SenuthDeelana.com
10 technology trends that will impact our lives in 2020
Image Credit: Getty Images
Presented by Pegasus Tech Ventures
2020 — it’s when the world will see transformational changes in how technology impacts our lives. Here’s a look at the top technology trends that will influence us.
Trend 1: Breakout moment of artificial intelligence (AI) in manufacturing
AI is now part of everyday life, driven by the emergence of a device ecosystem including Alexa, Siri, and Google Assistant. In 2020, emotion recognition and computer vision will scale and AI will have a breakout moment in manufacturing.
U.S. startups Vicarious, Kindred, and Osaro stand out in using AI technologies for manufacturing. Kindred’s technology is used to automate part of distribution for apparel brands such as GAP. Vicarious is attracting investment from Mark Zuckerberg, Jeff Bezos, and Elon Musk.
Trend 2: Practical deployment of Internet of Things (IoT)
IoT is hot but there are not as many mainstream applications today as some predicted. We anticipate that with 5G, the number of connected devices and mainstream IoT applications will reach scale.
Amazon recently launched Amazon Go, a system that uses IoT and machine vision technologies to enable consumers to shop without manual check out. Environments will use more sensors and vision technologies, enabling more scalable IoT solutions. Startups Standard Cognition (U.S.), Accel Robotics (U.S.), Trigo (Israel), Grabango (U.S.), and AiFi (U.S.) provide similar services as Amazon Go.
Trend 3: Increased demand for edge computing processing power
2020 will see the need for higher performance from edge computing hardware since better sensors and larger AI models now enable a host of new applications. There is a growing need to infer more data and then make decisions without sending data to the cloud.
Chip startups SambaNova (U.S.), Graphcore (U.K.), Cerebras (U.S.), Wave Computing (U.S.), and Syntiant (U.S.) have developed architectures to handle increased demand. High-performing AI chips, known as neuromorphic or brain chips, mimic the structure of the brain and process top AI algorithms.
Trend 4: Commercialization of quantum computing usage in mass scale
We expect 2020 to begin the quantum computing era. As data increases, quantum computing will target the biggest problems in industry, such as health care and energy. In 2020, the ability to handle big data will be required for cancer treatment, nuclear energy control, and DNA analysis. Corporations IBM, Google, Intel, Microsoft, and Alibaba have moved into quantum computing.
Startups Rigetti (U.S.), D-Wave Systems (Canada), and QC Ware (U.S.) are disrupting quantum computing. The technology will grow as it becomes easier to use with platforms such as Amazon Web Services by mid-2020.
Trend 5: Evolution of aerospace technologies
Mankind will begin its return to space in 2020, largely driven by the private sector. Since the Cold War, technological advances have slowed. Notable companies now making aerospace advancements include SpaceX and Blue Origin.
SpaceX is developing the rocket Starship, which will reuse the entire vehicle body. Starship may shorten intercontinental trips to 20-30 minutes via space. In China, the government and private enterprises (example: LinkSpace) are making progress in space.
Trend 6: New era of the internet — deployment of 5G and Starlink broadband internet technology
5G competition between the U.S. and China is entering the main stage in 2020. There will be a new competition about who can propel 5G faster into mass consumer use. Another hot topic is the Starlink Broadband business planned by SpaceX.
Until 2020, as many as 2,500 satellites will be launched. This marks a new era of broadband Internet for some users in North America. Starlink’s broadband Internet system will grow with 12,000 satellites through 2023, followed by the addition of 30,000 satellites. SpaceX will provide higher speed Internet starting in 2020.
Trend 7: Evolution of health care — predictions at the genome level
Under the theme of prevention, digital health care has seen much innovation. In the U.S., startups 23andMe and Color lead in genome analysis, while Genesis Healthcare stands out in Japan and Genoplan in Korea. These companies use genomic analysis to learn of diseases and provide prevention methods.
The evolution of AI will improve the quality of treatment. In 2020, many medical images taken using MRI, CT scans, and X-rays will be diagnosed using AI. Startups Enlitic and Zebra Medical Vision stand out as leaders.
Trend 8: Evolution in Agriculture – technology to Grow Crops Efficiently
In agriculture, companies that offer products using computer vision, AI, and big data stand out. In 2020, it will become common to monitor crop growth by computer vision Ceres Imaging (U.S.), Taranis (Israel), Farmwise (U.S.).
Robots, such as those by Abundant Robotics, that harvest plants and fruits will become more common. The technology for improving crop growth efficiency will also be enhanced by indoor farming companies, such as Bowery Farming (U.S.), funded by GV.
Trend 9: Evolution of autonomous driving technology
Autonomous driving is already a hot topic, although level 5 (fully autonomous) has not been realized. Some Tesla cars can be switched to autopilot mode on the highway, but this is only possible up to level 2 (driving support) or level 3 (operated by the driver in an emergency).
