Connect with us


DeepMind says reinforcement learning is ‘enough’ to reach general AI



Reinforcement learning

Elevate your enterprise data technology and strategy at Transform 2021.

In their decades-long chase to create artificial intelligence, computer scientists have designed and developed all kinds of complicated mechanisms and technologies to replicate vision, language, reasoning, motor skills, and other abilities associated with intelligent life. While these efforts have resulted in AI systems that can efficiently solve specific problems in limited environments, they fall short of developing the kind of general intelligence seen in humans and animals.

In a new paper submitted to the peer-reviewed Artificial Intelligence journal, scientists at U.K.-based AI lab DeepMind argue that intelligence and its associated abilities will emerge not from formulating and solving complicated problems but by sticking to a simple but powerful principle: reward maximization.

Titled “Reward is Enough,” the paper, which is still in pre-proof as of this writing, draws inspiration from studying the evolution of natural intelligence as well as drawing lessons from recent achievements in artificial intelligence. The authors suggest that reward maximization and trial-and-error experience are enough to develop behavior that exhibits the kind of abilities associated with intelligence. And from this, they conclude that reinforcement learning, a branch of AI that is based on reward maximization, can lead to the development of artificial general intelligence.

Two paths for AI

One common method for creating AI is to try to replicate elements of intelligent behavior in computers. For instance, our understanding of the mammal vision system has given rise to all kinds of AI systems that can categorize images, locate objects in photos, define the boundaries between objects, and more. Likewise, our understanding of language has helped in the development of various natural language processing systems, such as question answering, text generation, and machine translation.

These are all instances of narrow artificial intelligence, systems that have been designed to perform specific tasks instead of having general problem-solving abilities. Some scientists believe that assembling multiple narrow AI modules will produce higher intelligent systems. For example, you can have a software system that coordinates between separate computer vision, voice processing, NLP, and motor control modules to solve complicated problems that require a multitude of skills.

A different approach to creating AI, proposed by the DeepMind researchers, is to recreate the simple yet effective rule that has given rise to natural intelligence. “[We] consider an alternative hypothesis: that the generic objective of maximising reward is enough to drive behaviour that exhibits most if not all abilities that are studied in natural and artificial intelligence,” the researchers write.

This is basically how nature works. As far as science is concerned, there has been no top-down intelligent design in the complex organisms that we see around us. Billions of years of natural selection and random variation have filtered lifeforms for their fitness to survive and reproduce. Living beings that were better equipped to handle the challenges and situations in their environments managed to survive and reproduce. The rest were eliminated.

This simple yet efficient mechanism has led to the evolution of living beings with all kinds of skills and abilities to perceive, navigate, modify their environments, and communicate among themselves.

“The natural world faced by animals and humans, and presumably also the environments faced in the future by artificial agents, are inherently so complex that they require sophisticated abilities in order to succeed (for example, to survive) within those environments,” the researchers write. “Thus, success, as measured by maximising reward, demands a variety of abilities associated with intelligence. In such environments, any behaviour that maximises reward must necessarily exhibit those abilities. In this sense, the generic objective of reward maximization contains within it many or possibly even all the goals of intelligence.”

For example, consider a squirrel that seeks the reward of minimizing hunger. On the one hand, its sensory and motor skills help it locate and collect nuts when food is available. But a squirrel that can only find food is bound to die of hunger when food becomes scarce. This is why it also has planning skills and memory to cache the nuts and restore them in winter. And the squirrel has social skills and knowledge to ensure other animals don’t steal its nuts. If you zoom out, hunger minimization can be a subgoal of “staying alive,” which also requires skills such as detecting and hiding from dangerous animals, protecting oneself from environmental threats, and seeking better habitats with seasonal changes.

“When abilities associated with intelligence arise as solutions to a singular goal of reward maximisation, this may in fact provide a deeper understanding since it explains why such an ability arises,” the researchers write. “In contrast, when each ability is understood as the solution to its own specialised goal, the why question is side-stepped in order to focus upon what that ability does.”

Finally, the researchers argue that the “most general and scalable” way to maximize reward is through agents that learn through interaction with the environment.

Developing abilities through reward maximization

In the paper, the AI researchers provide some high-level examples of how “intelligence and associated abilities will implicitly arise in the service of maximising one of many possible reward signals, corresponding to the many pragmatic goals towards which natural or artificial intelligence may be directed.”

For example, sensory skills serve the need to survive in complicated environments. Object recognition enables animals to detect food, prey, friends, and threats, or find paths, shelters, and perches. Image segmentation enables them to tell the difference between different objects and avoid fatal mistakes such as running off a cliff or falling off a branch. Meanwhile, hearing helps detect threats where the animal can’t see or find prey when they’re camouflaged. Touch, taste, and smell also give the animal the advantage of having a richer sensory experience of the habitat and a greater chance of survival in dangerous environments.

