《ABI Research：2024人工智能對數字娛樂行業的變革性影響報分析報告（英文版）（16頁）.pdf》由會員分享，可在線閱讀，更多相關《ABI Research：2024人工智能對數字娛樂行業的變革性影響報分析報告（英文版）（16頁）.pdf（16頁珍藏版）》請在三個皮匠報告上搜索。

1、TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAININTRODUCTIONArtificial Intelligence(AI)is widely deployed across the digital entertainment value chain.But rapid innovation in Generative AI(GenAI)models and expanding AI accessibility will create transformati

2、ve commercial and technology value chain opportunities.Content creators,distributors,and consumption enablers will see new AI-driven use cases that target experience,personalization,network optimization,and digital monetization.This whitepaper explores how rapid AI innovation will have a transformat

3、ive impact on the digital entertainment industry over the next 10 years.THE STATE OF ARTIFICIAL INTELLIGENCEOVERVIEWThe AI market continues to rapidly expand,driven by the emergence of public-facing applications like ChatGPT built on giant GenAI foundation models.These models have increased the awar

4、eness and accessibility of GenAI by the wider public,as they are versatile and able to support a broad spectrum of applications across verticals.But GenAI is just one part of the ecosystem.Current AI applications are made up of three dominating fields:Natural Language Processing(NLP),Computer Vision

5、(CV),and Automatic Speech Recognition(ASR).Each of these fields is explained in Figure 1.CONTENTSINTRODUCTION.1THE STATE OF ARTIFICIAL INTELLIGENCE.1OVERVIEW.1KEY AI TRENDS.3TRANSFORMING DIGITAL ENTERTAINMENT WITH EMERGING AI-SUPPORTED TECHNOLOGIES.5ADVANCED CONTENT CREATION TOOLS FOR ALL.5NEW ENTER

6、TAINMENT EXPERIENCES.6DEEPER ALIGNMENT WITH CUSTOMERS AND HYPER-PERSONALIZATION.7CONTENT DELIVERY NETWORK OPTIMIZATION.8MONETIZATION OF DIGITAL ASSETS.8HOW CAN THE DIGITAL ENTERTAINMENT VALUE CHAIN MAKE THE MOST OUT OF AI?.9CONTENT CREATORS.9CONSUMPTION ENABLERS.12STRATEGIC RECOMMENDATIONS.13CONCLUS

7、ION.14TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINReece Hayden,Senior Analyst Eric Abbruzzese,Research Director Malik Saadi,Vice President,Strategic TechnologiesTRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINF

8、igure 1:AI Field Breakdown (Source:ABI Research)Each of these AI fields has predictive and GenAI models,which will transform the digital entertainment industry over the next 10 years.Generative AIGenAI models are not new,but since 2022,they have become widely available.GenAI models use Large Languag

9、e Models(LLMs)and transformer models trained on billions or even trillions of data points to develop parameters(weights and biases)within their neural network structure,enabling them to generate content in response to user prompts.GenAI innovation has,thus far,focused on developing larger foundation

10、 models to improve probabilistic convergence,enhancing accuracy and performance.Models are now incorporating multi-modality that can respond to prompts and produce outputs across various modalities like audio,text,video,and images,enhancing GenAIs applicability for digital entertainment applications

11、.This innovation is being complemented with new diffusion architectures(applied in models like DALL-E,which can generate images with state-of-the-art quality).Although GenAI is rapidly evolving,it still faces significant challenges around Intellectual Property(IP),data availability,performance,bias,

12、privacy,computational complexity,and consistency in quality.Figure 2:Three Key GenAI Innovation Trends (Source:ABI Research)1 2021 ABI ResearchFigure 1 AI frameworksNatural Language Processing(NLP):Enables machines to understand,interpret,and generate human language,e.g.,text,social media language,t

13、rend analysis.Computer Vision:Enables computers to interpret,understand,and analyze images and videos,e.g.,gesture tracking.Automatic Speech Recognition(ASR):Processes,understands,and converts human speech into text,e.g.,mobile digital assistants.Open ecosystem:Closed source and pre-trained models l

14、ike GPT-3.5/4 are still widely used but market concerns around customization,accessibility,barriers to entry,transparency,and speed of innovation are leading to open ecosystem innovation.Innovation goes beyond models(e.g.,LlaMA)to data sets and tools(AI Model Efficiency Toolkit).This trend is being

15、spearheaded by Meta with leading models and projects like PyTorch.Model compression:Balancing price and performance is governing investment in generative AI.Compression techniques like quantization,distillation,and sparsity induction aim to sustain performance,while maximizing processing efficiency.

