Chuhee LeeIndependent research
All publications
Cover art: one point surrounded by rings of small squares.
Horizon
2026–2031
Scope
US-led · global context
Lens
Entertainment and learning
Basis
Expert forecasts
ShiftsEntertainmentNovember 2026 edition · Research report

Media and Learning Outlook 2026–2031

How AI, wireless networks and new devices will change the way people watch, listen, play and learn by 2031, as expert forecasts see it.

Research questionHow will advances in AI, wireless communication and consumer devices change how people use media for entertainment and for learning by 2031?

By Chuhee Lee≈23 min readData as of 1 Nov 2026Next refresh: November 2027Change log

With research assistance from several AI models (Claude, Gemini, ChatGPT) · Views are the author’s own

Cite asChuhee Lee, Media and Learning Outlook 2026–2031, November 2026 edition (data as of 1 November 2026), https://blueplanet-ai.github.io/media-learning-outlook/, CC BY-NC 4.0.

The report in 60 seconds

Three takeaways · each with its source
  1. 29% vs 5% Fact

    Asking replaces browsing, more slowly than feared

    Deloitte expects about 29% of adults in developed countries to see an AI search summary every day in 2026, and Pew found users clicked a result on 8% of searches with an AI summary, against 15% without one. Measured effects are smaller than feared: AI Overviews cut search traffic to English Wikipedia by about 5%.

    Source: Deloitte [1]; Pew [2]; Khosravi and Yoganarasimhan [3]
  2. 2× vs −17% Interpretation

    Designed tutors teach; general chatbots help finish tasks

    54% of US teens have used chatbots for schoolwork. A Harvard randomized trial found an AI tutor more than doubled learning gains, but in a Turkish trial students who practiced with a general chatbot scored 17% lower on a closed-book exam. The evidence points to tutor design and teaching support, not access, deciding whether AI learning lasts.

    Source: Pew [4]; Kestin et al. [5]; Bastani et al. [6]
  3. >50% vs 1–3% Hypothesis

    Supply turns synthetic; attention stays selective

    More than half of the tracks uploaded to Deezer each day are AI-generated, yet in April they drew only 1–3% of streams. The author’s central case for 2031, “Ask to know, scroll to watch”, is that AI answers become the first stop for information and learning while entertainment stays in the feeds people already use, with named people and brands keeping most attention: not yet observed.

    Source: Deezer via TechCrunch [7] and AlternativeTo [8]

Fact: stated by the cited source. Interpretation: the author’s reading of the cited evidence. Hypothesis: expected from the evidence but not yet observed. Figures inside every takeaway are sourced.

At a glance

One-page summary

Headline findingExpert forecasts to 2031 point to a split in how people use media. For information and learning, people will increasingly ask an AI rather than look for themselves. For entertainment, AI works inside the feeds and apps they already scroll.

Summary

The short version

Expert forecasts to 2031 point to a split in how people use media. For information and learning, people will increasingly ask an AI rather than look for themselves. For entertainment, AI works inside the feeds and apps they already scroll: YouTube, Netflix, Spotify, games and Google itself. A growing share of what they get back will be made or remixed by AI. The author’s central case, “Ask to know, scroll to watch”, rests on that split. Faster networks and new devices enable it; neither is the main cause.

Asking replaces browsing, for information.

Deloitte expects about 29% of adults in developed countries to see an AI search summary every day in 2026, three times the share using a standalone AI app daily. Pew found users clicked a result on 8% of searches with an AI summary, against 15% without one. Measured effects are smaller than feared: AI Overviews cut search traffic to English Wikipedia by about 5%, while news leaders expect a 43% fall. Only 2% of US adults name an AI chatbot as a top-three news source.[1][2][3][9][10]

Learning becomes a conversation.

54% of US teens have used chatbots for schoolwork. A Harvard randomized trial found an AI tutor more than doubled learning gains against in-class active learning. But in a Turkish trial, students who practiced with a general chatbot scored 17% lower on a closed-book exam, and a tutor designed for learning only matched studying without AI.[4][5][6]

Supply turns synthetic; attention stays selective.

