNet official website Reinforcement Learning and Hierarchical RL Reinforcement Learning (RL) is a powerful tool for dealing with decision-making problems. It is applied to AI system training to make it more capable than ordinary people.In the computer simulation process, a system tries, fails, learns, experiments, and then tries again—this series of steps can be completed quickly, and each trial and error will correct its future attempts.AlphaGo, which we are familiar with, is based on the RL mechanism to learn how to decide to defeat human chess players.But this technology also has problems: when agents lack sufficient supervision (supervisor, simply supervising is setting output values/targets, it is easier to find hidden patterns in the data, and unsupervised learning is not set.The output value, which is the same concept in the “machine supervision” related to machine learning related technology, or the need to run a long-term complex task, may encounter difficulties.At this time, researchers will try to apply Hierarchical Reinforcement Learning – to be able to find high-level actions, to overcome learning difficulties in an orderly manner, and to finally master new tasks at a speed that is unexpected to humans.RL can enhance the “intelligence” of AI systems to enable cars to drive automatically under unconventional conditions, or to assist military drones to implement complex actions that have not been previously implemented.Continuous Learning Now, deep learning technology can help the system learn to complete complex tasks in a way that is closer to what people do, but these tasks are still very specific, such as defeating humans in a certain game.And they need to follow a strict program: collecting data, setting goals, and applying an algorithm.This process requires manual participation and can take a lot of time, especially in the early stages of supervised training.Continuous learning (CL) will focus on building skills that promote autonomous learning and incremental learning, and researchers will continue to expand their capabilities boundaries in the future.In the multitask learning film “The Karate Kid,” Gardener Miyaji promised to teach the boy Daniel Karate, but Daniel quickly got tired of day-to-day training.For Daniel, paint fences, cars, endless “waxing, sealing wax”… These things seem useless, and certainly can’t help him learn karate.Of course, all the last chores were proven to be related to karate, and such training helped him become a karate champion.The film was mentioned because researchers have recently trained intelligent systems like Daniel.When developers use machine learning, they try to solve a single specific problem in this way.They will oversee the fine tuning of the smart system and continually fix it until the system’s performance is as expected.But focusing on a single task often leads to invalid results—maybe there is a better solution than the one discovered by the researchers?As a result, a new field of research, multitasking, has emerged, allowing systems like Daniel to seek connections in a variety of related tasks to find out how to better solve problems.Generative Adversarial Networks (GANs) face-changing technology has been hot in 2019.Face-changing techniques based on generative confrontation networks (or GANs) are easy to implement.We can understand GAN as a Turing test without any human involvement.GAN is an unsupervised deep learning system consisting of two mutually antagonistic neural networks trained on the same data (eg, human images).For example, the first AI creates a photo of a woman who looks real, and the second AI compares the generated photo with the photo of the real woman.The first AI re-adjusts its generation process again and again based on the judgment of the second AI until the image of the woman who looks completely true is automatically generated.Thispersondoesnotexist.com website is using this technology to continuously generate realistic photos, but in fact these photos do not exist.GANs are also used to dynamize old photographs or paintings.This year, researchers at the Skolkovo Institute of Technology and the Samsung AI Center used the technology to make Mona Lisa shake her head and let Rasputin sing Beyonce’s “Halo.”Automated Machine Learning (AutoML) generated by ThisPersonDoesNotExist.com is a new machine learning method that matches raw data and models to display the most relevant information.Some institutions are freed from the traditional machine learning methods that are time consuming and difficult.Now, Google, Amazon and Microsoft offer many AutoML products and services.Customized Machine Learning Google’s Cloud AutoML helps users upload their own data construction models, even non-professionals can train machine learning.AI’s persistent bias (Ongoing Bias In AI) AI has serious prejudice is no secret.This issue is multifaceted.For example, the data set used to train AI is usually from Reddit or Amazon’s comments and places where Wikipedia itself is prejudiced.People who build models often don’t know they are biased.As our computer systems are used more and more for decision making, we may find ourselves being grouped into a group by an algorithm. Although it may not affect us, it may actually create a huge hidden danger.AI Bias Causes Civil Unrest In fact, every day you actively or passively create countless data (such as uploading and tagging photos on Facebook, driving to work, etc.).These data are often mined and used by algorithms that you don’t find and used to create ads, help potential advertisers predict our behavior, determine our mortgage rates, and even help law enforcement to predict if we might commit a crime.Researchers at universities such as the University of Maryland, Columbia University, Carnegie Mellon University, Massachusetts Institute of Technology, Princeton University, the University of California at Berkeley, and the International Institute of Computer Science are studying the side effects of automated decision making.You or someone you know may fall into the wrong side of the algorithm. For example, you may not qualify for the loan due to some opaque or difficult to understand reasons, you may not get a specific drug or you may not know the rent price.And more and more data is collected and sold to third parties without their knowledge.Trend 2: Computational News (Intelligent Journalism) News reports can use calculations to mine stories that would otherwise not be discovered.CS technology can help the journalism industry in two main ways: journalism using computational methods and journalism about computing.In July 