Last year, various surveys, including reliable indicators, have highlighted a significant decline in reading habits over the past decades.
The most striking evidence is not simply that people read less, but that their capacity for deep reading is weakening. According to OECD data, the proportion of 15-year-olds who fail to reach minimum reading proficiency has now risen to nearly one in four across advanced economies, with sharp declines in tasks requiring inference, evaluation, and integration of information across texts.
In the United States, NAEP scores show that average reading performance among 13-year-olds has fallen to its lowest level in decades, reversing long-standing gains. Laboratory studies mirror these trends: experiments comparing print and screen reading consistently find that readers of digital texts score 1030% lower on comprehension and recall, particularly for longer and conceptually demanding material.
{"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/tcp-photo-syndey-16X9.jpg","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/tcp-photo-syndey-1x1-2.jpg","eyebrow":"","headline":"Get more insights from Tomas Chamorro-Premuzic","dek":"Dr. Tomas Chamorro-Premuzic is a professor of organizational psychology at UCL and Columbia University, and the co-founder of DeeperSignals. He has authored 15 books and over 250 scientific articles on the psychology of talent, leadership, AI, and entrepreneurship. ","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"https:\/\/drtomas.com\/intro\/","theme":{"bg":"#2b2d30","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#3b3f46","buttonHoverBg":"#3b3f46","buttonText":"#ffffff"},"imageDesktopId":91424798,"imageMobileId":91424800,"shareable":false,"slug":""}}
Eye-tracking and cognitive load research further indicates that frequent digital readers engage in more skimming, less rereading, and shallower semantic processing. Crucially, these effects are not confined to weaker readers. Even highly educated adults now report shorter attention spans for long-form text and greater mental fatigue when reading complex arguments, suggesting that the decline of reading reflects not a loss of literacy, but an erosion of the cognitive endurance and attentional discipline that deep reading uniquely develops.
Not just children
To make matters worse, various robust data indicators show that adults are spending less time reading, especially for pleasure. For instance:
(1) A large time-use study analyzing diary data from over 236,000 Americans found that the share of adults who read for pleasure on an average day dropped from about 28% in 2003 to just 16% in 2023, a roughly 40% decline over two decades.
(2) That same research showed a steady annual fall of about 3% per year in the prevalence of daily leisure reading among U.S. adults.
(3) An earlier report by the World Economic Forum indicated average daily reading time in the U.S. declined from about 23 minutes per day in 2004 to around 16 minutes by 2019, even before the most recent decades drop.
(4) In the U.S., fewer adults now report reading books for pleasure: about 48.5% of adults said they read at least one book in the past year in 2022, down from 54.6% in 2012.
A real concern?
Should this really concern us? Perhaps not. After all, reading is just one medium through which humans have ingested information and exercised their minds, including for deep thinking. For most of history, knowledge travelled orally rather than silently on the page. Ancient cultures relied on storytelling, poetry, and song to preserve and transmit complex ideas: Homers epics were memorized and performed long before they were written down; Greek philosophy unfolded through dialogue rather than textbooks; and entire moral, legal, and scientific traditions were passed across generations through ritualized speech, music, and debate. From this perspective, the book is a relatively recent cognitive technology, not an eternal prerequisite for intelligence (consider that Socrates and his fellow philosophers were concerned by the invention of writing, thinking it may atrophy memory).
And today, once again, new media promise alternative routes to learning and thinking: immersive simulations, virtual and augmented reality, AI tutors, and even speculative neuro-technologies all claim to enhance understanding, creativity, or memory without requiring sustained reading at all. Perhaps these tools will indeed make us more knowledgeable or even smarter. Needless to say, not all reading is cognitively ennobling. Wading through a disposable airport romcom is unlikely to stretch the mind more than an unscripted, curious conversation with a stranger at a bar. The real question, then, is not whether reading is declining per se, but whether whatever replaces it can cultivate the same depth of attention, reflection, and intellectual effort that serious reading has historically demanded.
Digital diversions
To be sure, every person is different and even among those who are reading less, former reading time may be recycled or reutilized in many different ways. That said, there is a clear trend to devote more time and attention to the very technologies that have increasingly monopolized our focus over the past two decades. Time-use and media-consumption data strongly suggest that leisure reading has been displaced not by other cognitively demanding activities, but by screen-based media. In the United States, Bureau of Labor Statistics time-use surveys show that average daily reading for pleasure fell from about 23 minutes in the early 2000s to roughly 16 minutes by 2019, while time spent on digital devices and television increased steadily. Over the same period, social media use expanded rapidly: Pew Research Center reports that adult social media adoption rose from around 5% in 2005 to over 80%, with many users spending multiple hours per day on these platforms. Globally, Digital 2024 data indicate that adults now spend about 2.5 hours per day on social media and more than 6.5 hours per day consuming digital media overall, compared with a small and declining fraction of time devoted to reading books or long-form text.