Technology for understanding detailed road conditions by AI is evolving. Startups Prophesee (France), Perceptive Automata (U.S.), and Humanising Autonomy (U.K.) stand out as leaders. They will contribute towards achieving level 5 in 2020.
Trend 10: The U.S. and China put blockchain to practical use
As blockchain grows, payment-type venture companies and venture companies in security — such as Chainalysis, which develops money laundering countermeasure technology — are attracting attention.
In 2020, major institutions will introduce blockchain to prevent large-scale information leakage and Internet fraud. IBM set up an accelerator program specializing in blockchain. China approved the introduction of blockchain in services such as ICBC (China Industrial and Commercial Bank), Alibaba Group, China Southern Airlines, etc. In 2020, blockchain will be put to practical use.
Technologies changing the world
These 10 technologies will change the world in 2020. While there are concerns about workforce impact, there is little doubt that 2020 will be a time of innovation as people better leverage technology.
Anis Uzzaman is CEO of Pegasus Tech Ventures. Located in Silicon Valley, Pegasus Tech Ventures provides early stage and final round funding. With several multi-million dollar funds under management, Pegasus Tech Ventures focuses its investment in IT, Health IT, Artificial Intelligence, IoT, Robotics, Big Data, Virtual Reality, Augmented Reality, FinTech and Next Generation Technologies.
Anis has invested in over 170 startups in the U.S., Japan, and South East Asia. Some of the prominent U.S. startups in the Pegasus Tech Ventures portfolio include Vicarious, Color Genomics, Genius, Affectiva, Afero, x.ai, and ShareThis. Anis is an investor and board member of Tech in Asia, the largest tech media blog in Southeast Asia. Anis also sits on the board of directors of Affectiva, Lark, Asteria, and I AND C-Cruise.
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact sales@venturebeat.com.
McAfee: Start protecting against quantum computing hacks now
Above: Steve Grobman, CTO of McAfee, at RSA 2020.
Image Credit: RSA
McAfee’s chief technology officer warned that it’s time for companies to start worrying about quantum computing attacks that can break common forms of encryption available today, even if quantum computing isn’t going to be practical for a while.
Steve Grobman, CTO of the cybersecurity firm, made the remarks in a keynote address at RSA, the big security conference in San Francisco this week.
Is it important to protect today’s data from attacks that may be completed in the future?, Grobman asked. He noted that National Archives files related to national security from the Kennedy assassination nearly 60 years ago still have redactions for current national security risks today.
“We need quantum-resistant algorithms as soon as possible,” Grobman said.
Cloud computing is sweeping through the industry, and it will enable the use of quantum computing. And that’s a problem, as quantum computers may be able to break encryption techniques such as RSA encryption much faster than traditional computers can. Typically, encryption techniques make it easy to encode data but hugely difficult to decode it without the use of a special key. The security is possible only because of the huge amount of time it takes for a classical computer to do the computations.
Above: McAfee’s Steve Grobman is worried about quantum attacks now.
Image Credit: RSA/VentureBeat
Binary digits — ones and zeroes — are the basic components of information in classical computers. Quantum bits, or qubits, are built on a much smaller scale. And qubits can be in a state of 0, 1, or both at any given time. These computers can handle extremely complex calculations in parallel, but they require a huge amount of manufacturing precision just to be accurate. Quantum computers could be used to achieve breakthroughs in biology, chemistry, physics, and even cybersecurity.
Companies like Intel and IBM are working on improving the speed, accuracy, and cost. But it may take years before the improvements take hold and give quantum computing a chance to beat classical computers. If the quantum computing speeds up dramatically and arrives sooner than expected at practical prices, then the safety of today’s encryption techniques will be compromised, Grobman said.
“I’m realistic enough to know that nation-states will use quantum to break our public key cryptography system,” Grobman said. “Now I know what you are thinking: Quantum is not coming anytime soon. But we can’t think of quantum in terms of eventually or tomorrow. Because quantum is a real risk today. You must assume that adversaries are already accessing your most sensitive data. It’s encrypted, but they still find it valuable. They’re not worried about [decrypting] it today. They’re counting on quantum to do that in the future.”
Grobman said cybercriminals can siphon off data today and unlock it when quantum cryptoanalysis becomes practical, he said. So companies have to consider the sensitivity of their data and how long it must be protected. For names matched to social security numbers, that’s a long time, as an example.
Above: Data going back decades still has to be protected.