2019, the Washington Post established a team specializing in general elections. The team established a computational political journalism research and development laboratory and conducted experiments to support the data work of the Post before the 2020 election.The Computing News Lab at Stanford University has been developing new computing methods for public journalism.The California Civic Data Coalition is a joint venture between several media to track political funds. Computational Image Completion and Generation Now, every person with a smartphone can use computational photography tools..They can open the closed glasses in the group, find the best frames in the motion picture and clear the flaws in our selfies.All of this is real-time without the need to launch other photo editing software.Now we can seamlessly add or remove objects, change shadows, and more in the scene.Obviously, there are ethical requirements for journalists here – under what circumstances does the editing of the photos allow?Similarly, before using it for a story or story, a journalist should develop a technique that automatically shows that the photo has been edited.Automated Versioning A Swiss-based Tamedia reporter tried to use generation technology during his 2018 election.Tamedia automatically generated an article using a decision tree algorithm called Tobi, detailing the results of each city covered by 30 newspapers organized by private media, and simultaneously generating a total of 39,996 different languages.The version of the election report, each with an average of 250 words, was posted to Tamedia’s online platform.Each report marks that the report was written by an algorithm.With more attempts, news and entertainment media companies can develop multiple versions of the same content to reach a wider audience or mass-produce content.Tobi’s Twitter account generates natural language to adjust the level of reading (Natural Language Generation to Modulate Reading Level) Natural Language Generation (NLG) is a processing task that can generate human-specified language.NLG can be used to rewrite content at various reading levels, providing great possibilities for both book publishers and news media organizations.As book publishers and news organizations look for new sources of revenue, NLG will not only be used to write reports, but will also create different versions for readers with different levels of reading.This is because the basic corpus (the data that makes up the story) does not change, but the vocabulary and details can be adjusted.For example, reports on Berkshire Hathaway’s quarterly earnings can be expressed in many different ways, for financial professionals, high school economics courses, English as a second language learners, andNon-English speaking countries.Similarly, NLG can be used to automatically create high-volume tasks such as summaries and abstracts.Using NLG to customize and write different versions of the story can help media organizations scale their global business without hiring other people.However, NLG can also be used for fraud, which means that it will be regulated in the future.Data Mining Crowds Computational news technology enables journalists to query our passive data (such as our online activities, health records, locations, etc.) to learn or understand new things.Our data not only tracks our behavior, but anyone can use it to search, collect and analyze.We expect more news organizations, as well as marketers, political activists and other groups, to start using data creatively.Because our thinking influences action (such as searching for “What is the EU?”), our actions generate data that can be used to learn about us.Algorithmic Fact Checking Misleading false information pollutes the Internet and our social media environment, and everyday consumers have to fight against false information.Although we have done a lot of fact-finding work on a global scale, it turns out that the algorithms used to spread fake news are faster than human fact checkers.Researchers at the University of Texas at Arlington and Google have been studying automation techniques that use framework semantics.A framework is a diagram that describes a particular type of event, situation, object or relationship, and its participants.Researchers have extended a system called FrameNet to build a new framework specifically for features including automatic fact checking.On-Screen Fact Checking Researchers at Duke University and the University of Texas at Arlington invented ClaimBuster, which scores facts in any sentence.The system uses audio or video from live events and converts it to text, uses filters to identify statements about the facts, and then matches those statements to the database.ClaimBuster’s Synthetic Data researchers do not always have access to complete health, medical, transportation, and demographic data.Therefore, some people are developing and experimenting with synthetic data sets to perform meaningful analysis and training models in AI.However, synthetic data sets often miss important information or appear to be biased.Researchers at Data to AI Lab at the MIT Information and Decision Systems Laboratory are developing a machine learning system to automatically create synthetic data that can then be used to develop and test data science algorithms and models.They proposed the synthetic database Synthetic Data Vault (SDV), which is capable of learning and developing multivariate models for a variety of purposes.Journalists need to know when to use synthetic data, and news organizations should develop ethical guidelines on when to use synthetic data and how to inform consumers about the use of synthetic data.Synthetic and Generated Content Authentication AI can be used to discover which text is written by the algorithm rather than written by humans.Researchers at Harvard University and MIT-IBM Watson AI Lab have developed a tool for identifying when an algorithm generates text.The Giant Language model Test Room (GLTR) uses AI to determine common words in text, and can determine whether a sentence looks too predictive to be written like a real person.The tool is expected to be used to identify false or misleading news, robot-generated content, and counterfeits.Giant Language model Test Room Text Recognition (Motion) Trend 3: Cognitive System Voiceprints Voiceprints are unique features of a person’s voice.Now the new machine learning technology is connected to the huge data set of voice recording to help researchers identify the voiceprints that people produce when they speak.Researchers at Carnegie Mellon University have developed