16、Parameter pruning is another technique that is being utilized to extract specific behavioral parameters from giant models to create efficient smaller models.Retrieval Augmented Generation(RAG):Enables models to search and retrieve data from adjacent datasets.This improves the performance and accurac

17、y of“smaller”generative models with small parametric memories.Leveraging RAG can enable price-to-performance efficient usage of models and enable transparency through source attribution.Focused on optimizing performance,privacy,and costs at scaleTRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON

18、THE DIGITAL ENTERTAINMENT VALUE CHAINBut beyond technological innovation,enabling a GenAI-driven future will require a functional stakeholder community to overcome challenges such as data availability and IP.Subsequently,generative AI success in digital entertainment will hinge on effective business

19、 relationships as much as it does on technological innovation.Predictive AIPredictive AI models are deeply integrated within digital entertainment,particularly in recommendation use cases.These models analyze historical data to identify patterns,relationships,and correlations within the data that ca

20、n then be used to predict future outcomes.Recent innovations have focused on making predictive models more intuitive and transparent with NLP and GenAI integration.This enables AI to make and then explain predictions.As the market progresses,further innovation will focus on producing more effective

21、and timely predictions,which are explored in Figure 3.Figure 3:Predictive AI Innovation (Source:ABI Research)These advances will have widespread implications for the digital entertainment value chain.Most evidently,it will improve the efficacy of advertising with greater targeting and applicability,

22、while enabling creators to produce content that is relevant to the evolving interests of an audience within the context of emerging trends.KEY AI TRENDSThe following section explores key AI trends.The market is moving toward more accessibility,scalability,and privacy-focused solutions,which will acc

23、elerate deployments in the digital entertainment market.Open Approach Accelerates Ecosystem InnovationSignificant investment and innovation are focused on the open ecosystem,which will lower development costs and accelerate innovation across the GenAI value chain.Numerous areas of AI,especially GenA

24、I,are benefiting from an increasingly open approach:Open Foundation Models:Open foundation models like Metas LlaMA 2 and Mistral 7B,which offer similar performance to closed-source leaders like OpenAIs GPT-4,are giving more developers inexpensive access to GenAI,enabling a new wave of applications a

25、nd tools to be created.3 2021 ABI ResearchFigure 3 Predictive AI innovationTime Series AnalysisML has utilized static data for model development,however,increasingly training is being supported with time series data which helps models take account of changing tastes,new emerging trends.Real-time Upd

26、atesContinuous learning allow models to incrementally learn,and update model parameters based on new data without retraining.Enables models to adapt and change to new patterns ensuring insights align with incoming data flows.Ensemble LearningUtilizing multiple models or algorithms to obtain more acc

27、urate predictions.By overlaying/merging diverse model forecasts,this technique can help overcome errors/bias with collective intelligence.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAIN Datasets:Legitimate data accessibility remains a key barrier to GenAI

28、application development,especially in the digital entertainment value chain where IP remains a strongly contested topic(as exemplified by The New York Times lawsuit against OpenAI and Microsoft).Open datasets have emerged to support training and accelerate standardization of Research and Development

29、(R&D).These can bridge the data gap that may hinder AI adoption in digital entertainment.Platforms like Hugging Face enable sharing of more than 80,000 datasets.Open-Source Project Development Tools:These aim to improve developer access to code libraries and Machine Learning(ML)tools to support the

30、generation of creative use cases.Examples include Magenta(a collection of music creativity tools built on open-source models);openFrameworks(a toolkit and library for creative experimentation);three.js(coding library and Application Programming Interface(API)for Three-Dimensional(3D)computer graphic

31、s);and p5.js(open-source coding library for creativity).Lowering Application Development BarriersEnabling AI use cases within digital entertainment requires developing an ecosystem of applications targeting specific pain points.Figure 4 explores how the market continues to decrease the barriers to A

32、I application deployment.Figure 4:Tools to Lower Barriers to Developing AI Applications(Source:ABI Research)As the market quickly moves toward prompt-developed applications that can be generated through natural language prompts,more digital entertainment process experts within the digital entertainm

33、ent value chain will be able to quickly and cheaply deploy innovative and functional applications that target specific pain pointseven without programming skills.These tools are quickly maturing;for example,GPT Builder enables anyone to create a domain-targeted GPT-based application.Although,this re

34、mains basic,over the medium term,these low/no-code development tools will start to integrate multi-modality,enabling the development of more complex applications for the digital entertainment industry.Inferencing Moves to the DeviceSoftware Developer Kits(SDKs)Application Programmable Interface(APIs