More than half of the tracks uploaded to Deezer each day are AI-generated, yet in April they drew only 1–3% of streams. McKinsey’s scenario puts around 60 billion dollars of annual US film and TV distribution revenue up for redistribution within five years of mass AI adoption.[7][8][11]

Media starts to talk back, mostly for the young.

72% of US teens have tried AI companions, and Americans spent about 705 million hours on companion apps in the first quarter of 2026, two and a half times their time on dating apps. Across all ages the picture is narrower: only 1.9% of ChatGPT messages are about relationships and 0.4% about games and role play.[12][13][14]

Devices and networks enable; they do not lead.

IDC forecasts 27.3 million display-less smart glasses and 12.2 million display glasses in 2030, small next to phones and TVs. Ericsson forecasts mobile traffic 2.5 times higher by 2031 and first 6G services around 2030. Google is launching audio glasses before display glasses. The bigger network change is the uplink, as AI and cameras send more data up.[15][16][17][18]

The leading indicator: how US teens already use AI chatbots

Share of teens 13–17 · Pew, fieldwork Sep–Oct 2025

Share of US teens 13–17 who have used chatbots for…
Searching for information
57%
Help with schoolwork
54%
Fun or entertainment
47%
Creating or editing images or videoabout 4 in 10
~40%
Summarizing articles, books or videosabout 4 in 10
~40%
Casual conversation
16%
Emotional support or advice
12%

“About 4 in 10” is drawn at 40%. About 3 in 10 teens use a chatbot daily.[4] The author reads teens as the leading indicator: habits formed now carry into adulthood by 2031.

Chapter 1 of 9
Baseline

Where we start in 2026

Media time is capped and already moving onto the TV set and creator platforms; AI arrives in a market that grows slowly in money and not at all in hours.

Time. Americans 15 and older watched 2.61 hours of TV a day in 2025, the largest single leisure activity in the federal time diary. Ages 15 to 19 watched 1.48 hours and spent 1.53 hours on games and computer leisure; ages 65 to 74 watched 4.20.[19][20] Hours are close to a ceiling, so every shift in this report is a shift of share.

Screens. In August 2026 streaming took 49.6% of US TV-set time and YouTube alone a record 15.8%.[21] 61% of US internet households use a smart TV as their main streaming device.[22] McKinsey’s US baseline shows linear TV viewing falling 4% a year from 2022 to 2024 while streaming grew 13% and social video 14%.[11]

Money. PwC forecasts global entertainment and media revenue of 4.2 trillion dollars in 2030, growing 3.4% a year. Advertising grows faster (5.6%) than consumer spending (2.5%), traditional TV shrinks 1.1% a year, and PwC names AI-driven personalization of advertising as a main growth engine.[23]

Share of US TV-set time, August 2026

Percent · Nielsen Gauge

Streaming, all services
49.6%
YouTube (inside streaming)record share
15.8%
Cable
18.9%
Broadcast
17.5%

YouTube is part of the streaming total. Nielsen changes its Gauge method from September 2026, so later months will not be like-for-like.[21]

Author’s read · confidence: highThe five-year question is not whether people will consume more media, but which interface, which supplier and which device will hold a roughly fixed amount of time and a slowly growing amount of money.
Chapter 2 of 9
Shift 1 · driver: AI

Asking replaces browsing

For information, news and how-to learning, AI answers become the front door; for entertainment, the evidence points to AI inside the feeds people already use.

What experts forecast. Deloitte predicts that daily use of AI inside search will be three times daily use of any standalone AI tool in 2026: about 29% of adults in developed countries will see at least one AI search summary a day, and about 10% will use a standalone AI app daily.[1] Ericsson’s 2026 consumer survey across 27 markets finds 38% of smartphone users expect to hand daily tasks to AI by 2030, against about 4% using agentic AI today, and heavy AI users rising from about 10% to about 30% of smartphone users.[24]