a technique for generating 3D faces using voiceprints.Law enforcement agencies are applying the system to identify prank callers and deceive local dispatched special police in order to retaliate against others.Personality Recognition’s emerging predictive analytics tool records users’ data, behaviors, and preferences that reflect the user’s personality and predict the user’s reaction in any situation.In 2018, Cambridge Analytica used algorithmic analysis to help Donald Trump win the election.Political candidates, law firms, marketers, customer service representatives, and others are using new systems that view user mobile web activity, emails, and conversations to assess user profiles in real time.The ultimate goal is to determine the specific needs of the user.ElectronicArts is developing a system that can evaluate the character of multiplayer video gamers to better match players based on their game style, dialogue style and spending willingness.In the real world, insurance underwriters are trying to assess a user’s personality through magazines and websites subscribed to by users, photos posted on social media, etc., in order to determine their insurance risk.Some lenders have begun to use personality algorithms to predict future business transactions for users.(The data shows that if two people with the same professional and personal circumstances borrow at the same time, then people with higher college degrees are more likely to pay off their debts.) Emotional Recognition In 2018, Amazon applied for a new system.Patent, the system can detect the user’s physical and mental health based on the user’s past and present interactions.If Amazon detects that the user is ill, the user is advised to take the cough medicine within one hour.Automaker Kia debuted its real-time emotion recognition system (READ) at CES in 2019.The system can monitor the passenger’s facial expression, heart rate and skin electrical activity through sensors to adjust the interior environment to the passenger’s emotional state.Kia’s READ System Affective Computing emotional computing is an interdisciplinary field that covers computer science, psychology, and cognitive science.Researchers at the Massachusetts Institute of Technology are working on a machine learning algorithm that understands and responds to our emotions through skin electrical activity collected from our wearable devices (smart watches, fitness trackers).But data from other sources may also come in handy, such as our skin, face, and conversations with others.Analysts speculate that emotional computing will become a $25 billion industry by 2023.Trend 4: Synthetic media synthesis media has been everywhere, such as the virtual Japanese pop star Hatsune Miku (debuted in 2007), the British virtual band Gorrilaz created by artist Jamie Hewlett and musician Damon Albarn (issued in 1998)The first single).They are all media created or modified by algorithms.Hatsune Miku’s speech synthesis Speech Synthesis is also known as “synthesized speech” or “text-to-speech technology”, which mimics real human speech and applies it to various interfaces.By leveraging sufficient data and training, the speech synthesis system can understand the spectral frequency of any person and produce a digital voiceprint for a person.Synthesia is a company that uses this technology to dub through automatic re-shaping facial animations for internationally widely released movies.The actor’s facial expressions and mouth can be reworked to match the dubbing.The Modify Custom Voice generation algorithm creates synthetic sounds that sound like raw sounds and can be modulated to the desired pitch and pitch.Lynebird, a Montreal-based AI company, built a speech-imitation algorithm that generates synthetic speech that is hard to detect.The voice sample database it uses can be found either in public repositories (YouTube, Vimeo, Soundcloud) or by users.Over time, artificial intelligence not only learns to recognize intonation, but also recognizes emotional rhythms.As long as there are enough publicly available audio files to build the dataset, you can fake the conversation between yourself and your favorite star.It quickly matches and quickly generates personalized synthesized speech for each consumer.Lyrebird video editing voice fraud (Voice Fraud) synthetic media has always had a problem: unscrupulous people can use it to mislead people, deceive voice authentication systems and fake audio records.According to a study by call center software maker Pindrop, voice fraud alone caused $14 billion in damage to US companies with call centers last year.As part of the ASVspoof 2019 Challenge, Google has been working on synthesizing speech data sets, an open source code designed to help develop countermeasures against deceptive speech.Machine Image Completion If a computer system can access enough images, it can fill holes in the photo.This is a very practical application for professional photographers and everyone who wants to take a better selfie.If the foreground of the mountain is not clear or the skin is flawed, you can switch to another version to produce a perfect image.But how do we draw a line between reality and the improved scene?To what extent should the photo be modified without the label or disclosure?Online dating, journalists, marketers, and policy makers should think about these issues.Image generation is also a useful tool for law enforcement and military intelligence personnel – computers can now help them identify people or objects in video frames.Taking into account machine learning algorithms and biases on data sets, image generation may become part of the discussion about privacy and devices.AI researchers at the Skolkovo Institute of Science and Technology and the Samsung AI Center used machine image generation technology to move old photos and famous paintings (such as the Mona Lisa).The Deep Behaviors and Predictive Machine Vision researchers at the Massachusetts Institute of Technology CSAIL are not only training computers to recognize the content of the video, but also allowing the computer to predict what humans will do next.Computers can now watch videos and TV shows on YouTube predicting whether two people might hug, kiss, shake or clap.This research will make it easier for robots to navigate and interact with humans by learning our body language.It can also be used to sell, operate machines or learn.Researchers at Nvidia, a manufacturer of voice, sound and video algorithms, developed a new generation algorithm in 2018 that created realistic faces using a generative confrontation network..The algorithm