Image Credit: RSA
Grobman noted that the federal budget for quantum computing research is just $30 million. That’s 0.006% of the federal budget to solve a problem that is a threat to national security. Selecting the right algorithm isn’t easy, as 69 replacement algorithms have been proposed and 12 were broken or attacked within three weeks. After three years, the field has been narrowed to 26. Government and industry need to work together to retool systems that are based on quantum-vulnerable attacks, Grobman said.
“Let’s all commit to build post-quantum action plans that measure time and impact sensitivity so we’re ready to migrate systems,” Grobman said.
Meanwhile, Grobman also said that companies have to be aware of “cloud-native threats.” These are designed to attack the unique technical nature of the cloud. Cloud computing architects snap together new building blocks that build sophisticated technologies. But configuration errors happen, and since such systems are based on public networks, those errors open the door to catastrophic attacks. Some epidemiologists made this mistake, he said, and it exposed their data to the whole internet.
Grobman also compared how fast computer viruses spread to biological viruses such as the flu. He noted how, just a few years ago, the EternalBlue computer virus spread around the globe infecting a quarter of a million machines in 150, going “from an epidemic to a pandemic in one day.” But like the flu, those infections come back every year because a significant number of machines still aren’t patched.
The ability of computer malware creators to create variants exacerbates the severity of attacks. The CurveBall vulnerability this year can be exploited with just 10 lines of code. That makes it easy to use and create variants. The barrier for causing havoc is far lower today, exacerbating the challenge of fighting off the attacks. One attack, SHA-1 collision, in 2017 took 6,610 years of processor time to orchestrate an attack. Today, it can be done in less than 60 seconds. That puts digital signatures using hashes are at risk, Grobman said. Patching against such problems helps, but the implementation of patches across an entire user base is difficult.
On Monday, McAfee acquired Light Point Security to bring its browser isolation technology to its cloud security solution.
Securiti.ai named most innovative startup at RSA
Securiti.ai, a company that uses AI to help companies comply with privacy laws like GDPR in the European Union and CCPA in California, today was named winner of the RSA Sandbox Innovation award.
Securiti creates digital personas for individuals and finds copies of data shared across systems or with third-party vendors or partners to help companies comply with right-to-be-forgotten laws. It uses natural language for data requests and other forms of AI to allow those companies to assess vendors or internal systems.
This is the 15th year a panel of entrepreneurs and security professionals chose a most innovative startup at RSA, a security conference happening this week in San Francisco. Top 10 finalists in the contest have seen 56 acquisitions and more than $6.2 billion in investments since the award started, RSA said in a statement.
Securiti.ai raised a $50 million series B funding round led by General Catalyst last month, after the company launched in stealth with $31 million in funding last August.
“Privacy is a basic human right, and companies want to honor individual rights of privacy and data protection. Privacy compliance and operations are only getting more complex for businesses around the world, and we’re humbled that the judges recognized our vision for AI-powered PrivacyOps and data protection,” Securiti.ai CEO Rehan Jalil said in a statement today.
In a conversation onstage today during opening addresses, leading cryptographers talked about trends like SIM swap attacks in mobile devices and encryption debates with governments, but started by discussing machine learning, and how the right to be forgotten impacts the work of security professionals.
“The two major problems in AI, in my opinion — the first is we don’t know why they work so well and the second is we don’t know why they work so well,” Adi Shamir, The Weizmann Institute professor and RSA algorithm cocreator, said. RSA stands for Rivest–Shamir–Adleman.
How AI is transforming cybersecurity and may someday play a role in attacks was the topic of VentureBeat’s second special issue, released last week.
Citing the limitations of facial recognition software today, MIT professor Ron Rivest, who also cocreated the RSA algorithm, suggested bans and moratoriums on facial recognition software by government regulators should continue.
Princeton University professor Arvind Narayanan talked about differential privacy use by the U.S. Census. Differential privacy is also being used by companies like Apple and Google for techniques like federated learning, and he called for neutral people to advise governments and the public on how that kind of tech works.
Qualcomm and ZeroLight debut end-to-end 5G Boundless XR system
Image Credit: Qualcomm
There were plenty of questions left unanswered when Qualcomm announced its 5G-ready Snapdragon XR2 mixed reality chipset last December, including the timetable for actual headsets containing both the core XR2 chip and its optional 5G module. Today, Qualcomm and ZeroLight are announcing a complete end-to-end solution that will use the XR2 and 5G for Boundless XR in enterprise settings.
The broad premise of “Boundless XR” is to replace the wires typically needed for mixed reality headsets with high-bandwidth wireless connections, giving the lightweight, low-power headset access to the photorealistic rendering power of a 300-watt computer. While a somewhat obscure flavor of Wi-Fi known as WiGig or 60GHz Wi-Fi is one option, millimeter wave 5G is another, assuming there are 5G base stations nearby — something that enterprises are now considering with private 5G networks.