Rewards and environments also shape innate and learned knowledge in animals. For instance, hostile habitats ruled by predator animals such as lions and cheetahs reward ruminant species that have the innate knowledge to run away from threats since birth. Meanwhile, animals are also rewarded for their power to learn specific knowledge of their habitats, such as where to find food and shelter.

The researchers also discuss the reward-powered basis of language, social intelligence, imitation, and finally, general intelligence, which they describe as “maximising a singular reward in a single, complex environment.”

Here, they draw an analogy between natural intelligence and AGI: “An animal’s stream of experience is sufficiently rich and varied that it may demand a flexible ability to achieve a vast variety of subgoals (such as foraging, fighting, or fleeing), in order to succeed in maximising its overall reward (such as hunger or reproduction). Similarly, if an artificial agent’s stream of experience is sufficiently rich, then many goals (such as battery-life or survival) may implicitly require the ability to achieve an equally wide variety of subgoals, and the maximisation of reward should therefore be enough to yield an artificial general intelligence.”

Reinforcement learning for reward maximization

Reinforcement learning is a special branch of AI algorithms that is composed of three key elements: an environment, agents, and rewards.

By performing actions, the agent changes its own state and that of the environment. Based on how much those actions affect the goal the agent must achieve, it is rewarded or penalized. In many reinforcement learning problems, the agent has no initial knowledge of the environment and starts by taking random actions. Based on the feedback it receives, the agent learns to tune its actions and develop policies that maximize its reward.

In their paper, the researchers at DeepMind suggest reinforcement learning as the main algorithm that can replicate reward maximization as seen in nature and can eventually lead to artificial general intelligence.

“If an agent can continually adjust its behaviour so as to improve its cumulative reward, then any abilities that are repeatedly demanded by its environment must ultimately be produced in the agent’s behaviour,” the researchers write, adding that, in the course of maximizing for its reward, a good reinforcement learning agent could eventually learn perception, language, social intelligence and so forth.

In the paper, the researchers provide several examples that show how reinforcement learning agents were able to learn general skills in games and robotic environments.

However, the researchers stress that some fundamental challenges remain unsolved. For instance, they say, “We do not offer any theoretical guarantee on the sample efficiency of reinforcement learning agents.” Reinforcement learning is notoriously renowned for requiring huge amounts of data. For instance, a reinforcement learning agent might need centuries worth of gameplay to master a computer game. And AI researchers still haven’t figured out how to create reinforcement learning systems that can generalize their learnings across several domains. Therefore, slight changes to the environment often require the full retraining of the model.

The researchers also acknowledge that learning mechanisms for reward maximization is an unsolved problem that remains a central question to be further studied in reinforcement learning.

Strengths and weaknesses of reward maximization

Patricia Churchland, neuroscientist, philosopher, and professor emerita at the University of California, San Diego, described the ideas in the paper as “very carefully and insightfully worked out.”

However, Churchland pointed it out to possible flaws in the paper’s discussion about social decision-making. The DeepMind researchers focus on personal gains in social interactions. Churchland, who has recently written a book on the biological origins of moral intuitions, argues that attachment and bonding is a powerful factor in social decision-making of mammals and birds, which is why animals put themselves in great danger to protect their children.

“I have tended to see bonding, and hence other-care, as an extension of the ambit of what counts as oneself—‘me-and-mine,’” Churchland said. “In that case, a small modification to the [paper’s] hypothesis to allow for reward maximization to me-and-mine would work quite nicely, I think. Of course, we social animals have degrees of attachment—super strong to offspring, very strong to mates and kin, strong to friends and acquaintances etc., and the strength of types of attachments can vary depending on environment, and also on developmental stage.”

This is not a major criticism, Churchland said, and could likely be worked into the hypothesis quite gracefully.

“I am very impressed with the degree of detail in the paper, and how carefully they consider possible weaknesses,” Churchland said. “I may be wrong, but I tend to see this as a milestone.”

Data scientist Herbert Roitblat challenged the paper’s position that simple learning mechanisms and trial-and-error experience are enough to develop the abilities associated with intelligence. Roitblat argued that the theories presented in the paper face several challenges when it comes to implementing them in real life.

“If there are no time constraints, then trial and error learning might be enough, but otherwise we have the problem of an infinite number of monkeys typing for an infinite amount of time,” Roitblat said. The infinite monkey theorem states that a monkey hitting random keys on a typewriter for an infinite amount of time may eventually type any given text.

Roitblat is the author of Algorithms are Not Enough, in which he explains why all current AI algorithms, including reinforcement learning, require careful formulation of the problem and representations created by humans.

“Once the model and its intrinsic representation are set up, optimization or reinforcement could guide its evolution, but that does not mean that reinforcement is enough,” Roitblat said.

In the same vein, Roitblat added that the paper does not make any suggestions on how the reward, actions, and other elements of reinforcement learning are defined.

“Reinforcement learning assumes that the agent has a finite set of potential actions. A reward signal and value function have been specified. In other words, the problem of general intelligence is precisely to contribute those things that reinforcement learning requires as a pre-requisite,” Roitblat said. “So, if machine learning can all be reduced to some form of optimization to maximize some evaluative measure, then it must be true that reinforcement learning is relevant, but it is not very explanatory.”