can also adjust various elements in its system, such as age and freckle density.A team at the University of California, Berkeley, developed a software that automatically transmits the actions of a person in one video to someone in another video.We have been training computers to watch videos and judge the corresponding sounds in the real world, such as what a raft will make when it hits the sofa.The focus of this research is to help the system understand how objects interact in the real world.But the technology is also deceptive: In 2017, researchers at the University of Washington developed a model in which Obama published a speech that he had never published in real life, but it was enough to confuse.The phenomenon of composite video mixed with AI designed to provide real content will make the problem even more difficult.Trend 5: Optimizing For Voice Search Dialogue networks are everywhere.With voice control we can instantly get information and other content from smart speakers, car dashboards, TV remotes and digital assistants for smartphones.With the popularity of voice search, organizations such as publishers have new strategic thinking: Can content be optimized for voice search?Audioburstis’ technology can extract and analyze audio from sources such as broadcast, contextualize and search, and help AI powered cars provide listening services.The next generation of Next-Gen Native Video and Audio Story Formats is developing non-new narratives to attract viewers through the development of consumer technologies with responsive visual, tactile and audio interfaces.News media and entertainment organizations have begun to explore these novel, personalized, interactive and immersive narrative models.While this will undoubtedly provide a better experience for consumers, news and entertainment media companies must plan ahead so that they can be aggregated and delivered in a narration in the future.The “Black Mirror: Panda Sneckey” story uses the “interactive video” concept. Human-Machine Interfaces. Voice interfaces such as Siri, Alexa and Google Assistant are becoming more and more complex.But researchers have begun to look to the future: connecting humans and mammals directly to computers.People can communicate with ideas through human-computer interaction interfaces, providing new options for stroke and paralysis patients.Trend 6: Augmented reality (mixed reality) mixed reality, also known as extended reality, can digitally generate, enhance, or manipulate environments, including virtual reality and augmented reality.MR is typically experienced through a head mounted display (HMD) or mobile device.In the past decade, MR has attracted more and more attention.By 2020, we will see MR continue to grow in various industries and new markets.Many people hope that in the next decade, with the popularity of 5G as an MR catalyst, this technology will become part of our daily lives.Holograms Holographic images are projected images that appear to be in three dimensions.By 2024, the holographic image market is expected to exceed $5 billion, with advertising being relatively heavy.In consumer products, professional camera brand Red announced the launch of its Hologram 2 mobile phone, and Huawei and Samsung are also developing mobile devices with holographic images, including holographic communications.In the entertainment world, companies like Los Angeles’ BASE Hologram continue to launch “live” concerts featuring holographic projections of deceased artists. Whitney Houston will tour the 2020 show, but they are not yet in holographic images.Multi-angle rendering of the volume projection necessary for 3D holograms, so ticket sales are limited.In order to avoid controversy over animal rights issues, the German circus Roncalli has launched a holographic spectacles that can show animal performances, which are both cost-effective and not cruel.BASE Hologram restores the singer Buddy Holly’s 360-degree video 360-degree video with a dedicated camera, designed to capture omnidirectional material.After the video is rendered, the viewer can use the mouse, touch screen, or motion control gestures to rotate the view and view the recorded scene.YouTube, Facebook and Vimeo all offer 360-degree video, and major online media such as ABC, Fox and CNN also have dedicated digital channels for playing immersive content covering news, sports and entertainment, and more and moreThe media is close behind.GoPro, Insta360, Ricoh and Xiaomi’s portable 360-degree cameras will soon increase the number of 360-degree videos that users shoot and socially share.The dinosaur 360-degree panoramic video Augmented Reality (AR) on YouTube does not simulate the entire new environment like VR, but just places a few digital elements in your natural view.AR can use a head-mounted display or smart glasses experience, leading brands such as Google and Microsoft, as well as Magic Leapand, Vuzix and Meta products are under development or already on the market.Mobile devices provide the most convenient AR experience for the average consumer as companies compete to develop AR headsets.Some cinemas and NBA venues have introduced AR games that can be played on mobile devices. Quartz’s mobile app integrates AR functionality into some news reports, while Google uses AR to perform real-time translations of foreign languages.AR (AR as a Tool to Enhance Print) AR will effectively help the development of the printing industry, and media companies are using AR prints to further attract consumers.AR can integrate two different channels of content and advertising.Services such as blippar can add animations, models, or images that can only be unlocked using a smartphone.Companies like Max Factor and Net-A-Porter have allowed users to scan their favorite products with AR and buy them directly from their mobile phones.The success of AR-driven advertising campaigns means a win-win situation for media companies and advertisers, and brands will continue to use AR print ads due to lower prices.Virtual Reality (VR) Virtual reality is a computer simulation environment.Get an immersive experience with VR glasses or by embedding your phone in a specific device.In recent years, VR has made important developments in the field of entertainment content. New Emmy awards such as interactive media have emerged in an endless stream. Hollywood directors Darren Aronofsky and Robert Rodriguez are also shooting with VR devices.Google, Sony, Samsung and HTC are all selling VR headsets.In 2019, Facebook’s Oculus launched the Quest headset, a separate replacement for the Oculus Rift device, which can only be used when bundled with a PC.Nintendo has also recently entered the immersive market, launching the VR suite for its popular Switch gaming device.You can also construct a headset by sliding it into a special mask.The “standing” VR is viewed from a relatively fixed angle, unlike the “room size” VR, which allows viewers to walk more freely in physical space, and its digital environment reflects their real life.Oculus Quest Device Trend 7: Streamers Streamers in the fourth quarter of 2017 witnessed 500,000 consumers abandoning their cable and satellite TV services.The two main reasons why users continue to pay for TV are that they can watch live broadcasts and discounted Internet and cable service packages.