For this end-to-end solution, Qualcomm provides the reference design for XR2-based headsets and chips for 5G network hardware, while ZeroLight delivers the mixed reality application and edge cloud rendering server for the private 5G network. Millimeter wave frequencies will maximize the bandwidth and speeds between the headsets and servers, enabling the headset’s display resolution to be as high as possible, with nearly zero latency.
Rather than waiting for carrier-provided 5G infrastructure, this end-to-end 5G network will operate fully within the enterprise’s premises. Because of the tight connection between the headset and edge cloud server, real-time rendering can be split between them, with the headset handling input tracking and some latency-sensitive, low-power tasks while the server does heavier lifting.
The deal makes sense given that ZeroLight has specialized in creating premium automotive mixed reality experiences, and long planned to enable car dealerships to offer tetherless walkarounds of digital vehicles — or digitally augmented real vehicles — in real spaces. Current solutions require a tower-sized computer and/or a long wire to occupy physical space inside each area the dealership wants to use for demos; this solution moves all the hardware except the small headset off the show floor, freeing customers to move around.
ZeroLight is using Nvidia PC-based rendering hardware and CloudXR streaming to a 5G network. The initial retail demonstration uses multiple Pagani vehicles, including a choice of paint colors, alloys, and interior trims.
Equipped with this solution, a dealership will let a customer select all of the customization options for the vehicle, then walk around a photorealistic, fully customized vehicle even if there isn’t one on the lot. This will help customers make on-the-spot purchasing decisions that they might otherwise pass on, absent the chance to see things up close. Since over 35 companies are now participating in Qualcomm’s XR Enterprise Program, there’s a greater likelihood that similar applications will become available soon for remote collaboration, training, health care, and industrial use.
Qualcomm also says that it’s making a “superset” Snapdragon XR2 reference design broadly available to developers, including most of the major features the chipset supports, including 60GHz Wi-Fi and Ericsson-validated 5G wireless abilities, seven cameras, twin high-resolution LCD screens, 6DoF controllers, foveation, eye tracking, and hand tracking. Qualcomm is currently working with partner Goertek on streamlined industrial designs, showing how purpose-optimized VR and AR versions of the hardware will be at least as small and light as well-known existing solutions.
Electronic Arts pulls out of GDC over coronavirus concerns
Above: Electronic Arts doesn't want any of its staff traveling while fear of coronavirus remains high.
Image Credit: Electronic Arts
Electronic Arts is not going to officially participate in the Games Developers Conference in San Francisco in March. The massive publisher told staff on Monday, February 24, that it is limiting its presence at the annual gathering of game creators. This comes amid growing concern over the spread of coronavirus (COVID-19) to more countries in Europe and Asia.
EA has over 9,000 employees with some located in countries that are actively fighting the spread of the virus. Putting a hold on travel should limit the company’s exposure to coronavirus. That in turn should limit its exposure to the financial risk of dozens of employees missing work due to illness.
Here is the note that EA chief people officer Mala Singh sent out to EA employees last night:
“We will be limiting our presence at the Game Developers Conference (GDC) in San Francisco, March 16-20. We are cancelling our participation in official GDC events. And employees should not travel to San Francisco for the conference. For meetings that are set to take place outside of the conference and not within the event venue, we will evaluate on an individual basis if these should take place. If you have meetings planned around GDC and are unsure if you should move forward, please speak with your manager.”
While the CDC is not issuing any travel warnings for U.S. cities like San Francisco, EA doesn’t want to take any chances. It is also not the only company to pull out of GDC.
Sony pulled out of GDC and PAX East. Facebook and Oculus announced it would not participate as well. Japanese studios at Capcom, Square Enix, and Kojima Productions also cancelled plans to attend the conference.
I reached out to GDC for a comment. I’ll update this post with any new statement from the organizers.
EA staff work from home in affected countries
Fear of coronavirus is spreading as infections continue to show up in more countries around the world. While much of China is still dealing with lockdowns and quarantines, similar responses are beginning to take shape in South Korea, and Italy.
EA has employees in all of those countries. And for its part, it is doing its part to limit the spread of infection.
“In addition to the EA China team, employees in our South Korea and Italy offices are now working from home in alignment with local guidance following increased instances of exposure in those regions,” Singh wrote in her note to employees.
So EA is just one more company trying to continue business as usual while also looking over its shoulder at a potential pandemic.
Dota Underlords Season One kicks off with Battle Pass and more
Above: Some of the villains aiming to take over the city of White Spire in Dota Underlords Season One.
Dota Underlords is officially out of Early Access. Publisher Valve Software is launching Season One of its popular auto battler today. This content drop introduces a new Battle Pass and modes. You can start playing Dota Underlords for free on PC, Mac, Linux, iOS, and Android, and all of your progress carries over between those platforms.