What it does to media. In Pew’s browsing panel, 8% of visits with an AI summary led to a click on a result, against 15% without one, and 26% of those pages ended the session.[2] Google search traffic to more than 2,500 news sites worldwide fell 33% in a year (Chartbeat data), media leaders expect a further 43% fall over three years, and 76% plan to push journalists to behave more like creators, with YouTube their top off-platform priority.[9]

Smaller and more Google-shaped than feared. A causal study using the staggered rollout of Google’s AI Overviews found they cut search traffic to English Wikipedia by 5.45% (4.82% with a different comparison), an order of magnitude below publishers’ expectations.[3] In a study of 11,500 real queries, AI Overviews appeared on 51.5% of them and drew on sources that overlapped little with the organic results, leaning toward Google-owned content such as YouTube.[25] EMARKETER forecasts almost 190 million US users of AI tools by 2030, with Google’s Gemini and AI Mode passing ChatGPT in users in 2027.[26] For news, only 2% of US adults name an AI chatbot as a top-three source and 7% rely on AI tools a great deal or a fair amount.[10]

Learning and leisure too. More than 70% of ChatGPT messages are now not about work, and 10.2% are requests for tutoring or teaching.[14] 57% of US teens have used chatbots to search for information and about 4 in 10 to summarize articles, books or videos.[4]

Clicks when an AI summary appears

Share of Google search visits with a click · 900 US adults · Pew, March 2025

No AI summary on the page
15%
AI summary on the page
8%
Link inside the AI summary
1%

Google called the study’s method flawed and its queries unrepresentative.[2]

Expected against measured traffic loss

Change in search referrals · Reuters Institute survey of news leaders; causal estimate for Wikipedia

News leaders’ expectationsearch referrals, next three years
−43%
Measured: Wikipedia, Englishcausal estimate after AI Overviews
−5.45%
Measured: Wikipedia, vs Frenchalternative comparison
−4.82%

Different sites and measures: news publishers’ three-year expectation against a measured effect on an encyclopedia. The gap shows how much of the fear is not yet in the data.[9][3]

What cuts against this. Daily use of standalone AI apps is still around one adult in ten in Deloitte’s forecast, and only 9% of US adults are extremely or very confident that information from AI, chatbots or search is reliable.[27] For entertainment, people mostly browse and scroll (feeds, home screens, cover art) rather than formulate a request; Gemini and ChatGPT, asked to test this report, both made that point. The agentic forecasts are stated expectations, not observed behavior (see Agentic AI and the Consumer).[28]
Author’s forecast · confidence: mediumBy 2031 an AI answer, spoken or on screen, is the default first step for information and how-to learning among most Americans under 45, and much of it is Google’s answer rather than a new assistant’s. News follows more slowly because of trust. For entertainment, AI mostly works inside the feed and the home screen, and asking (“find me something like…”) stays a secondary way in.
Chapter 3 of 9
Shift 2 · driver: AI

A tutor in every pocket

Self-directed learning moves from searching and watching to conversing with an AI tutor; the evidence says designed tutors teach, while general chatbots mostly help people finish tasks.

Adoption is already broad. 54% of US teens have used chatbots for schoolwork and 10% do all or most of it with chatbot help. About a third of teens think AI would teach a skill better than a person, the only task Pew asked about where more teens favored AI.[4] Duolingo, whose paid tier now includes an AI video-call tutor, reported 58.7 million daily users in the second quarter of 2026, up 23% in a year.[29]

The best evidence is strong but narrow. In a Harvard physics course, a randomized crossover trial with 194 students found learning gains with a custom AI tutor more than double those of in-class active learning, an effect of 0.73 to 1.3 standard deviations, in a median 49 minutes. The tutor was fed the professors’ pre-written step-by-step solutions; it was one course, one session per topic and no test of retention.[5] In Nigeria, a six-week, teacher-guided after-school program using GPT-4 in nine public schools raised scores on its final assessment by 0.31 standard deviations (0.23 for English) and by 0.21 on the regular end-of-term exam. Only 759 of 1,328 assigned students sat the final test, and the largest gains went to stronger students.[30]