- Obviously these two reasons are not as big as the traditional cable TV can continue to advantage.Netflix and Amazon are the two major streaming media in the world.By 2020, everyone’s eyes will be on the new Disney+ service with Marvel and Star Wars, not to mention Hulu and ESPN+.In the future, we expect to see an increase in the share of more streaming media devices such as Amazon Fire Stick, Google Chromecast and Roku, and a steady decline in cable satellite TV subscriptions.Streaming media services will erode the local broadcast news market, and will also put long-term TV news programs at a disadvantage.Slash of OTT Streaming Services may be due to the success of HBO Now, and in 2019 all major online media are launching its top streaming services.Disney+ is launching its own OTT service and expects AT&T/DirecTV, Viacom and Discovery to offer new services or update existing services.The OTT streaming service market is crowded enough to be more saturated in the near future.Connected TVs have been available since May 2019. All Samsung TVs have integrated screen image and content projection for Apple devices and access to native Apple TV and iTunes apps.It actually bypasses a separate Apple TV device, but most other TVs must purchase Apple content from this device.This makes Samsung attract most Apple users.Networked TVs are expected to be more popular in the average home, which may be associated with popular streaming services or aggregation devices such as Amazon Prime Video, Roku, Hulu, YouTube, Showtime Anytime, iPlayer (UK only), All 4 (onlyLimited to UK), Playstation Now, HBO Now, DirecTV Now, iTunes and Netflix.Media organizations can use connected TV and exclusive integration as a competitive strategy to provide richer content to retain and expand their audience.Samsung Connected TV Trend 8: e-Sports e-sports is a fast-developing and confrontational digital game industry that can professionally produce games and target audiences through live streaming or face-to-face.Although this organized and confrontational game has been developed for decades, in recent years, the development of game technology and streaming media has led to a huge increase in its popularity and legitimacy.For example, Fortnite (fortress night) is one of the hottest games of last year, with a revenue of 2.4 billion in 2018 and more than 250 million users.Mixed Reality Arcades is like the video games of the 1980s. The form of games is becoming more and more popular, but most people can’t afford the equipment. So, like the game hall of the year, the mixed reality experience hall came into being..Nomadic, the founder of the MR arcade, has opened an offline MR gaming experience store in Orlando, Florida and San Rafael, California, where players can explore the room with VR helmets and backpacks.The Mixed Reality Experience Hall is everywhere, so everyone can participate in many games, but this time it doesn’t have to wait a long time.Virtual World Arcad offers a membership package with unlimited virtual reality time.In Tokyo, the Mixed Reality Experience offers more than just basic games, including swinging seat belts, flight platforms and simulated bungee jumping, flying, and even falling from skyscrapers.Virtual World Arcad Mixed Reality Experience Center Trend 9: Platforms forced to pick the platform. Due to the proliferation of fake news and hate speech, online platforms and social media will increasingly invest in platform management.: You can use manual review, or you can use an algorithm that detects hate or problem speech.Amnesty International launched a project called “Troll Patrol” on Twitter, which found that nearly 30 female politicians and journalists received tweets that were “insulting” every 30 seconds.In addition, the policy is clear and crucial.However, the real question is which stakeholders will be selected to define these rules, as any decision may be politicized.Restrictions on Bulk Messaging news organizations currently rely on third-party platforms to engage consumers.Limiting bulk messages will make it more difficult for news media to interact with the audience.Since December 2019, WhatsApp will no longer allow automatic or bulk push messages.In today’s era when media platforms are more popular than email, Facebook’s WhatsApp has been an important tool for news organizations to send news every day.The company announced that it will take legal action against any individuals or companies that send messages in bulk on the platform after the new restrictions are effective on December 7.These restrictions are largely related to future platform restrictions.WhatsApp has more than 400 million users in India, and false information and rumors have triggered a series of mob turmoil.According to data from the Pew Research Center at New Interest in America’s Local News Outlets, the number of employees in US newspaper newsrooms fell by 45% between 2008 and 2017, setting the number of US local newspaper mergers and closures.Record.News organizations are currently making a lot of efforts to rebuild the local journalism industry.In 2019, Google’s “News Program” worked with the Associated Press to build a tool for the newsroom to share content and its reporting plans.The company is also working with McClatchy to create a digital news site that covers the local area – the first site will be located in Youngstown, Ohio, where the daily newspaper “Vindicator” has just closed.In 2018, Facebook and the Lenfest School of Journalism funded the launch of a local news subscription accelerator.According to Facebook, the Accelerator project attracted thousands of digital subscriptions and email subscriptions to readers in 14 city newspapers participating in the program.The Subscription Economy Matures Whether you are a subscription, a member, or a donation, we all live in an era of audience economy.For media companies (especially the news media), this means that