Like other auto battlers, Dota Underlords is a team-based multiplayer game. You build a squad of minions with the goal of unlocking their synergies and upgrades to take out and outlive your opponents. But in Season One, Valve is adding new multiplayer and solo modes.
These new modes all relate to the ongoing narrative for Dota Underlords Season One. The city of White Spire has found itself in the midst of a power vacuum. Someone killed the city’s matriarch, and now smugglers and other unsavory types are fighting to take her place. Those include a demon summoner, a wine-guzzling anthropomorphic walrus, and more. In one of these new modes, you choose from the underlord heroes and then work through a series of stages and puzzles.
Above: Dota Underlords gets new modes for Season One.
Image Credit: Valve Software
Dota Underlords Season One is also getting a Battle Pass. This enables players to progress through tiers to unlock cosmetic items. If you purchase the premium Battle Pass, you will unlock more accessories at a faster rate. Valve first popularized the Battle Pass with its annual Compendium content for Dota 2. Now, it’s using that as the primary business model for Underlords.
Flock Freight raises $50 million to match shippers with freight transportation
Heavy-duty trucks play an indispensable role in transporting freight across contents — and the globe. In 2017 in the U.S., trucks carried 60.2% of the $614.0 billion of goods imported from Canada and Mexico. But while large enterprises appear to have a handle on an industry that was worth over $700 billion three years ago, small- and medium-sized businesses often struggle to find ground transportation suitable for smaller and partial freight loads.
If you ask Oren Zaslansky, the solution lies in a digital AI-driven marketplace that pools less-than-truckload (LTL) and partial-truckload (PTL) freight shipments so they can be shipped via a full truckload service. That’s the service Zaslansky’s startup, Flock Freight (previously AuptiX), provides, and it’s attracted the attention of investors including Google Ventures.
Flock Freight today announced that it’s raised $50 million in series B funding, bringing its total raised to over $70 million following an $18 million series A round and $2.5 million seed round. Zaslansky says it’ll use the capital infusion to expand FlockDirect, its marketplace. “Flock Freight is pioneering the hubless shipping market, benefiting both the shipper and the carrier,” said Zaslansky. “As we continue to prioritize exceptional service and affordability for shippers and increased revenue potential for carriers, we look forward to expanding Flock Freight and our partnerships with carriers throughout the U.S.”
For LTL, Flock Freight facilitates the travel of shipments on trucks to their intended destinations, eschewing the traditional hub-and-spoke freight transit model. In the case of PTL — which Flock Freight defines as a maximum of 24 pallets that take up to 48 feet in deck space, weigh under 40,000 pounds, and are headed to a single location — the platform finds as many as 10 trucks along a single route and pools them together into a single truckload to maximize cost savings.
Flock Freight says its driver network in the U.S. and Canada numbers in the thousands, each individual driver within which can be tracked in real time via a dashboard or email notifications. Furthermore, the company says its damage claim rate is 0.001% and that its on-time delivery rate is 97.5%, and that it’s able to reduce fuel emissions by up to 40% by eliminating the need to switch trucks or stop at warehouses.
Zaslansky makes the case that those stats set it apart from competitors in the freight logistics space. Uber offers a service called Uber Freight, to which it recently committed another $200 million as part of a major expansion. San Francisco-based startup KeepTruckin recently secured $149 million to further develop its shipment marketplace, and Next Trucking closed a $97 million investment. For its part, Convoy raised $400 million at a $2.75 billion valuation to make freight trucking more efficient.
“The LTL shipping industry is ripe for an overhaul, as slow service, paper logs, and missing goods have become the standard,” said Zaslansky. “The multi-billion-dollar industry and the millions of businesses it serves deserve modernization, and Flock Freight is well-positioned to make this happen, given our broad customer base, years of experience and proprietary technology.”
SignalFire and GLP Capital Partners co-led this latest round.
Amazon pilots AI-powered customer support agents
Above: The logo of Amazon is seen at the company logistics center in Boves, France, September 18, 2019.
Image Credit: Reuters / Pascal Rossignol
Might AI help improve customer service for the millions of people who shop on Amazon.com? Amazon intends to find out. In a blog post, the Seattle tech giant revealed that it’s testing two AI-based systems to handle incoming shopper inquiries. One fields requests from customers automatically and without human intervention, while the other helps human service agents respond more quickly and easily.
As Jared Kramer, an applied-science manager in Amazon.com’s customer service technical management organization, explained in a blog post, the automated agents use machine learning rather than rules and refer requests they can’t handle to human representatives, enabling them to tackle a broader range of interactions. That’s as opposed to Amazon.com’s old flow chart system that specified responses to particular inputs.