Ben Dickson is a software engineer and the founder of TechTalks. He writes about technology, business, and politics. 

This story originally appeared on Copyright 2021


VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *


Intel launches more silicon and software for 5G wireless networks



Intel launches more silicon and software for 5G wireless networks

Elevate your enterprise data technology and strategy at Transform 2021.

Intel made the case today that its silicon chips and software are accelerating 5G wireless networks at the edge, and the big chipmaker is launching new chips to further improve its position in such 5G technologies as virtual radio access networks (vRAN).

Intel VP Dan Rodriguez made the announcements in a keynote speech for the virtual Mobile World Congress event. By 2023, experts expect 75% of data will be created outside of the datacenter — at the edge in factories, hospitals, retail stores, and across cities. Developers want to converge various capabilities at the edge, such as AI, analytics, media, and networking, and Intel wants to be there with the right technology.

In a recent survey of 511 information technology decision-makers, over 78% said they believe 5G technology is crucial to keeping pace with innovation, and nearly 80% said 5G technologies will affect their businesses, Intel reported.

With this in mind, Rodriguez said Reliance Jio, Deutsche Telekom, and Dish Wireless are transforming their networks on Intel architecture. The vRAN promises cloud-like agility and automation capabilities that can help optimize the RAN performance and ultimately improve the experience for users.

Intel is also expanding its family of Agilex FPGA (field programmable gate array), or highly programmable chips. The company is adding a new FPGA with integrated cryptography acceleration that can support MACSec in 5G applications. This adds another layer of security to vRAN at the fronthaul, midhaul, and backhaul levels.

Above: Intel is unveiling new 5G wireless network tech at MWC 2021.

Image Credit: Intel

Intel also said the Intel Ethernet 800 Series family is expanding with the company’s first SyncE capable Ethernet Adapter designed for space-constrained systems on the edge and well-suited for both high-bandwidth 4G and 5G RAN, as well as time- and latency-sensitive applications in industrial, financial, energy, and other sectors.

Intel summed up the tech as its Intel Network Platform, a technology foundation that aims to reduce development complexity, accelerate time to market, and help customers and partners take advantage of features in Intel hardware — from core to access to edge. Intel says its Intel Network Platform includes system-level reference architectures, drivers, and software building blocks that enable rapid development and delivery of Intel-powered network solutions and an easier, faster path to developing and optimizing network software.

Rodriguez said nearly all commercial vRAN deployments are running on Intel technology. In the years ahead, Intel sees global vRAN base station deployments scaling from hundreds to “hundreds of thousands,” and eventually millions.

Why it matters

intel 2018 mwc 2 12

Above: Intel’s Mobile World Congress in 2018.

Intel said operators of 5G networks want a more agile, flexible infrastructure to unleash the full possibilities of 5G and edge as they address increased network demands from more connected devices. At the same time, global digitalization is creating new opportunities to use the potential of 5G, edge, artificial intelligence (AI), and cloud to reshape industries ranging from manufacturing to retail, health care, education, and more.

Decision-makers also revealed that they view edge as one of the top three use cases for 5G in the next two years. With Intel’s portfolio delivering silicon and optimized software solutions, the company can tap into an estimated $65 billion edge silicon opportunity by 2025. Intel technology is already deployed in over 35,000 end customer edge implementations.

Network deployments

Operators like Deutsche Telekom, Dish Wireless, and Reliance Jio are relying on Intel technology. Reliance Jio announced it will participate in co-innovations with Intel in 5G radio and wireless core and collaborate in areas that include AI, cloud, and edge computing, which will help with 5G deployment.

Deutsche Telekom is using Intel FlexRAN technology with accelerators in O-RAN Town, in the O-RAN network it is deploying in  Neubrandenburg, Germany — a city of 65,000 people spread out over 33 square miles. The company is relying on Intel as a technology partner to deliver high-performance RAN at scale.

Dish Wireless is relying on Intel’s contributions to the 5G ecosystem as it builds out the first cloud-native 5G network in the U.S. Its inaugural launch in Las Vegas, as well as its nationwide network, will be deployed on infrastructure powered by Intel technology in the network core, access, and edge.

Cohere is pioneering a new approach to improving spectrum utilization by leveraging capabilities in FlexRAN. It is integrating and optimizing spectrum multiplier software in the RAN intelligent controller. Cohere’s testing shows its Delay Doppler spatial multiplexing technology is improving channel estimation and delivering up to a 2 times improvement in spectrum utilization for operators. That’s what Vodafone has seen in 700Mhz testing in its labs.

And Cellnex Telecom — with support from Intel, Lenovo, and Nearby Computing — is delivering edge capabilities based on Intel Smart Edge Open. This will allow Cellnex to act faster on data, provide service-level management, improve quality of service, and deliver a more consistent experience to its end users. Deployed in Barcelona, this solution will extend to more markets using the blueprint developed with Intel and Nearby Computing.