The warning is just as clear. In a randomized trial with about 1,000 Turkish high school students, cited in the OECD’s Digital Education Outlook 2026, access to GPT-4 raised math practice scores by 48% (general chatbot) and 127% (tutor designed for learning). On a closed-book exam afterwards, the general chatbot group scored 17% lower than students who never had AI, and the tutor group scored about the same as them. Students using the general chatbot did not notice they had learned less.[6][31] The OECD warns that AI used without teaching guidance may improve performance without real learning gains.[31]

Experts lean hopeful but split. In Elon University’s canvassing of about 300 technology experts, 42% expect AI to change curiosity and the capacity to learn for the better by 2035 and 29% for the worse, while 50% expect the capacity to think deeply to get worse.[32]

Performance is not learning

Math scores against students without AI · Türkiye, ~1,000 students · randomized trial

Practice, with AI available
General-purpose chatbotvs practice without AI
+48%
Tutor designed for learningvs practice without AI
+127%
Closed-book exam, without AI
General-purpose chatbotvs students who never had AI
−17%
Tutor designed for learningno measurable difference
≈0

The −17% bar shows the size of a drop. The tutor group’s exam result was not statistically different from zero.[6]

What cuts against this. Every positive trial so far used educator-built content or teacher supervision and lasted from one session to six weeks; most people use general chatbots on their own. School phone limits are spreading: 33 states had enacted K-12 cellphone bans or limits by March 2026.[33] No study yet follows self-directed adult learners over years.
Author’s forecast · adoption: medium-high · outcomes: lowBy 2031 most self-directed learning in the US (languages, skills, homework help, explainers) starts with a conversation, often by voice, and AI turns any document or video into a lesson, a quiz or an audio explainer. Whether this makes people more skilled depends on tutor design and verified course content, and so far the best results come with a teacher or institution behind the tutor.
Chapter 4 of 9
Shift 3 · driver: AI

Synthetic supply, curated demand

AI multiplies the supply of music, video and stories far faster than demand; human creators, labels and recommendation systems decide what gets watched.

Supply. Deezer said in July 2026 that more than half of its daily uploads are AI-generated, about 90,000 tracks a day; in April the share was 44% and the volume about 75,000.[7][8] YouTube says more than a million channels used its AI creation tools every day in December 2025, and it plans tools to make Shorts with your own likeness, simple games from a text prompt, and music.[34] Netflix is using AI for recommendations, ad formats and production, and said in April it would add a vertical mobile feed that carries video podcasts.[35]

Demand. In April, AI tracks drew only 1–3% of Deezer streams, and 85% of those streams were flagged as fraud. In Deezer’s survey 97% of respondents could not tell AI music from human music, and 80% wanted AI tracks labeled.[8] On a fresh YouTube account, Kapwing counted 21% of the first 500 Shorts as AI-generated “slop”; YouTube has made reducing such content a 2026 priority.[36][34]

Money. McKinsey estimates AI could touch workflows behind about 20% of original content spend within five years and that, if AI use overtakes incumbent technology, around 60 billion dollars of annual US distribution revenue could be redistributed within five years of mass adoption. In its scenario, open platforms gaining 5% more of US TV and film viewing hours would cut distribution revenue by 13.2 billion dollars and add 7.5 billion to the platforms. It also notes AI output is not yet of premium production quality.[11] Deloitte expects generative video to empower independent creators, and microdrama in-app revenue to roughly double to 7.8 billion dollars in 2026.[1]

Retreats. OpenAI discontinued its Sora app in 2026, and Disney left a planned 1 billion dollar investment and character license with it.[37][38]

AI music on Deezer: uploads against listening

Percent · company figures

Share of daily uploads, April 2026about 75,000 tracks a day
44%
Share of daily uploads, July 2026about 90,000 tracks a day
over 50%
Share of streams, April 2026
1–3%

The streams bar is drawn at the 2% midpoint of the reported 1–3% range.[7][8]

What cuts against this. The figures come from platforms and vendors with a stake in the story, and no independent audit measures how much of what people watch is AI-made. If quality reaches premium level faster than McKinsey assumes, AI content could win attention directly rather than through creators.
Author’s forecast · confidence: mediumBy 2031 most newly uploaded music, short video and illustration is AI-made or AI-assisted, while most listening and viewing time still goes to named people and brands. Labels, provenance and the recommender become the main quality filter, so trust in the curator matters more than access to content.
Chapter 5 of 9
Shift 4 · driver: AI

Media you talk to

Entertainment becomes two-way: AI companions, interactive characters and prompt-made games grow fastest among the young; fully generated games stay early.