“It is difficult to determine what types of conversational models other customer service systems are running, but we are unaware of any announced deployments of end-to-end, neural-network-based dialogue models like ours,” wrote Kramer. “And we are working continually to expand the breadth and complexity of the conversations our models can engage in, to make customer service queries as efficient as possible for our customers.”
Above: An example of how we converted raw interaction transcripts into training examples, which include relevant information from the customer’s account profile.
Image Credit: Amazon
Amazon says that in the customer-facing system, it’s using a template ranker — where an AI model chooses among hand-authored response templates — that allows it to control the automated agent’s vocabulary. (It plans to soon begin testing a generative model that crafts responses to replies on the fly.) The templates are general forms of sentences, with variables for things like product names, dates, delivery times, and prices, and the model is able to incorporate new templates with little additional work because it’s pretrained on a data set of interactions between customers and representatives. Basically, because the template ranker has seen many responses that don’t fit its templates, it’s learned over time several general principles for ranking arbitrary sentences.
Researchers at Amazon trained separate versions of each model for two types of interactions: return refund status requests and order cancellations. As an input, the order cancellation model receives not only the dialogue context but also some information about the customer’s account profile. In addition to the context and the profile information, the response ranker receives a candidate response as input, and it uses what’s known as an attention mechanism to determine words in previous messages that are particularly useful for ranking the response.
During randomized trials that compared the agents to existing rule-based systems with a metric called automation rate, the new agents significantly outperformed the old ones, according to Kramer. “Automation rate combines two factors: whether the automated agent successfully completes a transaction (without referring it to a customer service representative) and whether the customer contacts customer service a second time within 24 hours,” he said. “According to that metric, the new agents significantly outperform the old ones.”
Amazon’s chatbot adoption is on-trend — Gartner predicts they’ll power 85% of all customer service interactions by the year 2020, and there’s a good reason for the continued growth. Roughly 62% of customers are open to the use of AI to improve their experiences and an estimated 30% of U.S. service positions could be automated through chatbots, saving an estimated $23 billion in annual salaries.
Newzoo: Global esports will top $1 billion in 2020, with China as the top market
Above: Newzoo's latest revised forecast for esports revenue growth.
Image Credit: Newzoo
Global esports revenues will surpass $1 billion in 2020 for the first time — without counting broadcasting platform revenues, according to market researcher Newzoo.
China is the largest market by revenues, with total revenues of $385.1 million in 2020, followed by North America, with total revenues of $252.8 million. Newzoo noted that it has re-evaluated the size of the esports market, based on improved methodologies. Some media have been critical of Newzoo’s hype around esports in the past.
Globally, the total esports audience will grow to 495.0 million people in 2020, Newzoo said. Esports Enthusiasts (people who watch more than once a month) make up 222.9 million of this number.
In 2020, $822.4 million in revenues—or three-quarters of the total market—will come from media rights and sponsorship.
Above: Esports audience growth
Image Credit: Newzoo
In the coming year, the global esports economy will generate revenues of $1.1 billion, a year-on-year growth of 15.7%. Most of these revenues (74.8%) will come from sponsorships and media rights, which will total $822.4 million, a 17.2% increase from last year.
Consumer spending on tickets and merchandise will total $121.7 million, while another $116.3 million will come from game publishers’ investments into the esports space, via supporting tournaments through partnerships or as white-label projects with professional tournament organizers.
Newzoo also said the global esports audience will reach 495.0 million this year, made up of 222.9 million Esports Enthusiasts and a further 272.2 million Occasional Viewers. In 2020, the average revenue per Esports Enthusiast will be $4.94, up 2.8% from 2019.
“As the esports market matures, new monetization methods will be implemented and improved upon,” said Remer Rietkerk, head of esports at Newzoo, in the report. “Likewise, the number of local events, leagues, and media rights deals will increase; therefore, we anticipate the average revenue per fan to grow to $5.27 by 2023.”
Above: Newzoo’s forecast for the top revenue streams for esports.
Image Credit: Newzoo
Mobile has unlocked esports for emerging markets—a trend visible in countries like Vietnam, where titles like PUBG Mobile and Garena Free Fire have exploded in popularity.
As such, emerging esports markets will show the highest compound annual growth rate (CAGR for 2018-2023), with regions such as Southeast Asia (24.0% CAGR), Japan (20.4%), and Latin America (17.9%) accelerating to close the gaps between themselves and older, more developed esports markets.
China will remain the largest esports market in 2020, with revenues of $385.1 million. These revenues will grow with a CAGR (2018-2023) of 17.0% to reach $540.0 million by 2023. Most of these revenues will come from sponsorships, which will grow from $187.1 million in 2019 to $222.4 million in 2020.
