Key takeaways
- Content supply is expanding faster than focused attention, intensifying competition for people’s time.
- Short dramas and simultaneous TV-and-phone use show how entertainment is adapting to fragmented attention.
- AI can weaken independent learning when it supplies answers, while thoughtfully designed tutors can improve outcomes.
- Gaming offers a playbook for measuring behaviour, testing changes and improving experiences beyond launch.
- Opportunities include attention measurement, adaptive learning, trusted curation and stronger relationships with fans.
“A wealth of information creates a poverty of attention.”
Herbert Simon wrote that in 1971, decades before smartphones, infinite feeds and generative AI. His observation has become an increasingly useful way to think about the next wave of technology companies. We are making information cheaper and easier to produce, while competing for a resource that remains stubbornly scarce: the attention of the person receiving it.
AI accelerates both sides of this competition. It helps creators produce more articles, videos, music, games and advertisements. It also helps platforms select, personalise and distribute that content more effectively. The result is a growing imbalance between what we can generate and what anyone has time to meaningfully consume.
At Remagine Ventures, where we invest at the intersection of AI and consumer behaviour, I think this creates an important shift in where value will accrue. Creation tools will continue to matter. But the ability to earn attention, understand its quality and turn it into a satisfying experience may become more defensible than the ability to produce another piece of content.
The evidence suggests that this shift is already happening.
BCG’s October 2026 report, Media’s New Fight for Eyeballs, Profits, and Control, compares focused and background media consumption between 2018 and 2025. Its finding is striking: active media time remained relatively flat, while growth came largely from media consumed alongside other activities. In its survey, 61% of respondents said spending more time on one media format meant spending less on another.

This distinction matters commercially. A programme playing on a television, a podcast running during work and a video someone actively watches may all register consumption. They do not necessarily receive the same attention, create the same memories or generate the same willingness to pay.
We have become very good at counting exposure. Understanding what that exposure achieves is harder.
Consider the living room. Omdia’s 2026 research found that nearly three in four US TV viewers regularly use their smartphones for other media while watching television. This extends well beyond teenagers: among 55–64-year-olds, simultaneous media use rose from 42% in 2023 to 56% in 2026.
The television may still be on, but the phone is competing for the next moment of interest. A slow scene now competes with a feed that can offer something new immediately. For producers, advertisers and streaming services, that creates pressure to earn attention repeatedly within a viewing session.
Short dramas are a particularly revealing response. These are serialised stories designed for mobile viewing, with compressed episodes, high emotional stakes and frequent cliffhangers. Their structure makes the decision to continue small and easy, even when the cumulative time spent becomes substantial.
According to Sensor Tower’s June 2026 report, short-drama apps exceeded 850 million downloads globally in the first quarter, up 140% year on year. Estimated in-app purchase revenue reached approximately $750 million, up 20%. By April, average daily time spent had reached 25 minutes, compared with roughly 35 minutes for traditional streaming apps in its measurement. These are mobile-app estimates, rather than a measure of the entire television market.
The gap between download growth and revenue growth is worth noticing. Capturing interest is one challenge; acquiring viewers profitably and retaining paying customers is another.

To me, short dramas look like a meeting point between entertainment and mobile gaming. The creative work draws someone into a story, while the product experience determines how they discover it, progress through it and pay for more. AI could make localisation, dubbing and production cheaper, but those savings will also make it easier for competitors to enter. Distribution, distinctive intellectual property and repeat viewing become critical.
It would be a mistake, however, to conclude that audiences have simply lost the ability to engage with anything substantial.
Edison Research’s Infinite Dial 2026 found that 58% of Americans aged 12 and older had listened to or watched a podcast in the previous month, a record for its survey. That does not tell us how deeply they listened, but it complicates the idea that every successful format must become shorter.
Similarly, McKinsey’s 2025 attention research found high reported focus among book readers and console and PC gamers. Books registered 81% on its high-focus measure; console and PC gaming reached 73%, compared with 57% for streaming video. These are survey measures, but they suggest that people still devote sustained attention to experiences they find worthwhile.
The more useful question is what makes people willing to concentrate. Participation, personal relevance, curiosity and emotional investment all offer routes beyond simply shortening everything.
AI introduces another complication: we can increasingly consume an answer without doing the work that once produced understanding.
For publishers, that changes distribution. In Pew Research Center’s analysis of browsing by 900 US adults in March 2025, users clicked a traditional search result in 8% of visits with an AI summary, compared with 15% without one. This was an observational comparison, not an experiment proving that summaries caused the difference. Still, it illustrates the commercial tension: an answer can satisfy a reader without sending them to the source.