35、)Low or Near Zero Code ToolingToolkit and software framework to build applications optimized to specific hardware platforms and operating systems.Set of defined rules that enable applications to communicate with each other and cause an action e.g.,ChatGPT is an API for underlying GPT model.Develop A

36、I applications without extensive coding knowledge.Prompt Applications Build fine-tuned AI models and applications with natural language prompts.Commitment to an open ecosystem continues to emerge as vendors aim to lower developmental barriers and unlock market innovation.However,challenges remain fo

37、r stakeholders with limited ML expertise.They must prioritize developing expertise to maximize the commercial value of open-source AI tools.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINThe first wave of GenAI deployment has leveraged the cloud for traini

38、ng and inferencing.As the market looks to scale,networking costs and limitations will create a bottleneck.Increasingly,the market is pivoting toward on-device AI and deploying hardware capable of running inferencing locally for application-specific,compressed LLMs.This not only reduces network bandw

39、idth use,but also enables the development of personal AI models.These new models will help support the development of customer identities or twins that can be used to support content hyper personalization.Federated Learning Frameworks That Aim to Mitigate Data Privacy RiskStakeholder concerns over m

40、isuse or leakage of data is a primary challenge within AI deployment.Increasingly,customers are concerned about how their IP is being used to train or inform AI models.Federated learning is a key framework that is becoming more accessible and increasingly used to improve data privacy by enabling loc

41、al training of AI models without centrally pooling data.This framework has enormous benefits,especially for the digital entertainment industry:Customer Data Anonymization:Used to preserve confidential information prior to model training.This ensures that customer-specific or personal information is

42、not leaked into other AI models during the training process.DifferentialPrivacy:Ensuring data privacy,while sharing information across various domains.This can be achieved by the including noise or randomness in model training.Homomorphic Encryption:Training computation can be performed on encrypted

43、 data without decrypting.This ensures data remain confidential during decentralized training,limiting the risk of data exposure.Many of these key AI trends are still emerging within the digital entertainment industry;however,as the technology matures,expect to see significant growth in adoption with

44、in the next 1 to 2 years.The earliest deployers have been consumption enablers that are beginning to move inferencing to the device level,while they have already utilized training frameworks like federated learning to reduce consumer AI data privacy concerns.TRANSFORMING DIGITAL ENTERTAINMENT WITH E

45、MERGING AI-SUPPORTED TECHNOLOGIESOver the next 10 years,ABI Research expects AI-augmented technologies to revolutionize the digital entertainment industry.The following section explores key AI-driven technology changes that will emerge across the short(1 to 2 years),medium(3 to 6 years),and long(7 t

46、o 10 years)terms.ADVANCED CONTENT CREATION TOOLS FOR ALLUnderpinning AI-led advancements in digital entertainment is the increasing accessibility of models,tools,and datasets for anyone to use.The emergence and accessibility of large foundation models is one such advancement that has triggered innov

47、ation within GenAI with applications being developed and deployed across use cases.This is extending to multi-modality Local,on-device AI processing can enable lower latency inferencing with significant applicability for content creators looking to build more immersive experiences across film/TV,gam

48、ing,advertising,and sports.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINwith Large Vision Models(LVMs)and diffusion models that can be prompted and produce outputs in audio,text,image,and video.In parallel,prompt engineering“for all”is having a profound

49、effect on AI,as usage enables anyone to optimize AI deployments without ML expertise.Prompt engineering is designing specific queries to optimize and improve output accuracy from AI models.Easier access to prompt engineering techniques will enable low-skill users to enhance content creation to devel

50、op richer output and improve accuracy.The next wave of innovation will be created by application builders like OpenAIs GPT Builder,NVIDIA GenAI Foundry,and Google Clouds GenAI App Builder.These will enable anyone with simple natural language or voice prompts to build AI applications to perform speci

51、fic tasks across multiple domains,including multimedia content creation,gaming,or simple content scripting.For example,script writers could build an application to automate idea creation or to translate ideas into simple immersive storyboards.These short-term applications will retain significant hum

52、an supervision with a focus on putting AI in human hands,while the medium and long run will be characterized by a shift to closed-or semi-closed-loop automation enabled by autonomous AI agents.These agents can make rational,intelligent choices based on incoming data from end users without human supe

53、rvision.They can plan,track,and perform tasks without prompting.Their value lies in the ability to intelligently automate mundane or repetitive tasks,enabling humans to focus on creativity or complex tasks.ABI Research expects AI agents to be used across various facets of digital entertainment.For e