Companions are already mass. 72% of US teens have tried an AI companion and 52% use one regularly; entertainment is the most common reason (30%).[12] Sensor Tower data put US time in AI companion apps at about 705 million hours in the first quarter of 2026, against about 280 million in dating and social discovery apps.[13] 47% of US teens have used chatbots for fun or entertainment.[4]

But it is concentrated. In a study of about 1.1 million ChatGPT conversations, only 1.9% of messages were about relationships and personal reflection and 0.4% about games and role play.[14] MIT and OpenAI found emotional engagement absent from most ChatGPT conversations and concentrated in a small group of heavy users; voice use was linked to better wellbeing when brief and worse outcomes with prolonged daily use.[39]

Interactive worlds are early. Google DeepMind’s Genie 4 world model lets people create and change explorable worlds by text while they play, but it supports only “a few minutes of continuous interaction” and is in a research preview for testers.[40] Game makers are wary: 52% of professionals in GDC’s 2026 survey said generative AI is hurting the industry and 7% said it helps, though 36% use it.[41] YouTube plans text-to-game tools in 2026.[34] Newzoo forecasts players rising from 3.70 billion in 2026 to 4.07 billion in 2029.[42]

US time in AI companion apps against dating apps

Hours per quarter · Sensor Tower via OfficeChai

Q1 2025
AI companion apps
~580m h
Dating and social discovery
~330m h
Q1 2026
AI companion apps
~705m h
Dating and social discovery
~280m h

Vendor app-panel data; methods not published with the figures.[13]

What cuts against this. Companion use among minors is drawing regulation: the FTC ordered seven companion-chatbot companies to report on safety and age limits in September 2025, and California’s SB 243, signed in October 2025, requires safety protocols and reminders to minors every three hours that they are talking to an AI.[43][44] US adults are wary too: in an Elon University survey, 55% expect AI to harm social and emotional intelligence over the next ten years, against 9% who expect it to help.[45] World models are expensive to run and not yet able to hold a game together for more than a few minutes.
Author’s forecast · companions: medium · generative games: lowBy 2031 conversational characters, inside chat apps, games and kids’ and fan media, are a mainstream entertainment category for under-25s. Games that are generated while you play remain a niche next to conventional games with AI-driven characters.
Chapter 6 of 9
Shift 5 · driver: devices

Beyond the phone

New devices add an always-listening AI layer around the phone and TV rather than replacing them; display glasses stay small through 2030.

Glasses. IDC forecasts display-less smart glasses rising from 13.6 million units in 2026 to 27.3 million in 2030, display glasses from 3 million to 12.2 million and mixed-reality headsets from 3.2 million to 10.4 million.[15] ABI Research expects about 32 million consumer AR glasses in 2030, about 28 million of them without a display.[46] In the first half of 2026, 96% of AI glasses shipped had no display.[47] Google’s first Android XR eyewear, with Samsung, Gentle Monster and Warby Parker, starts with audio glasses this fall; display glasses come later.[16] Most glasses, in other words, are cameras, microphones and speakers for an assistant, not screens for video.