For the reader, the question goes beyond whether a publisher gets a click. What happens when summaries and generated answers routinely replace reading, comparing evidence and constructing an argument?
“AI is dumbing us down” is an understandable concern, but it overstates what the evidence establishes. Watching short dramas or browsing a phone during television does not demonstrate declining intelligence. Nor did those behaviours begin with generative AI.
There is, however, evidence that certain uses of AI reduce the mental effort people apply, and that getting better results with a tool can conceal weaker performance without it.
A Microsoft Research study presented at CHI 2025 surveyed 319 knowledge workers. Higher confidence in generative AI was associated with less reported critical thinking. The researchers also found that thinking shifted towards checking, integrating and overseeing AI outputs. This was self-reported evidence, rather than proof of permanent cognitive decline, but it raises an important concern about what users choose to delegate.
Stronger causal evidence comes from a randomised study published in PNAS, involving nearly 1,000 high-school mathematics students. Students given a standard GPT-style assistant performed better during practice, but scored 17% worse than the control group on a subsequent exam without assistance. A tutor designed to provide hints and safeguard learning largely eliminated that harm, although it did not produce a significant improvement over the control group on the unassisted exam.
The implication is specific and important: improved performance while using AI is not necessarily evidence of learning.
The same study also shows why design matters: students interacted more substantively with the tutor built to guide them than with the version that readily supplied solutions.

There is encouraging evidence too. A separate 2025 randomised trial in a Harvard physics course found that a carefully designed AI tutor produced greater immediate learning gains in less time than the comparison classroom lessons. That result does not establish superiority across every subject or over the long term. It does show that design choices can change the outcome.
My concern is that we are making it increasingly easy to bypass the effort through which understanding develops. A tool that removes administrative friction is useful. A tool that removes every opportunity to reason, practise or struggle may create a different kind of dependency.
That is also where the business opportunity becomes more interesting.
Gaming provides a useful precedent. Developers have long treated the experience as something they can measure and improve after launch. They analyse where players abandon an onboarding sequence, how difficulty affects progression, which features bring people back and whether changes improve retention.
Unity, for example, offers tools for testing different game configurations and measuring their effects on retention and revenue. The feedback cycle connects a product decision to observed behaviour, rather than relying entirely on creative intuition.

Other media businesses already experiment with recommendations, thumbnails and pricing. AI could extend that approach deeper into the experience: adapting an explanation to a learner, localising a story for a market, helping a viewer discover a relevant series or changing how a fan interacts with a fictional world.
But optimising an experience for users requires deciding what “better” means. More minutes can indicate enjoyment, confusion or difficulty finding the exit. Completing a lesson can indicate mastery or simply that the answer was supplied. Watching another episode can reflect enthusiasm without establishing a durable customer relationship.
The companies that measure these distinctions well could have a meaningful advantage. I see several opportunities:
| Opportunity | Product and customer | Evidence of value |
|---|---|---|
| Better attention measurement | Tools for media owners and advertisers that combine behavioural signals with consented research and experiments | Incremental recall, conversion and repeat use, beyond impressions |
| Experimentation for media | Software for publishers and studios to test onboarding, discovery, episode packaging and monetisation | Higher retention and contribution margin across viewer cohorts |
| Adaptive learning | Tutors and training products that adjust explanations, practice and feedback | Knowledge retained and tasks completed independently |
| Personal discovery and filtering | Paid assistants that curate information, entertainment or shopping around a user’s stated goals | Less search time, better decisions and continued willingness to pay |
| Direct relationships with fans | Community, membership, commerce and interactive experiences around creators and intellectual property | Renewals, repeat purchases and audience relationships beyond rented feeds |
These opportunities have different economics. An analytics company needs access to reliable data and a way to show that its recommendations cause improvement. A consumer assistant must save enough time or improve enough decisions to justify a subscription. A short-drama studio needs content economics that survive acquisition costs, rather than impressive download numbers alone.
I am particularly interested in products that can connect understanding to action. A dashboard that reports a drop-off is useful. A system that helps a team diagnose the cause, test a change and measure the result becomes part of how the business operates.
The potential advantage is the accumulated knowledge of what works for a particular audience and context, supported by experiments and integrated into the customer’s workflow. Generating a hundred variants is increasingly accessible. Knowing which differences matter, and why, is much harder.
There is also room for businesses that deliberately ask for less attention. A research assistant that helps someone identify the three documents worth reading, or a shopping service that narrows a confusing choice, can create value by shortening the interaction. A subscription model may align better with that outcome than a business dependent on maximising advertising inventory.
For education, the opportunity is especially compelling. We can borrow gaming’s progression, feedback and sense of achievement while measuring whether the learner becomes more capable. The valuable product is one people return to because they are improving, with evidence that survives when the assistant is switched off.
For creators and media owners, I expect trusted relationships and distinctive intellectual property to become more important as production expands. Personalisation can help someone find a story they love. It cannot guarantee that the story gives them a reason to care. A recognisable voice, a compelling world and a community with shared interests remain difficult to manufacture on demand.
As an investor, I would ask a founder building in this space three questions: whose attention are you earning, what value does that person receive, and how do you know the experience is improving?
Simon’s observation has become a product-design challenge as much as an economic one. AI gives us extraordinary capacity to produce, personalise and persuade. The opportunity is to use that capacity to make people’s time more worthwhile.
The companies I want to back will be able to demonstrate that they do.
- AI Can Make More Content. Who Will Pay Attention? - November 17, 2026
- Weekly Firgun Newsletter – November 13 2026 - November 13, 2026
- AI Safety: Who Should We Believe? - November 10, 2026