54、xample,AI agents can autonomously develop and distribute targeted,contextualized social media ads in real time based on trends,target audience,mood,and attributes without human supervision.Taking this a step further,the emerging concept of Q-learning is likely to play a part,leveraging reinforcement

55、 learning to enable foundational models to“self-learn”and correct their own actions based on human-like reasoning.This will help move the market toward closed-loop automation through which AI models can perform end-to-end tasks and continue to learn based on the outcome,similar to how humans perform

56、 tasks now.NEW ENTERTAINMENT EXPERIENCESExtended Reality(XR)that encompasses Virtual Reality(VR)and Augmented Reality(AR)is increasingly playing a role in digital entertainment.AI already plays a significant role,as machine vision models are used for visual tracking(i.e.,head,hand,and spatial tracki

57、ng).They are necessary for capturing movement and enhancing visual accuracy,but AI opportunities extend far beyond this.The short-term focus is on enabling XR to sense,understand,and interpret the world like humanswith spatial mapping and object recognition.Live translation will also be enabled with

58、 NLP and GenAI model integration,supporting human-to-human interactions and enhancing content delivery(e.g.,for TV and films).There will be increased usage of local 3D content generation with Deep Learning(DL)techniques like Neural Radiance Fields(NeRF).These models,deployed natively on XR,will enab

59、le consumers to create new,hyper-personalized digital environments or assets in real time,responding to various external stimuli.In the medium term,expect greater support and improvements in 3D volumetric video generation,enabling wider use of holograms for more immersive experiences.Innovation will

60、 focus on improving realism,enabling techniques to create greater clarity in images and videos.Multi-modal generative tools are quickly removing AI development barriers to enable access for all stakeholders across the value chain.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTE

61、RTAINMENT VALUE CHAINThis will help spatial computing gain traction.Spatial computing seamlessly converges physical and virtual objects,allowing for a more immersive/natural interaction between users and digital assets.Spatial computing will use CV(for gesture,body language,and object detection),Gen

62、AI(for analysis and content generation),and ASR(for voice recognition).In the long term,XR and spatial computing will benefit from interactive AI,which will enable end users to prompt content generation through natural language or gestures,creating a more intuitive way to communicate with AI.It will

63、 also significantly impact content consumption experiences:pervasive screens(being able to view content anywhere);environment interactivity(e.g.,consumers wearing smart glasses could look at a film ad and ask to see a trailer);and digital items embedded in the physical world.Each iteration will enab

64、le content creators to develop new and more immersive experiences.Democratization of 3D digital asset generation will also significantly impact the development and feasibility of the metaverse.GenAI models will enable developers to build and populate a metaverse quickly at a much lower cost.This wil

65、l also support in-metaverse game development with enhanced Non-Player Character(NPC)humanization,game dynamism,realism,and customized/immersive in-game experiences.The metaverse will also benefit from human scanning,enabling personalized avatar development.Enabling these new AI-driven experiences wi

66、thin spatial computing and the metaverse will require seamless multi-device capabilities(e.g.,spatial capture on mobile devices).The short term will see innovation focused on making cross-ecosystem and cross-operating system experiences possible.It will also aim to support seamless handoffs to ensur

67、e consistent and continuous user experiences across devices.This will transform over the medium term into multi-device experience sharing,allowing different users to engage and interact with digital platforms or universes in real time across devices,enabling deeper,community-focused experiences.DEEP

68、ER ALIGNMENT WITH CUSTOMERS AND HYPER-PERSONALIZATIONThe next 10 years will see huge strides in customer analytics and hyper-personalization beyond narrow predictive models used today.Figure 5 provides ABI Researchs long-term expectations for AI-driven personalization.Figure 5:Moving from Personaliz

69、ation to Hyper-Personalization(Source:ABI Research)Hyper-personalization will use“digital identities”developed by integrating all customer data to Now1-3 yrs4-10 yrs10+yrs3-5 yrs4-5 yrsPast choices&genres/categoriesNatural language Real-time mood/sentimentDaily experiencesContext awareAll data from

70、user daily life influences recommendations(actions,communications,audio inputs,visual indicators).Data is extracted from multiple devices.Utilize sensors(microphone or face scanning)to determine user mood/sentiment at viewing time influences recommendation.Determines content preferences by looking a

71、t inputs across the entire day.Bringing in additional environmental information from XR and spatial computing.Self-sovereignidentityUser has control over their personal data and experiences which then informs content personalization.AI manages data availability for platforms.Leverages natural langua