Computers and new AI devices. Gartner expected AI PCs to be 55% of PCs shipped in 2026 and “the norm” by 2029.[48] OpenAI’s first device, designed with Jony Ive, is reported to be a 200 to 300 dollar smart speaker with a camera for early 2027, with glasses not before 2028.[49]

The TV stays central. The smart TV is the main streaming device for 61% of US internet households, while rising memory costs make TV price increases “unavoidable” in TrendForce’s forecast.[22][50]

Global shipments forecast, 2026 to 2030

Million units · IDC, June 2026 · bars show 2030

Display-less smart glasses13.6m in 2026
27.3m
Display glasses (AR/XR)3.0m in 2026
12.2m
Mixed-reality headsets3.2m in 2026
10.4m

Global forecasts; they say nothing directly about US media time.[15]

What cuts against this. A breakthrough display or a launch from Apple or Samsung could pull display glasses forward. The two glasses forecasts agree on display-less volume in 2030 (IDC 27.3 million, ABI about 28 million) but define display devices differently.
Author’s forecast · confidence: mediumBy 2031 the phone and the TV still carry most media hours. Glasses and earbuds become the way people ask, listen and learn on the move (audio explainers, translation, “what am I looking at”), which favors audio and short answers over long video.
Chapter 7 of 9
Shift 6 · driver: wireless

Networks that keep up

Traffic keeps growing but networks are not the constraint in the US; the change that matters is upload, as cameras and AI send more data up.

Volume. Ericsson forecasts global mobile network traffic of 515 exabytes a month in 2031, about 2.5 times the end-2025 level, with smartphone use rising from 22 to 42 gigabytes a month. Video was about 75% of mobile data at the end of 2025.[17]

Direction. Uplink grew faster than downlink at 43 of 55 operators Ericsson studied, and its AI scenarios suggest uplink could be three times higher or more in 2031 than in 2025.[18]

Generations. Ericsson expects 6.4 billion 5G subscriptions and 180 million 6G subscriptions by 2031, with first 6G services around 2030 and early launches in markets such as the US, China, Japan and South Korea.[51][18] Dell’Oro expects Wi-Fi 7 in more than 90% of the Wi-Fi equipment market by 2029.[52] Wi-Fi 8 (IEEE 802.11bn), with certification planned for December 2027, targets 25% lower tail latency, 25% fewer dropped packets when roaming, and better device-to-device links such as phone to TV or headset to PC.[53] Most AI media and learning happens indoors, so home Wi-Fi matters as much as cellular. 96.9% of Americans can already get fixed broadband at 100/20 Mbps.[54]

Global mobile network traffic

Exabytes per month, including fixed wireless · Ericsson Mobility Report, June 2026

End of 2025
203 EB/month
2031 forecast
515 EB/month
Author’s forecast · confidence: medium-highThrough 2031 wireless shapes media at the edges (live AI on glasses, cloud games, sharing video) rather than at the center. 6G arrives too late in the window to change mass behavior; uplink capacity, home Wi-Fi reliability and latency for live AI assistants matter more than peak download speed.
Chapter 8 of 9
Counterforces

What holds it back

Seven forces slow the shifts: a fixed time budget, low trust, labeling and child-safety rules, schools pulling back, household budgets, the pull of live events and the cost of AI video.

ForceEvidenceEffect on the shiftsKind
Fixed time2.61 hours of TV a day, half of all leisure time.[19]New AI media must take time from existing media.Survey
Trust9% of US adults are extremely or very confident that information from AI, chatbots or search is reliable; 49% have little or no confidence. 39% say AI use in news would reduce their trust.[27][10]Favors labeled, curated sources and human creators.Survey
Labels and rulesEU AI Act Article 50 requires machine-readable marking of AI-generated content and disclosure of deepfakes from 2 August 2026 (2 December 2026 for systems already on the market); a California law and an FTC inquiry target companion chatbots for minors.[55][44][43]Makes AI content visible; companion apps for minors face limits.Observed
Schools33 states with K-12 phone bans or limits; OECD warns of performance without learning.[33][31]Slows AI learning habits during school hours; shifts them home.Observed
Budgets41% of US consumers cancelled a paid streaming video service in the past six months; PwC sees subscription fatigue and more bundling.[56][23]AI features arrive inside ad tiers and bundles, not as new paid services.Survey
Live and sharedLive and experience-led revenue grows 5.2% a year to 2030, faster than total media.[23]Shared events keep value as personal AI media spreads.Forecast
Economics of AI videoOpenAI closed Sora; Disney exited its deal.[37][38]Generated video may stay a tool for creators, not a consumer app.Company
Author’s read · confidence: mediumNone of these forces stops the shifts; together they make them slower and more uneven by age than the forecasts from platforms and device makers suggest.
Chapter 9 of 9
Judgment, not forecast

Scenarios to 2031

The author’s central case splits by purpose: ask to know, scroll to watch. Two alternatives push one side further; a pushback would slow both.