Intel said its network business grew 20% between 2019 and 2020, from $5 billion to $6 billion. The company’s strong position is the result of early investments in hardware and software.

Intel predicted a bright future for the industry. As 5G blooms to meet its full potential alongside edge computing, experts expect artificial intelligence, the cloud, and smart cities will become the norm. Factory automation is also expected to flourish with Industry 4.0 and retail locations will redesign the shopping experience. And for consumers, cloud gaming and virtual and augmented reality over mobile networks will become an everyday experience, Rodriguez said.


VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

Continue Reading


Disney interview: Big games coming with Avatar and Pirates of the Caribbean



Disney interview: Big games coming with Avatar and Pirates of the Caribbean

Elevate your enterprise data technology and strategy at Transform 2021.

Disney had a big week at the Electronic Entertainment Expo (E3) with the announcement of Ubisoft’s new open-world game, Avatar: Frontiers of Pandora. The title has cinematic graphics that replicate the imagery of the movie and the environments of the beautiful moon of Pandora.

Microsoft’s Rare studio also announced that characters from the Pirates of the Caribbean films, like Jack Sparrow and Davy Jones, will be integrated into Sea of Thieves. Both are examples of Disney’s return to triple-A games after changes to its strategy for games over the years.

Disney had triple-A games in the past when it had its own game studios. But it closed down or sold off the studios, and more recently it has been licensing its properties to outside companies, mostly mobile game publishers such as Glu and Jam City. And now it’s clear that Disney has been licensing its properties out for triple-A games as well.

I talked with Sean Shoptaw, senior vice president of Walt Disney Games, and Luigi Priore, vice president of Disney and Pixar Games, about Disney’s presence at E3 and the latest on its strategy for games.

Here’s an edited transcript of our interview.

Above: Pandora looks beautiful as an open world.

Image Credit: Disney

GamesBeat: What’s new for Disney Games?

Sean Shoptaw: I guess that’s a pretty loaded question. There’s a lot going on. It’s been a great week at E3 with some of the announcements you’ve seen. The business is doing well. We’re super excited about the products we’ve announced, and a lot of the products we have in the market already as well. We’re very excited about the status of games at Disney.

GamesBeat: What was announced altogether this week?

Shoptaw: The Avatar title and the Sea of Thieves integration were the two big ones so far.

GamesBeat: How long has Avatar been in the making now?

Priore: That predates us, obviously, because Disney didn’t acquire the 20th Century Fox properties until a couple of years ago. That started well before the acquisition. The great thing moving forward is we’ve been lucky enough to be able to work with Ubisoft and the team at Massive with Lightstorm, James Cameron’s production company, and Jon Landau, who worked on Avatar: Frontiers of Pandora. They announced Massive was working on it a while ago, but this is the name announcement and the first glimpse of what that game is going to be. We’ve gotten very good responses.

GamesBeat: The animation almost feels like it is the movie.

Shoptaw: Yeah, after the trailer, people are finding out that it’s very cinematic. The quality is extremely high. We’re super excited about that title.

GamesBeat: How does that relate to the movie releases, the next Avatar movies? Are they slated for particular dates yet?

Priore: Yeah, the next one is holiday 2022. Sean can get into our general strategy, but on licensing games like this these days — there was a time 15 or 20 years ago where playing the movie was something. You bought the game and played the movie. Things like the classic Aladdin game on Sega Genesis. You played the film. That was popular at the time, but gamers expect more now. They want to interact with their favorite characters and worlds, but they want to play new stories and do new things with those characters and worlds.

On Avatar it’s the same thing. What James Cameron and Jon Landau created is an amazing science fiction world. Pandora is awesome. They have great heroes. It’s a great playground to play in. This is a brand new story with new characters. It’s going to become part of the canon. The whole idea is to have it be part of the storyline of that giant franchise on Pandora, but it’s not a “play the movie” game. It’s an all new open world, new characters. That’s why it’s called Frontiers of Pandora. It takes place on another frontier, another area of the moon of Pandora.

avatar 2

Above: The environs of Pandora.

Image Credit: Disney

GamesBeat: How much will we recognize it? Is it a replication of the movie world, or is it more Ubisoft’s imagining of a new part of the world?

Priore: No, we’re working directly with the filmmakers. Jon Landau is involved almost every day on this. This is the same world. It’s just that you’re going to meet new characters, new clans of Na’vi, and your role is going to be different. I don’t want to go too much into it because we didn’t announce everything yet. But it’s a whole new story with new characters on the same planet, in the same canon. Jedi: Fallen Order was a new story about a new Jedi in the Star Wars canon. It’s the same idea here.

GamesBeat: On your level, how are you involved, compared to Ubisoft’s responsibility?

Priore: Massive is the developer. They’re one of the best in class at open world games. Division, Division II, amazing games. They’re working with the FoxNext team and Lightstorm, working directly with the filmmakers. Where we come in is we’ve brought our expertise in working on IP, working on games. We’ve talked about this a bit. We have a collection of producers, game designers, artists, writers that work together with our partners to get the best out of what they want to do.