business motivation needs to be consistent with the needs of consumers.However, the risk is that the dissemination of subscription content will overwhelm the viewer’s willingness (or ability) to pay.If this happens, publishers will need to continue to look for new sources of revenue.Subscription products have become ubiquitous: Conde Nast announced in January that it will turn all its magazine sites (previously advertising revenue) into paid subscriptions by the end of the year.Newspapers across the country are launching new subscription products or improving existing services.Disney is rapidly developing a subscription video service that will be launched in the US in November, covering videos from Disney, ESPN and Hulu.But even though it is widely believed that digital publishing will be the future of development, paper subscriptions are still important for small local publishers.Arkansas Life magazine continued its one-year publication time after successfully launching a paper subscription in January.The magazine plans to cut publishing and move more of its business online, but it still needs paper subscribers to earn revenue.Offline Connections As users move to mobile devices, developers should also ensure that their apps are available offline.Netflix, Youtube, and Amazon Prime are now building offline browsing, allowing users to cache videos and watch them later.News content aggregation applications, such as Google, Smartnews, and Apple, also want to make the most of user time, even when the wifi signal is weak.The Washington Post’s progressive web app (which can be understood as a WeChat applet, but it runs on a web browser) reduces the load time of mobile web pages from 4 seconds to 80 milliseconds, allowing users toYou can also read the news while you are offline.The Progressive Web App is much faster than a normal browser to open a web page. Journalism as a Service In addition to traditional news products, news organizations are now also providing news services.This shift allows the media to fully realize its content value.The news service is mainly for groups working in the field of knowledge, including universities, legal startups, data science companies, businesses, hospitals and even technology giants.The content accumulated by the media is actually a data resource that can be structured, cleaned, and used by various organizations.The news as a service consists of several parts: news reports, APIs, a database that can be used by the editorial department and paid third parties; a calendar plug-in combined with news events; a system that automatically generates reports using content and databases accumulated by media organizations.Services can be provided off the social media platform, reducing the share of the news organization’s external profits, so that the services it provides can be fully realized.Pop-Up Newsrooms and Limited-Edition News Products news agencies are using pop-up newsrooms and limited-edition products to engage audiences.An integrated, collaborative newsroom can focus on a single topic or project to expand reach and discover deeper stories.Limited podcasts, news, and events identify readers interested in this and test their new ideas.In March 2019, BuzzFeed released a one-day publication in New York, which attracted a lot of attention on the Internet. This limited product may be an opportunity to develop a new audience.Media Consolidation In the United States, digital users are increasing, traditional media companies’ profits continue to decline, and advertising-based revenue models are difficult to maintain, especially for local media.The US Federal Communications Commission’s (FCC) deregulation policy has also paved the way for mergers and acquisitions and vertical integration of large media groups.In 2019, a huge number of mergers and acquisitions occurred: Disney and Fox completed the merger in March, including the transfer of control over Hulu; the court finally approved AT&T’s acquisition of Warner Media in February, dismissing the judge’sTrading is the view of anti-competitive behavior.The Monetizing Chat-Based Journalism information exchange platform has become a new center for social behavior, including information distribution and economic transaction behavior, and the two can be combined into one.The centralization of this channel creates new opportunities for reaching readers, and readers have the ability to pay in small amounts.Tencent WeChat is one of the first software to incorporate the micropayment system into its information exchange platform.Key opinion leaders such as former journalists, film critics and industry experts can now earn income through the WeChat public account.Some columnists earn up to $4,500 per article.Other major platforms, including Facebook, are also adding similar features.The End of Attention Metrics, the end of Attention Metrics, is no longer easy to measure, and traffic fraud is repeatedly prohibited.In November 2018, the US Department of Justice sued eight people for large-scale advertising fraud, with fraudulent advertising revenue totaling $36 million.They cleverly use robots to fake clicks and even mouse movements to mimic human consumers.This is a serious problem for publishers who rely on advertising revenue and advertisers who need to meet customer metrics.The decentralization and cooperation of web3.0 accelerated the development of the third generation Internet.Web 1.0 was the beginning of the Internet era, introducing static web pages, e-commerce, and email.Web 2.0 enables decentralized collaboration by introducing social networks, shared economies, cloud computing, and dynamic self-sustaining content repositories such as Wikipedia and Github.Distributed categorization drives large-scale collaboration and brings Web 3.0.In the Web 3.0 era, collaboration and decentralized creation can accelerate development for two reasons: collecting, mining, and understanding unstructured data is easier based on technologies such as data mining, natural language generation (NLP), and text parsing;AI and machine learning, machines can work directly with each other.In the end, the machines will be able to train each other.A similar project is already in progress.In the media field, Otoy creates a decentralized distributed network of partners, sharing idle processing resources and reducing the cost of visual design.In addition, in the Web 3.0 era, the media can also establish a micro-payment system, or enable users to have more control over their privacy and data.Trend 10: Media Consolidation spatial computing mixes data from the real world with personal data