Digital goods will be the fastest-growing revenue stream toward 2023, growing from $7.1 million in 2020 to $17.2 million by 2023. North America will be the second-largest region in terms of revenues with $252.5 million, followed by Western Europe as the third-most revenue-generating region with $201.2 million in 2020. China will be host to the largest esports audience with 162.6 million in 2020, followed by North America with an audience of 57.2 million.
Rietkerk said, “Our data highlights that 2019 was a seminal year for many teams, with tremendous growth in traditional revenue streams such as sponsorship. Meanwhile, leagues have been moving toward a ‘homestand’ system in which teams play at their own venues. This potentially opens the door to increased matchday revenues for teams, including returns from ticketing and concessions, as well as larger merchandise revenues.”
And Rietkerk added, “The market is also maturing in entirely new ways, with innovative revenue streams starting to develop, such as streaming and digital goods. These are new ways to monetize that are not available to traditional sports; they also demonstrate a growing understanding of the competitive advantages esports has over sports. These revenue streams have become pioneering ways for teams, organizers, and publishers to grow the business.”
With all of the dynamic changes, Rietkerk acknowledged that Newzoo needs to keep its model for the industry fresh.
In 2019, there were 885 major events. Together, they generated $56.3 million in ticket revenues, up from $54.7 million in 2018. Total prize money in 2019 reached $167.4 million, a slight increase from 2018’s $150.8 million.
Above: Esports audience size
Image Credit: Newzoo
The League of Legends World Championship was 2019’s biggest tournament by live viewership hours on Twitch and YouTube, with 105.5 million hours. The Overwatch League was the most-watched league by live viewership hours on Twitch and YouTube, generating 104.1 million hours.
The esports audience will grow to 495.0 million globally in 2020. Esports Enthusiasts will account for 222.9 million of this number, up 25 million year on year, and will increase with a CAGR (2018-2023) of 11.3% to 295.4 million in 2023.
Meanwhile, the number of global Occasional Viewers will hit 272.2 million in 2020, up from 2019’s 245.2 million. This number will grow with a CAGR (2018-2023) of 9.6% to 351.1 million in 2023.
In 2020, 2.0 billion people will be aware of esports worldwide, an increase from 2019’s 1.8 billion. China will continue to be the country/market that will contribute most to this number, with 530.4 million esports-aware people.
OpenAI’s Jeff Clune on deep learning’s Achilles' heel and a faster path to AGI
Exclusive
Above: OpenAI's Jeff Clune talks about ANML at the ReWork Deep Learning Summit held Jan. 30 in San Francisco, California
Image Credit: Khari Johnson / VentureBeat
Neural networks learn differently from people. If a human comes back to a sport after years away, they might be rusty but they will still remember much of what they learned decades ago. A typical neural network, on the other hand, will forget the last thing it was trained to do. Virtually all neural networks today suffer from this “catastrophic forgetting.”
It’s the Achilles’ heel of machine learning, OpenAI research scientist Jeff Clune told VentureBeat, because it prevents machine learning systems from “continual learning,” the ability to remember previous tasks. But some systems can be taught to remember.
Before joining OpenAI last month to lead its multi-agent team, Clune worked with researchers from Uber AI Labs and the University of Vermont. This week, they collectively shared ANML (a neuromodulated meta-learning algorithm), which is able to learn 600 sequential tasks with minimal catastrophic forgetting.
“This is relatively unheard-of in machine learning. To my knowledge, it’s the longest sequence of tasks that AI has been able to do experiments to, and at the end of it, it’s still pretty good at all the tasks that it saw,” Clune said. “I think that these sorts of advances will be used in almost every situation where we use AI. It will just make AI better.”
Clune helped cofound Uber AI Labs in 2017, following the acquisition of Geometric Intelligence, and is one of seven coauthors of a paper called “Learning to Continually Learn” published Monday on arXiv.
Teaching AI systems to learn and remember thousands of tasks is what the paper coauthors call a long-standing “grand challenge” of AI. Such systems can enable the creation of AI systems that can handle and remember a range of tasks, and Clune believes AI like ANML is key to achieving a faster path to the greatest challenge: artificial general intelligence (AGI).
In another paper Clune wrote before joining OpenAI — a startup with billions in funding aimed at creating the world’s first AGI — he argued a faster path to AGI can be achieved through improving meta-learning algorithm architectures, the algorithms themselves, and the automatic generation of training environments.
“If you had a system that was searching for architectures, creating better and better learning algorithms, and automatically creating its own learning challenges and solving them and then going on to harder challenges … [If you] put those three pillars together … you have what I call an ‘AI-generating algorithm.’ That’s an alternative path to AGI that I think will ultimately be faster,” Clune told VentureBeat.