Although we just joined this game production recently, since we acquired the 20th Century properties, we’re working directly with Massive and Lightstorm to help them make the best game possible. It’s our job to make sure that Massive has everything they need and that the brand is as authentic as possible working with Lightstorm.

GamesBeat: It still feels like there are so many opportunities for Disney in games. How do you approach which ones to take on, how many of them to do on what platforms?

Shoptaw: There’s no shortage of inbound interest to work with all of our franchises, thankfully. That’s something we’re grateful for. We try to take the approach that — we need to align our partnerships around people’s passions for IP. When we sit down and meet with a developer or publisher about an idea, a lot of that is driven by their passion to go make a specific game with a specific IP. Ideally we’re matching that up with a best-in-class partner. To the point about Massive, about EA, about the partnerships you see now and will see in future, it’s about matching that passion with best-in-class partners to go make what we hope are the best games we’ve made for whatever genre or IP it might be.

That, to us, is the recipe. It’s about working with high-quality partners that have passion for Disney IP, whatever it may be. It gets to be a much easier conversation once you’re in that world, where you see that passion. They have a track record of developing high-quality products. Then it’s about figuring out exactly what the execution is going to be, working closely with Luigi and our other teams internally to map to what ultimately is the final product. But that really is, at the top, our focus, to match people’s passions and the highest quality of partner we can find to go make a certain game.

GamesBeat: There’s a lot more coming than what we’ve seen here at E3, I’m sure.

Shoptaw: As I said yesterday, our slate has never been better. We’ve never been more excited about the slate we have. Some of that’s been announced and some hasn’t. But we feel like we’ve been fortunate to do some exciting partnerships with partners that have a high bar on quality and thankfully have a passion for our IP. We look at our pipeline of product and it’s never been healthier. The quality bar has never been higher.

GamesBeat: Star Wars: Hunters is another one of those coming.

Priore: Very excited about Hunters. Both mobile and Switch, which is very exciting for us. We’ve wanted to get more content on the Switch. We’re excited about what that game represents within the Star Wars universe. We think it’s a unique take, both creatively and from a genre perspective. It’s a very differentiated experience, one we haven’t seen so far in Star Wars.

avatar 3 1

Above: Avatar: Frontiers of Pandora is coming in 2022.

Image Credit: Disney

GamesBeat: Zynga is an interesting choice there. They haven’t done a console game before. When I was talking to them about their Harry Potter game, though, I was pretty stunned by how much work went into that. Several years, the biggest team they ever had. How much they put into all the animation and everything else that keeps players immersed in that universe was very interesting. It wasn’t as much of a surprise to see them do a Star Wars game.

Priore: They came to us with a good idea, with a team that we had a lot of respect for. They have a lot of passion for Star Wars. It made a lot of sense to us as we sat down and mapped out what a game could look like here. You’ll see that passion and quality in the final product. As I said, I think it’s a unique take on Star Wars, and knock on wood, our fans will agree. We’re pretty bullish on that game, excited for the world to see it.

GamesBeat: Is there anything else announced in Star Wars?

Shoptaw: We announced the Massive title as well not too long ago. We’ll do an open world Star Wars game with Massive. Similar to Zynga, we feel like it fits a need within the Star Wars universe that hasn’t been fulfilled, and we felt Massive was a perfect partner to execute on it. We’re huge fans of David [Polfeldt] and the team. We aligned quickly on a vision and an experience for Star Wars that, again, fans and gamers will flock to, hopefully. We feel good about the team making it, and we think the idea behind it is great.

GamesBeat: I take it that it’s just not the time to show a glimpse of that?

Shoptaw: We’re still a little ways off, but at the right time I think people will see why we’re so excited about it. We had Avatar to show this time. We didn’t want to show too much at once. With Star Wars, we’ve seen such a great response to Star Wars recently. Jedi: Fallen Order continues to perform. We just hit the 20 million user milestone recently. That title was another great example of telling a truly original story within that universe, something that hadn’t been told before. Allowing people to go be a Jedi and play a fun game like that has proven to work well and continues to resonate.

We’re not looking to flood the market and put one game on top of another. We want to be disciplined and focused on the best experiences. It’s not about making as many games as we can possibly make. It’s about making the right games with the right partners. When we do that, we see that we’re able to have a good amount of success. We feel fortunate about that. We’ll continue to do things that we think fans and gamers will be excited about with the right partners in the right genres on the right platforms. If we can keep that discipline I think we’ll continue to raise the bar on quality and continue to deliver products that will meet the moment, meet the level of quality that we want.

GamesBeat: What’s the strategy around platforms, especially mobile?

avatar 4 1

Above: The Avatar game has been years in the making.