72、ge prompts and GenAI to provide more accurate recommendations without relying on metadata or tags.Based on historic choices and metadata,provides basic trend recommendations.Figure 9 Transformation of personalization from now to futureAI will play a key role in shaping the success of spatial computi

73、ng as it supports more inuitive and natural interactions between humans and digital environments.Multi-device will be a central tenet of new digital entertainment experiences,as consumers want seamless handoffs and consistent multi-user,common experiences within emerging platforms like the metaverse

74、.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINprovide a better understanding of preferences.In the short term,they will be siloed within different digital platforms;however,“common digital identities”will increasingly be aggregated on devices based on al

75、l digital interactions.In the long term,these will consider other inputs from interactions and actions in the physical world,including the ability to sense mood changes.Voice assistants will benefit from digital identity development.Initially,personalization will be limited to personality trait cust

76、omization informed by digital actions and behaviors captured through direct interactions with platforms.This will extend to voice assistants that will be hyper-personalized with the ability to sense and respond to mood changes,as well as personalizing decisions based on the users entire digital iden

77、tity.CONTENT DELIVERY NETWORK OPTIMIZATIONWith new AI-enabled technologies like the metaverse,spatial computing,and complex customer entertainment with advanced 3D visual content,upstream and downstream data flows will multiply,placing a greater strain on network bandwidth.Solving this emerging prob

78、lem will require AI-enabled networking.These use cases are explored in Table 1.Table 1:AI-Enabled Networking Use Cases (Source:ABI Research)Short TermMedium TermLong TermAI is used to manage AI compute from cloud to edge considering performance,price,and resource availability.Intelligent network han

79、doff is used to optimize content delivery.Content Aware Encoding(CAE)deployed to assess incoming video streams,determine low interest pixels,lower their quality,and decrease bandwidth usage.AI-everywhere with intelligent handling of workloads across devices,the edge,and multi-cloud.AI manages connec

80、tion points to streamline data delivery and lower latency in distributed environments.AI managing federated Content Delivery Networks(CDNs)to share resources and enable infrastructure monetization by selling excess capacity.Full autonomous networking with AI integrated to optimize content delivery.M

81、ONETIZATION OF DIGITAL ASSETSAI-technological advancements will bring additional opportunities for monetizing digital content.Figure 6 highlights eight key monetization opportunities.Figure 6:AI-Supported Monetization of Digital Content (Source:ABI Research)Market for Personalized Digital ContentUse

82、r IdentificationContent Tracking,Verification,Sourcing&AttributionMake advertising and content more immersive with 3D graphics and virtual changing rooms.Generative models will increasingly be able to tailor digital content to the end user.For example,AI models that are trained on family member char

83、acteristics.ASR and ACR will be able to sense user identity to avoid problems in OTT/pay tv around password sharing.AI can analyze images,audio,text,and videos to compare with datasets to identify similarities and flag copyright infringement.This can also be applied to sourcing and attribution.Makin

84、g Entertainment More ShoppableAI-Generated Content Models/DataHigher Number of Digital TouchpointsCosts automatically fluctuate in real time to consider traffic or to consider who is the end user.AI agents understand how much members of a household are willing to pay for the same service.Like skins

85、in gaming,customers can buy additional models/datasets to create new personalized content.Stakeholders can increase the number of touchpoints with customers across different areas.Interoperability between platforms is key to exchange user data.Congestion or Congestion PricingNew Experience UpgradesC

86、ustomers can pay to upgrade content experiences(e.g.,3D or holographic generation models.)Comprehensive digital identities will eventually encompass all of the users physical and digital interactions.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINHOW CAN T

87、HE DIGITAL ENTERTAINMENT VALUE CHAIN MAKE THE MOST OUT OF AI?The AI-led technology transformation explored here highlights content creators,distributors,and consumption enablers,along with new digital entertainment use cases and commercial opportunities.ABI Research breaks down these opportunities a

88、nd answers this question:how can stakeholders avoid pitfalls and make the most out of AI-led technology transformation?CONTENT CREATORSDemocratized access to multi-modal GenAI models is having and will continue to have a significant impact on content creators.The core long-term impact will be the si

89、gnificant decline in cost and time for content creation.This has been steadily declining for years,but AI-generated 3D content and immersive story-boarding will accelerate this process.Figure 7 explores how AI accessibility will impact and create value for content creators over the next 10 years.Fig

90、ure 7:AI Opportunities for Content Creation (Source:ABI Research)AI accessibility will create enormous value for content creators in the short,medium,and long term with hundreds of valuable use cases emerging across different verticals,as explored in Table 2.12 2021 ABI ResearchFigure 12 how will co

91、ntent creators see value over time Value createdTimeNow2yrs4yrs6yrs8yrs10yrs+Creativity Augmentation ToolsVoice dubbing,virtual scouting,3D storyboarding,automatic dialogue replacementReal-time AI ApplicationsDirectors using Augmented Reality(AR)to generate storyboard.Virtual Production Processinge.