ANML has made progress by meta-learning solutions to problems instead of manually engineering solutions. This is in keeping with a move toward searching for algorithm architectures to find state-of-the-art results rather than hand-coding algorithms from scratch.
Last week, an MIT Tech Review article argued that OpenAI lacks a clear plan for reaching AGI, as some of its breakthroughs are the product of computational resources and technical innovations developed in other labs. The average OpenAI employee, the article said, believes it will take 15 years to reach AGI.
An OpenAI spokesperson told VentureBeat that Clune’s vision of AI-generating algorithms is in line with the organization’s research interests and previous work, like a model for a robotic hand to solve a Rubik’s Cube. But the individual did not share an opinion on Clune’s theory about a faster path to AGI.
The spokesperson also declined to answer questions about a roadmap to AGI or comment on the MIT Tech Review article, but said that Clune, as head of the multi-agent team, will focus on AI-generating algorithms, multiple interacting agents, open-ended algorithms, automatically generating training environments, and other versions of deep reinforcement learning.
“They hired me to pursue this vision and they’re interested — a lot of their work is very aligned with this vision, and they like the idea and hired me to keep working on it in part because it’s aligned with work they’ve published,” Clune said.
ANML, neuromodulation, and the human brain
ANML achieves state-of-the-art continual learning results by meta-learning the parameters of a neurmodulation network. Neuromodulation is a process found in the human brain, where neurons can inhibit or excite other neurons in the brain, including inciting them to learn.
Above: Meta-test training classification accuracy
In his lab at the University of Wyoming, Clune and colleagues demonstrated that they could completely overcome catastrophic forgetting, but only for smaller, simpler networks. ANML scales catastrophic forgetting reduction to a deep learning model with 6 million parameters.
It also expands upon OML, a model that was introduced by Martha White and Khurram Javed at NeurIPS 2019 and is capable of completing up to 200 tasks without catastrophic forgetting.
But Clune said ANML differs from OML because his team realized turning learning on and off is not enough on its own at scale; it’s also necessary to modulate the activation of neurons.
“What we do in this work is we allow the network to have more power. The neuromodulatory network can kind of change the activation pattern in the normal brain, if you will, the brain whose job it is to do tasks like ride a bike, play chess, or recognize images. It can change that kind of activity and say ‘I only want to hear from the chess-playing part of your network right now,’ and then that indirectly allows it to control where learning happens. Because if only the chess-playing part of the network is active, then according to stochastic gradient descent, which is the algorithm that you use for all deep learning, then learning will only happen more or less in that network.”
In initial ANML work, the model uses computer vision to recognize human handwriting. For next steps, Clune said he and others at Uber AI Labs and the University of Vermont will scale ANML to try to accomplish more complex tasks. Work on ANML is supported by DARPA’s Lifelong Learning machines award.
YouPorn launches 'TikTok-inspired' video app that learns when you swipe
Above: Swyp by YouPorn
YouPorn is launching what it calls a “TikTok-inspired” web app that uses machine learning (ML) to learn and adapt to each user’s viewing preferences. TikTok is a video-focused social network operated by China’s ByteDance and became one of the top 10 most downloaded apps of the past decade.
Los Angeles-headquartered YouPorn — owned by MindGeek, which operates myriad related sites, including Pornhub — is one of the most popular pornography websites in the world and among the most visited websites overall, with a current Alexa ranking of 287. With YouPorn Swyp, the adult entertainment giant is introducing a new mobile-focused format for discovering content that has been personalized based on the user’s scrolling and swiping activity.
Users swipe up and down to view previews, and then swipe left to watch the full video. Moreover, all videos — both in preview and full-view mode — play automatically, negating the need to click on play buttons. This is very much in line with how people typically consume videos online these days — it’s all about removing friction and increasing engagement.
“Swyp was designed to be an easier, more visually enticing way to watch porn and discover new content,” said YouPorn VP Charlie Hughes.
Above: YouPorn Swyp is for the TikTok generation
Last February, YouPorn launched progressive web apps, circumventing strict app store rules to make it easier for users to enjoy a native app-like experience on mobile. Swyp also benefits from this, meaning users can easily create a home screen icon on their iOS or Android device and experience faster loading times compared to a standard mobile website.
Both YouPorn and Pornhub are no strangers to leveraging AI and ML techniques. Back in 2018, YouPorn launched personalized weekly video playlists powered by ML, while Pornhub uses computer vision to identify porn stars and automate the content tagging process.
“At YouPorn, we are always trying to help our users discover their ideal adult entertainment experience,” Hughes said. “That is why we leverage more categories and are the first to deliver machine learning recommendations across all areas of the site. Now with the addition of YouPorn Swyp, it is easier than ever for users to explore content based on their specific interests, thus always improving the recommendations within Swyp itself and everywhere on our site.”