How good are such forecasts? Infrastructure forecasts have aged well: in 2020 Ericsson forecast 3.5 billion 5G subscriptions by the end of 2026, and it counted 3.1 billion by the first quarter of 2026, on track.[57][18] Forecasts of new behavior have not: in 2022 Gartner predicted that 25% of people would spend at least an hour a day in the metaverse by 2026.[58] This report therefore trusts network and device forecasts more than forecasts of new habits.

ScenarioWhat 2031 looks likeEvidence forEvidence againstConfidence
Ask to know, scroll to watch
central case
AI answers, mostly Google’s and the big assistants’, are the first stop for information and how-to learning. Entertainment discovery stays in feeds and home screens, with AI inside YouTube, Netflix, Spotify and games.29% daily AI-summary viewers forecast for 2026; 54% of teens use chatbots for schoolwork; Google AI forecast to pass ChatGPT in users by 2027; 1M+ YouTube channels use AI tools daily.Measured traffic loss so far is about 5%; only 2% name AI a top news source; learning gains need designed tutors.Medium
AI front door everywhereAssistants on phones, glasses and TVs also become the main way into entertainment; they can play and buy inside other apps.Ericsson: 38% expect to delegate daily tasks by 2030; audio glasses favor asking; teen chatbot use for fun at 47%.Platforms keep their apps closed; people browse rather than ask for entertainment; agentic use is expectation, not behavior.Low
Platforms absorb everythingEven for information, AI stays inside search, YouTube and existing apps; standalone assistants stay niche.Standalone AI app daily use about 10%; AI Overviews favor Google-owned sources; Sora closed.ChatGPT-style assistants already handle many non-work questions and homework.Medium
PushbackTrust problems, “slop”, school limits and companion rules slow adoption; AI stays behind the scenes.9% extremely or very confident in AI information; EU labeling rules; FTC companion inquiry; 33 states limit phones in school.Teen and companion usage is broad and rising; people use tools they say they distrust.Low to medium

Confidence describes how well the evidence supports each story, not a probability. The central case changed after the Gemini and ChatGPT research rounds: both rated AI inside existing platforms better supported than an AI front door for everything.

What would change the picture

Each November

Daily use of AI answers against standalone AI apps

Deloitte’s TMT Predictions and EMARKETER. Growth inside search supports the central case; standalone assistants pulling ahead support AI front door everywhere.[1][26]

As published

Measured traffic effects of AI answers

Causal studies such as the Wikipedia estimate, and Reuters Institute traffic data each January. Losses well beyond 5% would mean asking is replacing browsing faster than expected.[3][9]

Each year

Trust in AI information and news

AP-NORC and Gallup. Rising confidence supports the central case; falling confidence supports Pushback.[27][10]

As published

Retention in AI-tutoring trials

OECD and randomized trials. Gains on exams without AI, over a semester or longer and without a teacher, would make self-directed AI learning a durable habit.[31][6]

Quarterly

AI share of listening and viewing, not just uploads

Deezer and platform labeling disclosures, now backed by EU marking rules. A rising share of streams would mean AI content wins attention directly.[8][55]

Each June and November

Display glasses and uplink traffic

IDC forecasts and Ericsson Mobility Reports. Display glasses beating forecasts, or uplink growing faster than expected, would pull on-the-move media forward.[15][18]

Method, limits and flags

Read before citing

Question. How will advances in AI, wireless communication and consumer devices change how people use media for entertainment and for self-directed learning by 2031?

Scope. US consumers first, with global forecasts where US-only ones do not exist. Media means video, audio, music, games, news, social and interactive formats. Learning means self-directed and lifelong learning; schools appear only where they shape consumer habits.