and digital content.Augmented reality overlays digital content projections into real-world environments, using spatial computing to import environments and perform fast calculations.Digital forms of walls, floors, and sofas are created in this way, and they still follow the laws of physics.Magic Leap is a well-known space computing company whose goggles transform the real environment into a computing environment.In the next 10 to 15 years, Magic Leap hopes to develop a city-scale spatial computing environment that allows people to interact with the real world and the digital world at the same time.Magic Digital’s digital copy of Digital Twins refers to the reproduction of a real-world object in the virtual world and allowing it to act in accordance with the instructions.Volumetric Fields of Vision Spatial computing does not capture only 2D dimensions, but also captures multidimensional data such as depth and volume.The Dynamic Light fields light field measures the light flowing in all directions.For traditional computer graphics, you can generally generate a light field by taking a high-definition photo or building a model based on the data.In spatial computing, the light field must be dynamic.Researchers are working on a dynamic light field that collects light data and then projects digital objects in a manner that mimics the human line of sight.Spatial Computing Clouds Currently, space computing systems still require hardware.But researchers want to build more features in the cloud.Storing environmental data, past experiments, and other programs between multiple users and devices in the cloud will help develop spatial computing systems.Trend 11: Intelligent electronic products, including smart watches, smart mirrors, smart doorbells, etc., can play an important role in people’s access to news and entertainment.As of now, almost all wearable devices require a smartphone or computer to view and report data, adjust settings, and archive information.But this will change as smartphones fade out of sight.Smart Camera News Networks Amazon’s Ring Smart Doorbell System includes a software called Neighbors that allows users to post their recorded videos and encourage them to post videos of suspicious activity, crime and other issues in the community.And photos.As of August 2019, a total of 225 police departments across the United States were able to request video footage from Ring Doorbell users.Even users who are not Ring can download the app for free and view the published video.The app allows the uploader to edit and annotate the video, and the news media can supplement the news with these videos about theft, fire and other events.But the application can also cause some problems.For example, racial discrimination, deliberately mislabeling other people’s photos, reporting “suspicious” activities without any real evidence.Neighbors App Smart ER Glasses In July 2019, Magic Leap selected and released the first 3D puzzle game containing the world’s most famous landscapes from its independent creator program.Spatial computing systems and ER glasses can project light directly into the user’s eyes, making digital objects appear to exist in the real world.Although Magic Leap’s glasses are not yet available to the general public, their developer platform and corporate partner ecosystem are growing rapidly.Microsoft recently demonstrated an impressive real-time application.The woman who was originally speaking in English was recreated into a realistic holographic image, and the same speech was expressed in Japanese.In-ear electronic devices (Hearables / Earables) act as our personal assistants, and colleagues who play music translate conversations for us in real time.Apple’s second-generation AirPods are compatible with Siri, and Samsung’s Galaxy Buds can provide voice assistance through its Bixby assistant.Jabra’s Elite Sport earbuds provide real-time fitness guidance, heart rate sensing and VO2 Max testing, as well as anti-sweat and water resistance.Soul Electronics’ Run Free Bio Pro earbuds capture a wealth of data such as running rhythm, gait symmetry and heart rate.Smart Watches, Rings and Bracelets Motiv Rings are fitness trackers that monitor the number of steps, heart rate and other activities as well as gestures.Users can program it to automatically log in to Amazon, Google, and Facebook accounts, and control Alexa without speaking.The Oura ring collects biometric data to give users advice on optimizing sleep and attention.PayPal’s ring can be paid by simply waving on the NFC terminal.In December 2018, the FCC approved testing of radar tracking micro-motion chips.Soli chips can be embedded in glasses, rings, and bracelets to monitor user data at any time.Motiv Ring’s voice technology interfaces such as Siri, Alexa, and Google Assistant are becoming more and more complex, but researchers have begun to contemplate connecting humans and mammals directly to computers.People can communicate with each other through these human-machine interfaces, providing new options for stroke and paralysis patients.Biomedical scientists at the University of Minnesota and Carnegie Mellon University have developed a sensor interface that allows patients to use the brainwave sensor on their head to move the cursor and control the robotic arm on the screen.Digital Addiction For digital products, it’s important to develop the habit of using digital products.More and more studies have found that these habits may have a negative impact on mental health.Through 33 studies on how people read, it has been found that readers perform more efficiently when reading from paper than screen reading.Some new products aim to find a technical solution to solve digital addiction, but how to mitigate this symptom for a long time is not yet clear.Trend 12: Blockchain Technologies is a new technology for storing and sharing information on distributed ledgers, where all transactions and identity information are protected.Cryptocurrencies blockchain technology came to an inflection point in 2017.In addition to the development of marginal digital currency to bitcoin, it has gradually become the focus of the public, a new way to share and store information.Although the technology is still evolving, its wide range of applications may affect a range of industries.In the summer of 2019, Facebook released Libra, a blockchain digital currency supported by a consortium of 28 member organizations covering the payments, telecommunications, financial technology and venture capital industries.Blockchain technology has not yet entered the mainstream so far, and with the maturity of