Image Credit: Disney

Shoptaw: Mobile is a huge market globally. We’re always going to have more mobile products than we have console products, just by the nature of the platform. It’s pretty simple. We want to be where it makes sense for our IP to be, across genres, across markets. That might mean local products like Twisted Wonderland in Japan, which is a very unique, specific take on Disney in a market that is hugely passionate about Disney specifically. That execution is a great example of being very locally focused, an execution we know is going to resonate with a certain market. We certainly have regional looks as well, products that make sense in certain parts of the world. Asia is a good example. And then we have a fair amount of products that are global.

We look at it through a local, regional, and global lens. We want to make sure we match franchises and IP with markets in genres that resonate most powerfully. Twisted Wonderland is an incredible example of a local execution. A lot of our titles, obviously, are global, and they’ve been massive successes across markets. We’ll continue to look at big global opportunities like Galaxy of Heroes with EA. Obviously the Marvel portfolio has had a lot of incredible success across mobile.

We’re not one size fits all. We’ll focus on the right execution in the right market with the right partner and the right genre. We don’t want to flood the market, again, with a bunch of duplicative titles, or just put our brand on any title that we get some interest in. We’re going to be disciplined, and we’re going to make sure we apply that sort of strategic thought to every game we do, regardless of market. That approach over the last few years for us has shown that it works well, and we’ll continue to have that view of the world. It needs to make sense. It needs to be really high quality.

Even if we think we’re missing something, if there’s an opportunity for a genre or a certain IP is underserved, we’re not going to rush and just do a game because we think we need to. We will wait and make the right game with the right partner. That’s as important as getting any games out there. That’s something we’re focused on as much as we are getting products to market and satisfying the demand that we fortunately have for our IP. We’ll continue to be disciplined.

GamesBeat: Did the pandemic change your thinking in any ways?

Shoptaw: No. Fortunately the game industry overall, and certainly our business within Disney, had been doing very well prior to COVID. People’s perception was that video games benefited a lot from people staying home, working from home. There’s certainly some truth to that. But video games have been growing rapidly as an industry prior to COVID. It would have continued to grow rapidly if we never had COVID. So it hasn’t changed any strategic thinking for us. Fortunately our products and releases, nothing was impacted too dramatically by COVID. Again, strategically it hasn’t changed our view of the world.

avatar 5 1

Above: The humans are the enemy in Avatar.

Image Credit: Disney

GamesBeat: It seems like the video game opportunity is a lot more clear than it used to be in the wake of the pandemic. I’ve been writing all these stories about how much more money is coming into the game industry. I think it’s $49 billion in the first five months of this year in terms of investments and acquisitions and public offerings. That compares to $33 billion for all of last year. At the same time I know the movie industry is contracting. Does it make some sense to argue the case for games as a bigger slice of the pie going forward, a bigger opportunity? Is it time to double down on video games?

Shoptaw: We look at games as that pillar, regardless of what the model is. For us we feel like playing in the space where we’re playing gives us the highest quality products that we can scale across the world. When you look at internal development, obviously that comes with a considerable amount of investment and volume to go hit the aspirations that we have in this space. Again, that’s to work and deliver the best products across the world — console, mobile, PC.

Generally there’s no shortage of investment still happening on the linear side. To your point around film, streaming has taken a considerable bite out of that traditional film apple. But the investment in linear content is still extremely material. I don’t think that’s been diminished in any way. From a games perspective, again, our focus has been, and will continue to be, on quality, on being able to scale this business and meet the demand that exists in video games.

We feel like right now, that strategy is to go license and work with the best partners in the world to deliver on that demand. We’ll continue to do that as long as we can meet that bar of quality, of volume, and making sure that our reach is where we need it to be. Again, we’re fortunate to have the IP that we do. We owe it to consumers, fans, and gamers to make sure we’re delivering at that level. That will continue to be our focus.

We’re excited about where this business is and where it’s going. We think it is a pillar, regardless of model. As long as we’re delivering products like we are, games will continue to be a foundational part of the overall entertainment medium. Certainly from a Disney perspective we do that very thoughtfully. We’ve given a lot of attention and focus to it internally. You’re seeing those results in products today, and you’ll continue to see them in the future.

GamesBeat: Can you tell me a little about the Sea of Thieves integration?

Priore: We’re excited. The team at Rare — this goes back to what we were saying about best-in-class partners. They’ve made the best pirate game ever with Sea of Thieves. We’re excited to have A Pirate’s Life, something authentic to Pirates of the Caribbean that’s also authentic to Sea of Thieves. It lines up with what Sean was saying about doing it the right way, making it authentic to what we do at Disney. Just as we said about Avatar or Star Wars, we want to do that all the time, and we feel like we’re having success with that.

I’ve been here a long time. I’ve been at Disney in games for 25 years. I’ve been on the roller coaster, and I’ve never been more excited about the opportunities we have lined up. You’re seeing some of them, whether it’s Massive and Ubisoft with Avatar or Rare and Microsoft with Sea of Thieves and Pirates of the Caribbean. We’re excited about what’s coming next.