Sources. Research papers, research institutes and analysts’ forecasts first, then company disclosures and major news and trade publications. Each reference carries its type and date. Observed, survey, company and forecast figures are kept apart and labeled.

How the research was done. An earlier draft of this report, researched with Claude, Gemini and ChatGPT, measured US video viewing; its verified figures now form the 2026 baseline in chapter 1. For the forecast, Claude gathered expert forecasts, papers and news; Gemini (research papers, government, standards) and ChatGPT (analysts, companies, major publications) then answered the same written brief, and Claude verified and merged their findings. Framing, the six shifts, the scenarios and conclusions are the author’s.

What this report is not. The “author’s forecast” boxes and the four scenarios are judgments based on the evidence cited. Confidence labels describe how well the evidence supports a statement, not a probability. Signposts give a direction and a source, not a numeric threshold.

Flags and open questions

  1. Forecasts, not measurements. As agreed for this edition, the report leans on what analysts, researchers and companies forecast to 2030–2031 and does not require each claim to be validated by an independent third party. Many forecasters sell to the industries they forecast (Ericsson, IDC, ABI, Deloitte, McKinsey, PwC, EMARKETER), and company figures (Deezer, YouTube, Netflix, Duolingo, Google) are claims. Each is labeled.
  2. Research round with Gemini and ChatGPT. Both models answered the same written brief. Claude opened the sources behind every new figure used here. Their main effect was to change the central case: both rated AI inside existing platforms better supported than an AI front door for all media.
  3. Corrections from that round. The Turkish figures (+48%, +127%) measure practice with AI available, not learning; on the exam without AI the tutor group did no better than students without AI. The Nigeria result of 0.31 standard deviations is a composite (0.23 for English) from a teacher-guided program with high attrition. ChatGPT reported that 3% of US adults used AI as a news source in the past week; Gallup’s 3% refers to adults 50 and older, so the report uses Gallup’s 2% (top-three source) and 7% (rely on AI).
  4. Not used because not verified. Figures from a Qualcomm consumer survey, a Qualcomm readout of 3GPP 6G standards meetings, Omdia glasses shipments, an Omdia satellite direct-to-device forecast, IEA data-center electricity projections, an S&P Global Kagan glasses survey and a Google and Eedi tutoring trial could not be opened in full or confirmed, and are left out.
  5. Secondary reports. Where the primary page could not be opened, a named secondary report is cited: Pew’s AI-summary study via Search Engine Roundtable; Gartner’s AI PC forecast via ITPro; Sensor Tower via OfficeChai; Deezer’s April figures via AlternativeTo; the FTC orders and California SB 243 via DLA Piper; OpenAI’s device plans via MacRumors reporting The Information (not confirmed by OpenAI).
  6. Preprints and co-authors. The Wikipedia traffic study and the AI Overviews audit are preprints. The ChatGPT usage paper and the MIT study were co-authored by OpenAI researchers.
  7. Teen chart. Pew gives “about 4 in 10” for two uses; the chart draws them at 40%.
  8. Baseline. Nielsen changes its Gauge method from September 2026, so the August 2026 shares in chapter 1 will not compare directly with later months.
  9. Not found. No source found for: a forecast of US hours by medium to 2031; an independent measure of how much consumed media is AI-made; trials longer than a semester of adults learning with AI on their own; daily minutes of media or learning on smart glasses; how often people check content-provenance labels; US in-car media (see the site’s report on media on the move).

Change log

  • November 2026 edition
  • Added
    First edition, data as of 1 November 2026. Rescoped before publication from a US video benchmark to a forecast of media and learning.

About this report

Personal research by Chuhee Lee. Research, source verification and drafting were carried out with several AI models, including Claude (Anthropic), Gemini (Google) and ChatGPT (OpenAI). Every factual claim is sourced to the references listed below; topic selection, framing and conclusions are the author’s. These are personal views, not those of any employer.

References

Continue down the map

ForcesShiftsSignalsOpportunities
Follow new editions

Each report is refreshed on a schedule. Know when it changes.

Next refresh dates are listed on the home page.