technology in 2020, blockchain technology will continue to receive attention.Facebook released the cryptocurrency Libra’s Self-Sovereign Identity identity management system has evolved from government-issued ID cards to email accounts and social media accounts.Each person has an average of 27 to 130 online accounts.Companies like Google, Yahoo, and Facebook have built business models that manage large amounts of data on behalf of users.But in 2019 there were 15 mega-data breaches affecting 2 billion accounts in government, healthcare, finance and technology, involving organizations such as Facebook, CapitalOne, the Singapore Ministry of Health and the Bulgarian Tax Agency.Blockchain and distributed ledger technology introduces a new approach to identity management: self-sovereign identity.Autonomy has interoperability and portability across applications, devices, and platforms.Autonomy has two main advantages: security and control.For media companies, self-sovereign status will involve areas such as payment, authentication, royalties, and digital advertising.Tokens For Smart Royalties and Freelancers provide a new way to track content ownership and licensing through the use of smart contracts through blockchain networks such as Ethereum.A smart contract is a self-executing protocol, and the terms of the agreement are written directly into the code.For example, each time a song is played, the money is automatically deducted from the audience and flows to the artists.Blockchain technology can be used to better maintain copyright.In the future, centralized content platforms will have a platform for providing maximum ownership and rewards to creators.In this platform, the creator will receive most of the revenue, rather than providing most of the revenue to the distribution platform.At the same time, the creator will retain more ownership and interact directly with the audience.Blockchains and distributed ledgers are changing the incentive structure for content management and consumption patterns, and voting to support content in exchange for compensation.This will affect many industries, from online gaming to fashion to retail, from tourism to automakers, and even to the political movements of 2020.Blockchain Decentralization and Copyright Protection The Content Provenance and Permanent Archiving blockchain technology creates a shared permanent ledger where no content can be deleted.Therefore, adding the original content or index to the blockchain is one way for journalists to permanently save their content and trace it back.Trend 13: Security and privacy Entertainment, media and technology companies will continue to face new security and privacy challenges throughout 2020.More important than ever, every organization must take proactive steps to protect user and company data, perform penetration tests on a regular basis to identify vulnerabilities, and periodically update passwords.Every organization should make a crisis plan for the worst.Right To Eavesdrop/ Be Eavesdropped On As more and more mobile phones, mobile devices, and intelligent software connect to the Internet of Things, the interaction between these devices and the companies that manufacture them will continue to expand.Our devices are no longer just communicating with each other, but trying to understand us and talk about us.News and entertainment companies need to determine whether these devices violate ethical guidelines when exchanging consumer data.When communicating data between smart devices, we can’t ensure that they are beyond the moral bottom line and privately accessing our information.Intelligent Voice Assistant is exposed to “eavesdropping” users Encrypted Messaging Networks In the past year, journalists have used closed-encrypted messaging networks such as Keybase and Signal.However, many news organizations still do not have guidelines on how to use these networks.In response to social media hacking and government-funded surveillance programs around the world, private networks will continue to be popular in 2020.Media Trolls refer to cyberspace-specific spam, hate speech, swearing and humiliation.Twitter, Facebook, and Instagram have updated community standards to limit hate speech.Reddit prohibits groups such as r / Incels from violating the site’s community standards (although there is still a lot of terrible content on the rest of the site).Authenticity cyberspace is really difficult to identify.Facebook has partnered with the Poynter Institute International Facts Inspection Network to combat false news on its platform.However, this partnership itself is difficult to monitor and further illustrates Facebook’s dominant influence on digital media.The authenticity of the media has expanded from fake news to fake video.Deepfake is a computer-generated facial interchange video that originated in Redditin and accumulated more than 80,000 subscribers before shutting down.Deepfake’s Data Ownership information sovereignty includes not only IP and copyright, but also personal behavior, health data and network activities in today’s smart devices.For example, photos uploaded to Facebook and videos posted on YouTube.At the end of the day we may feel the speed of the technology update iteration slows down, but it is undeniable that we are gradually entering a new era of technology with artificial intelligence as the mainstream.In this new era, we still face a severe test: verifying the authenticity of the news is still tricky, and all kinds of fraudulent applications have become more subtle; in the new world of synthetic media, machine writing has liberated manpower, but broughtMore serious algorithmic specification problems; digital subscription failures, audience loss plagued all traditional media and new media.How to capture the limited attention of the audience in this crowded market will continue to linger in the hearts of every media person in the next few years.However, in the face of this uncertain future, no one has given up.The General Data Protection Regulations adopted by the European Union have been recognized around the world, and various countries are promoting legislation related to network and data security. Algorithm writing has been thought to replace journalists, but now it helps the media to dig deeper and deeper.The story; blockchain technology has experienced a downturn, but it remains the primary choice for protecting media copyrights and traceable data.The forward development of technology is an irreversible trend.However, no matter how unpredictable the world is, the initial use of technology to improve human life has never changed..