GamesBeat: Call of Duty has an interesting funnel these days, where they start with Call of Duty Mobile. They have 500 million people that way. They have Warzone, a free-to-play console and PC game, 100 million players. That feeds into Cold War, a $60 packaged game that sold 40% better than the previous game in the series. It seems like no accident. You widen that funnel and eventually you widen the market for the franchise’s premium games. It seems like only the biggest companies can do that. I don’t know if Disney has looked at that strategy as well, where there’s a purpose to each game in that funnel.

Shoptaw: People’s strategic view of a game and that game’s purpose are going to differ greatly. If you’re developing a game like Call of Duty, that’s a significant franchise and an incredibly successful one. There’s a lot of ways you can continue to funnel users and grow that pie across platforms.

From a Disney perspective it’s obviously different. We’re working with partners to create experiences. Our strategy is, again, to bring as high a quality of product as we can to market. It’s not about platform-building. We’re not doing this vertically, building out platforms and doing things that might be the strategy of a big game developer.

For us, we’re certainly open to playing in a space that creates these multiplatform experiences that drive audiences in meaningful ways across products. It’s something we’d be happy to engage on if that kind of execution made sense for a franchise of ours. But again, our focus is generally tied to working with partners that can go elevate the IP, that can bring it to consumers in new, unique, innovative ways. If that outcome happens, to your initial question, that’s great. But it’s not core to our strategy because we’re not a developer. We don’t think about it through that lens. If they can leverage our IP in a similar way to Call of Duty, sure, we’re happy to engage on that conversation.


GamesBeat’s creed when covering the game industry is “where passion meets business.” What does this mean? We want to tell you how the news matters to you — not just as a decision-maker at a game studio, but also as a fan of games. Whether you read our articles, listen to our podcasts, or watch our videos, GamesBeat will help you learn about the industry and enjoy engaging with it.

How will you do that? Membership includes access to:

  • Newsletters, such as DeanBeat
  • The wonderful, educational, and fun speakers at our events
  • Networking opportunities
  • Special members-only interviews, chats, and “open office” events with GamesBeat staff
  • Chatting with community members, GamesBeat staff, and other guests in our Discord
  • And maybe even a fun prize or two
  • Introductions to like-minded parties

Become a member

Continue Reading


Ambarella unveils two new AI chip families for 4K security cameras



Ambarella unveils two new AI chip families for 4K security cameras

Elevate your enterprise data technology and strategy at Transform 2021.

Ambarella unveiled two new AI chip families for 4K security cameras today as it pushes further into computer vision.

The Santa Clara, California-based company is introducing its new CV5S and CV52S security chips as the latest in its system-on-chip portfolio based on the CVflow architecture. The chips use an advanced 5-nanometer manufacturing process, where the width between circuits is five-billionths of a meter.

With the combination of new designs and better miniaturization from the manufacturing process, the SoCs can support simultaneous 4K encoding and advanced AI processing in a single low-power design, which provides great AI SoC performance per watt of power consumed, said Fermi Wang, CEO of Ambarella, in an interview with VentureBeat.

“If you have a camera, you want to cover a very big space, and then you want to use a 4K camera to cover all 360 degrees,” Wang said. “You want to have a single chip to talk to the 4K sensors. You want to process all of the videos together and analyze the video.”

The CV5S family targets security camera applications that require multiple sensors for 360-degree coverage, over a wide area and with a long range, such as outdoor city environments or large buildings. Ambarella designed the CV52S family for single-sensor security cameras with advanced AI performance that need to more clearly identify individuals or objects in a scene, including faces and license plate numbers over long distances, such as ITS traffic cameras.

Camera applications

Above: Ambarella’s CV5S AI chips can power the latest security cameras.

Image Credit: Ambarella

The cameras can be used for detecting the faces of criminals in crowds or recognizing license plates. There are of course privacy concerns about that. They can also send an alert if a crowd is forming in a part of a city, and they can monitor packages left behind in stations, airports, and more. Applications also include managing traffic congestion, detecting vehicle accidents, automating speed control, locating missing or stolen vehicles, monitoring queues in retail environments, better managing retail product placement, enhancing warehouse tracking — generally providing more actionable intelligence at both the store and corporate levels. And these more invasive applications can be countered with the ability to set up privacy masks, like continuously obscuring certain portions of larger scenes, preserving privacy.

“We are generating revenue based on all of the CVflow family today,” Wang said. “If you look at all the possible applications that we can address here — our current security camera market, automotive markets — you can see that there’s a lot more opportunities. We’re talking about smart homes, smart cities, smart retail, and also in the future robotics. There are many, many applications that we can address moving forward.”

John Lorenz, senior technology and market analyst at Yole Développement, said in a statement that security system designers want higher resolution cameras, more channels, and faster AI. He said Ambarella’s new chips are competitive in the security chip market, which is expected to exceed $4 billion by 2025, with two-thirds of that being chips with AI capabilities.