92、g.,AI generated actorsSemi-Autonomous WorkflowsNear closed-loop content creationOn-demand Hyper PersonalizationAdvertisements personalized to customer sentiment/experienceAutonomous AI WorkflowsAdvertisements created,produced,and released autonomously with AI agentTRANSFORMATIONAL IMPACT OF ARTIFICI

93、AL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAIN Table 2:AI Opportunities for Content Creators (Source:ABI Research)GamingAdvertisingSportsFilm/TV/PodcastsQuick WinsQuality Assurance(QA)testing.Non-player character humanization.Digital asset development and game environment creation.Wider ac

94、cessibility of content creation to non-creatives(e.g.,marketers)with AI and pre-approved templates to remain on-brand.Generate new artwork in 2D and 3D.Generate storyboards.Virtual scouting.Personalize mass market campaigns to specific regions and audiences.Enable new XR experiences with shoppable c

95、ontent.Immersive and customizable experiences.Real-time and customizable on-demand statistics.Real-time language translation.Script generation.AI-enabled lighting changes,weather condition changes.Automatic dialogue replace.2D to 3D generation storyboarding.Automated closed captions.AI-automated med

96、ia campaigns.Automated podcast development(voicing,editing,and scripting).Medium-Term ValueUsers customization of digital content.Avatar scanning and integration.Real-time,hyper-personalized adaptive story lines with auto open-world generation.Create more complex and immersive worlds.Transform ads f

97、rom 2D to 3D in XR.Virtual,customizable,and interactive storyboard.Custom ads development.Immersive ads across different devices/digital environments.AI-generated assets,actors,and sets.AI-enabled ad directors.Generate new immersive ad experiences with spatial computing.In-game hyper-personalized ad

98、s targeted to specific viewers.AI-enabled content director.Increased frequency of ads with predictive models.Virtual,interactive storyboard.AI director.AI-generated actors.User-prompted adaptive content.Holographic set design.Long-Term OpportunitiesFull automation of game creation and testing cycle.

99、Deeper customization and monetization of digital content.Integration of biometrics and avatars into game experience.Personalized game world generation.Adaptive gaming experiences that change with mood/sentiment/daily life.Interactable content with story changing based on users inputs.Fully autonomou

100、s reactive ad creation,management,and monitoring with AI agents.Hyper-personalization of ads based on digital identity.Holographic sport viewing with spatial computing.User-requested content generation through natural language prompting or mood tracking,e.g.,create a TV show based on user preference

101、(genre,mood,themes,actors,or time period).Virtual actors/characters,set pieces,etc.made available on a per-use basis for user-generated content.Interactable content with story changing based on users inputs.CONTENT DISTRIBUTORSBasic AI algorithms are already being used by content distributors.But ac

102、cessibility to more complex GenAI models will create huge opportunities to discover insights,engage more effectively with customers,and develop new products/services.userid:145584,docid:525875,date:2024-09-26,TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAIN

103、OTT and Pay TV Over-the-Top(OTT)and pay TV will immediately see opportunities in the short term to optimize networking and content delivery with AI video transcoding,AI upscaling,network simulations with digital twins,and enhanced streaming through automated network load balancing and intelligent ha

104、ndoff.Moving forward,AI will be used to manage and optimize(quality of experience,traffic costs/contracts,etc.)content distribution and storage from origin servers to the cloud and edge,including multi and federated CDNs.This process will increasingly become automated from end-to-end in the long run

105、.AI agents will be able to flexibly move content to optimal locations based on dynamic commercial and technical criteria.OTT and pay TV providers will also see significant long-term opportunities in video encoding to lower networking burdens.These tools will initially compress individual scenes to r

106、educe the file size,while maintaining consistent quality.However,in the medium term,CAE will leverage AI algorithms to determine less important pixels and optimize compression based on human perception,further reducing file sizes.Beyond network optimization,AI presents opportunities to augment OTT a

107、nd pay TV services.One big opportunity over the next 10 years is to transform content search functions through hyper-personalization:Quick Wins:Leverage AI techniques to automate indexing;integrate NLP and GenAI models into search functionality to enable deeper content searches without using metadat

108、a labels.Medium-Term Value:Deploy multi-modality functionality to enable users to search based on sound,images,and video to receive similar content.Long-Term Opportunity:Search informed by user inputs and digital identity to ensure content result is hyper-personalized to one or more viewers.Hyper-pe

109、rsonalization will extend to recommendations.Recommendations have become a cornerstone of OTT and pay TV;increasingly,distributors will improve them by deploying GenAI.Generative models do not compare or rely on metadata and viewing histories,but can analyze content more effectively and align the co

110、ntent with the customers comprehensive digital identity gathered from interactions across platforms.Eventually,this will extend to recommendations based on perceived customer mood and sentiment determined through device sensors,virtual assistants,and other incoming data flows.Content protection will

111、 be another significant AI use case for OTT and pay TV.Shared memberships and content piracy are significant revenue drains on OTT and pay TV,as they enable consumers to circumvent“pay walls.”AI will help support content tracking and sourcing by comparing new content against existing databases,and e

112、ventually,machine vision models may be used to track and verify the identity of end users to ensure that they match the subscription holder.Social Media and Marketing Integrating GenAI with social media is already underway with new features being released that support content generation(e.g.,AI-powe

113、red LinkedIn post generation).Short-term opportunities will focus on new features that include automation and enhanced User-Generated GenAI will go beyond meta data and labels to enhance content search functions and personalization.AI deployments can help content distributors overcome some membershi

114、p leakage by tracking users and content,and eventually using sensors to analyze who is viewing content.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINContent(UGC).GenAI models will also offer support for regulatory tracking that extends beyond key words an

115、d understands post sentiment.This will heighten accuracy,especially when performed autonomously using AI agents.This can be extended to brand monitoring for stakeholders looking to gain a dynamic understanding of how consumers perceive their brand.Social media and marketing will also benefit from AI

116、-driven hyper-personalization.AI marketing agents will be able to track user activity across devices and leverage digital identities to contextualize ads and improve Return on Investment(ROI).A key will be user opt-in,which will be required across many regions to enable granular tracking.Virtual ass

117、istants could offer a solution by providing a more efficient way to opt-in for ads within certain brands or platforms.The long-term trajectory toward spatial computing will offer significant commercial opportunities.Marketers will be able to align ads more effectively based on both a consumer and lo

118、cation.They will also be able to create new immersive and interactive ad experiences with 3D generation and even holographic content,with XR offering a new way in which to deliver content.CONSUMPTION ENABLERSAI-led technology innovation will significantly impact the consumer device market.As complex

119、 AI models(multi-modal GenAI)become more accessible and strides are made in the accessibility of spatial computing,consumer devices will play a vital role in supporting these complex experiences.This will create both software and hardware challenges as explored in Figure 8.Figure 8:What Impact Will

120、AI in Digital Entertainment Have on End Devices?(Source:ABI Research)Although AI innovation may create challenges,it will also open opportunities for consumer devices to move beyond simply hosting entertainment to a more active role;enabling new experiences and optimizing content delivery as explore

121、d in Figure 9.As AI integration may increase device prices,it is critical that consumption enablers build new experiences and applications to build a stronger value proposition to justify the investment.Larger processing&memory demandsIncreased power consumptionDemand for higher resolution screens a

122、cross long-tail of devices Support for 3D content deliveryRequired audio and visual sensors for human trackingDigital identity aggregationPersonal,local AI modelsEnhanced data securityMulti-device continuous experiencesAlways on digital assistantsHardwareSoftwareEcosystem interconnectivityTRANSFORMA

123、TIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINFigure 9:How Will AI-Enabled Devices Create Value?(Source:ABI Research)STRATEGIC RECOMMENDATIONSThe next 10 years will offer plenty of AI-led opportunities for stakeholders within the digital entertainment value chain.T

124、his transformation will require a comprehensive strategy to prepare solid foundations and mitigate risks.Figure 10 explores steps that stakeholders should take to make the most out of AI-led sector transformation.Figure 10:Key Strategic Steps to Build a Strong Foundation for AI Deployment (Source:AB

125、I Research)Local content enhancement Multi-device,multi-user continuous immersive experiencesMulti-sensory experiencesEnableOptimizeMood aware generally intelligent assistantsLocal hyper-personalized content generationAI processing efficiencyCustomer data collection Digital identity aggregationPerso

126、nal AI model developmentCustomer device experiencesFigure 15Understand OpportunityPrepare FoundationManage TransformationIdentify quick wins and long-term opportunities.Build roadmap to align AI innovation with use cases.Evaluate return-on-investment(ROI).Explore risks around regulation and conduct