The new CV5S SoC family supports multi-imager camera designs and can simultaneously process and encode four imager channels of up to 8 megapixels (MP), or 4K resolution, each at 30 frames per second, while performing advanced AI on each 4K imager. These SoCs double the encoding resolution and memory bandwidth while consuming 30% less power than Ambarella’s prior generation.

ambarella Photo Fermi Wang CEO of Ambarella

Above: Fermi Wang is CEO of Ambarella.

Image Credit: Ambarella

The new CV52S SoC family targets single-sensor security cameras and supports 4K resolution at 60fps, while providing four times the AI computer vision performance, two times the central processing unit (CPU) performance, and 50% more memory bandwidth than its predecessors. This increase in neural network (NN) performance enables more AI processing to be performed at the edge, instead of in the cloud.

“Because you do all the video analytics at the edge, the full video doesn’t need to leave the camera,” Wang said. “You only pass the data that you analyze along to the cloud.”

That’s important as you don’t want traffic from self-driving cars to clog up the wireless bandwidth connections to datacenters.

“The biggest difference approach is that we talked about in the past, we call it ‘algorithm first.’ Basically, when we do the video compression or video processing, or image processing, even and then the controller computer vision for the deep neural network or AI processor, we can see the flow, we try to determine the hardware architecture as well,” Wang said. “For the architecture, we look at what kind of algorithm we want to implement first. So then, after we go through all the areas of study, we understand how an application works, and the portion of the area that takes the most computation performance, and where we can optimize without losing the performance or accuracy of the algorithm.”

ambarella Photo Ambarella Edge AIoT Market

Above: Ambarella is making AI chips for a spectrum of devices.

Image Credit: Ambarella

He added, “And after going through all the tradeoffs, we create the architecture not only try to deliver the best performance, but also deliver the best power consumption as well. I think we are very differentiated and can compete.”

In addition to security, there are many other AI-based internet of things (IoT) applications that can take advantage of the high resolution and advanced AI processing provided by these new SoC families. For example, smart cities can leverage high edge AI performance and image resolution for improved traffic management, accident detection, and automated speed control, as well as the rapid location of missing and stolen vehicles.

“If there’s an accident where there is traffic congestion, or if you need to find a missing vehicle, then you can get enough information to monitor all of the smart city requirements,” Wang said. “You need to do real-time management.”

Likewise, smart retail operations can use this resolution and advanced AI to better manage product placement, adjust cashier staffing for real-time line management, enhance warehouse product tracking, and provide more actionable intelligence at both the store and corporate levels.

The chip families share features such as a software development kit (SDK) for the security camera market, CVflow development tools, and dual Arm A76 1.6GHz CPUs with 1MB of L3 cache memory and a two times performance gain over prior generations. It also has enhanced image signal processing with high-dynamic range, ISO low-light, dewarping, and rotation performance.

They also have on-chip privacy masking to block out a portion of the captured scene, connector interfaces, on-chip cybersecurity hardware with secure boot, OTP, and Arm TrustZone technology. They can support up to 14 cameras and a variety of memory.

The CV5S and CV52S SoC families are expected to be available for sampling in October.


ambarella Photo Ambarella Portfolio

Above: Ambarella’s chip portfolio.

Image Credit: Ambarella

Ambarella was founded in 2005. It has evolved over the years from a video processor chip design firm to a designer of computer vision chips for a variety of markets. It started with video processors for cameras and video cameras. Then it transitioned to making AI-based chips for automobiles and security cameras.

“We went through many different markets, some good, some bad,” Wang said. “We did chips for camcorders and GoPro sports cameras, DJI drone cameras, and eventually there was no innovation in these markets.”

The markets that have lasted longer include security cameras, which require ever-increasing levels of resolution and quality, and automotive cameras, which started in 2011 and continue today as cars need more video sensors and AI processing to distinguish driving hazards.

“Over time, we believes that video analytics will become very important, where you can interpret what the computer vision shows,” Wang said.

Deep learning neural networks are necessary for that work, and that has put a lot of pressure on better AI processing while at the same time delivering better efficiency with low power consumption and lower costs.

ambarella Photo Ambarella SoC Architecture

Above: Ambarella’s CVflow architecture is the basis for a lot of chip families.

Image Credit: Ambarella

“We start our CVflow family with 10-nanometer production, and today we go to five nanometers,” Wang said. “And also we build tons of different software, including a tool to convert any neural network designed by our customers.”

In the past several years, Ambarella has spent $500 million on research and development on computer vision, and it reported last year’s revenue for the segment at $25 million, Wang said. Analysts are expecting the company to hit $75 million in computer vision revenue in 2021.

“It’s not only just a product anymore,” Wang said. “It’s really a revenue generator for our customers. We proved that the investment was really important, and I’m glad we went through that.”

The company has about 40 mass-production customers now. And they are asking for better and better performance.

“If you use 8K performance to process multiple video streams at the same time, then it becomes a mainstream product,” Wang said. “In fact, I can say that in security cameras, people want to connect four to six to eight cameras to one single chip.”


VentureBeat’s mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact.

Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:

  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more

Become a member

Continue Reading