127、legal due diligence.Build policy to navigate deployments.Educate and get buy in from internal(c-suite)and external(customers)stakeholders around AI vision.Support upskilling of employees.Evaluate and transform data policy to align with AI requirements.Choose partnerships to align with AI roadmap.Mon

128、itor Proofs of Concepts(PoCs)and evaluate ROI.Roll out solutions to customers.Ensure continuous monitoring.Maintain communication with customers and deploy feedback loop.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINBut even with strong foundations,signif

129、icant risks and ethical concerns must be addressed:Autonomous AI:Autonomous AI agents will help streamline content creation and distribution with clear time/cost advantages.However,hallucination(whereby AI models make incorrect predictions,create false positives/negatives)and data bias will bring ac

130、curacy and perceptional risk.Stakeholders must,at least for now,maintain human touch points and oversight to manage risk and reward of AI deployments.Bias:Data are the most important variables in AI deployment,especially in digital entertainment use cases.Inaccurate,narrow,or biased data will create

131、 content that perpetuates stereotypes.It is important to ensure that AI models are exposed to wide and diverse datasets within training.Job Destruction:Although AI will mostly augment,rather than replace employees,stakeholders cannot ignore that AI will displace some legacy job roles.Stakeholders sh

132、ould embrace the need to evolve with AI and implement strategies to upskill employees and prepare them for the AI era.Content creators will benefit enormously from AI,as it will augment,rather than replace them,as humans will still inject creativity and operate AI tools.Intellectual Property Rights(

133、IPR):Most generative models are not yet capable of generating new creative content,but they are proficient at reframing existing ideas to fit prompts.Therefore,any content generated will be based on underlying datasets that models have been exposed to,which could create IPR problems.This will extend

134、 to identity ownership,e.g.,using celebrity visual or voice AI doppelgangers for ad development.Many still question if or how identity ownership restrictions can be enforced.Stakeholders must be careful to understand these challenges,implement clear internal governance to mitigate risk,and look to i

135、nfluence emerging global or regional IPR AI regulation.Given the speed of development around copyrights,the market should expect a lack of direction around dataset access and ownership of generated assets,especially over the next 10 years.Customer Buy-in:Enabling advanced AI customer-facing use case

136、s will rely on increased collection and analysis of customer data.Given the data privacy concerns,stakeholders must look to build trust with customers through clear data regulations and transparent AI development processes.In addition,carefully explaining the proposition of new AI-enabled experience

137、s will help display the value of increased data analytics.This will become increasingly important as the market moves toward customer“digital identities.”Competition:Decreasing creative barriers will enable anyone without technical skills to start building.Also,content distributors will see monetiza

138、tion opportunities by moving downward and creating content in-house.Addressing emerging challengers requires effective usage of AI in-house and a clearly differentiated approach that does not target mass-market content creation.Customer trust and buy-in will be one of biggest barriers for stakeholde

139、rs trying to create value with AI,given the data privacy concerns.TRANSFORMATIONAL IMPACT OF ARTIFICIAL INTELLIGENCE ON THE DIGITAL ENTERTAINMENT VALUE CHAINCONCLUSIONAI will transform each layer in the digital entertainment value chain over the next 10 years.GenAI advancements and accessibility wil

140、l likely be the main drivers with immersive digital experiences,content creation tools,and hyper-personalization quickly becoming pervasive.Stakeholders must adapt their existing strategies,culture,and resources to reflect the new pace of innovation and to comprehend and acclimatize to significant c

141、ommercial,technical,ethical,and legal risks associated with universal AI deployment.We Empower Technology Innovation and Strategic Implementation.ABI Research is uniquely positioned at the intersection of end-market companies and technology solution providers,serving as the bridge that seamlessly co

142、nnects these two segments by driving successful technology implementations and delivering strategies that are proven to attract and retain customers.2024 ABI Research.Used by permission.Disclaimer:Permission granted to reference,reprint or reissue ABI products is expressly not an endorsement of any

143、kind for any company,product,or strategy.ABI Research is an independent producer of market analysis and insight and this ABI Research product is the result of objective research by ABI Research staff at the time of data collection.ABI Research was not compensated in any way to produce this informati

144、on and the opinions of ABI Research or its analysts on any subject are continually revised based on the most current data available.The information contained herein has been obtained from sources believed to be reliable.ABI Research disclaims all warranties,express or implied,with respect to this research,including any warranties of merchantability or fitness for a particular purpose.Published February 2024157 Columbus AvenueNew York,NY 10023Tel:+1 516-624-