While time spent reading traditional text has declined, many adults are substituting other sustained listening activites that share some cognitive benefits of reading; for example, Edison Researchs Infinite Dial reports that the share of Americans ages 12 and older who listen to podcasts weekly has grown from about 11% in 2013 to over 60% in 2024, with average weekly listening around seven hours, suggesting deeper engagement than typical short-form scrolling. Audiobook consumption has also risen sharply: the Audiobook Publishers Association and APA Foundation data show that nearly 50% of American adults listened to an audiobook in the past year, with frequent listeners averaging more than 6 hours per week, offering another way to engage with complex narrative and informational content. These trends indicate that although reading declines are real, listening to long-form spoken content (whether through podcasts or audiobooks) is increasingly filling part of the gap, providing extended attention to ideas, storytelling, and analysis in ways that resemble some of readings cognitive and reflective benefits.
Unique benefits
And yet, cognitive and developmental psychology remind us of the distinctive benefits of traditional reading, especially when it comes to thoughtful immersion and deep processing of text. Decades of research converge on at least five lessons worth remembering.
First, sustained reading strengthens attention and cognitive endurance, training the ability to concentrate for extended periods without external stimulation, a capacity that is closely linked to academic achievement and complex problem-solving.
Second, reading supports deeper comprehension and critical thinking: compared with fragmented or audiovisual media, linear text promotes inferential reasoning, abstraction, and the integration of ideas across time.
Third, regular reading expands vocabulary and conceptual knowledge, which in turn predicts reasoning ability (especially verbal and crystallized intelligence), learning speed, and even long-term occupational outcomes.
Fourth, reading fiction in particular has been shown to enhance perspective-taking and social cognition, improving peoples ability to understand others emotions, intentions, and mental states.
Finally, early and sustained exposure to reading plays a foundational role in brain development, literacy, and self-regulation, with long-lasting effects on educational attainment and cognitive resilience across the lifespan.
None of this means that reading is the only route to thinking, or that newer media are inherently inferior, but it does suggest that some cognitive benefits are unusually hard to replicate without sustained engagement with text.
And if you made it this far, thank you for reading this.
{"blockType":"mv-promo-block","data":{"imageDesktopUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/tcp-photo-syndey-16X9.jpg","imageMobileUrl":"https:\/\/images.fastcompany.com\/image\/upload\/f_webp,q_auto,c_fit\/wp-cms-2\/2025\/10\/tcp-photo-syndey-1x1-2.jpg","eyebrow":"","headline":"Get more insights from Tomas Chamorro-Premuzic","dek":"Dr. Tomas Chamorro-Premuzic is a professor of organizational psychology at UCL and Columbia University, and the co-founder of DeeperSignals. He has authored 15 books and over 250 scientific articles on the psychology of talent, leadership, AI, and entrepreneurship. ","subhed":"","description":"","ctaText":"Learn More","ctaUrl":"https:\/\/drtomas.com\/intro\/","theme":{"bg":"#2b2d30","text":"#ffffff","eyebrow":"#9aa2aa","subhed":"#ffffff","buttonBg":"#3b3f46","buttonHoverBg":"#3b3f46","buttonText":"#ffffff"},"imageDesktopId":91424798,"imageMobileId":91424800,"shareable":false,"slug":""}}
Foldable phones have spent years trying to justify themselves. Some were too fragile, others too bulky, and most felt like solutions in search of a problem. The Galaxy Z TriFold is Samsungs clearest attempt yet to answer a more reasonable question: Can one device replace the phone-tablet combo without becoming a chore to carry?
Coming to the United States later this month, the TriFold folds twice, opens into a 10-inch screen, and closes back into a pocketable form. Its an assertive design, but not a novelty play. Samsung seems very aware that this kind of device only makes sense for a specific kind of user.
[Photo: Emily Price]
The double fold is the trick, but the software does the real work
The headline feature is the dual hinge. Closed, the TriFold behaves like a premium smartphone. Open it fully, and it becomes a genuinely usable tablet-size workspace.
That space matters. You can run three apps side by side, resize them, and keep them anchored even when calls or notifications interrupt. Samsungs task bar lets you jump back into complex layouts without rebuilding them, which is a small thing until youve lost your place mid-task one too many times.
We had a chance to try the phone first hand at a Consumer Electronics Show (CES) preview. The first time you open the device, the folding mechanism, in particular, stands out. Fully open, you might not even notice youre holding a phone rather than a tablet. The three separate screens blend together seamlessly.
Samsung has also added guardrails. The phone will warn you if youre folding it the wrong way when you go to put it awaywhich feels less like hand-holding and more like protecting an expensive mistake.
Editing photos is where the bigger screen actually shows off
The TriFolds size gives Samsungs photo tools room to breathe, especially its generative editing features. Blake Gaiser, head of smartphone product management, says the difference is immediately obvious once you start using them.
We’re really well known for what we call generative editingbeing able to remove things from a photo, Gaiser told me during a demo this week. He took a photo that included a person, and then was able to select and remove that person from the photo in seconds. It understands everything that I want to pick out here, and I’m able to take all the pixels out of that.
He points to something thats easy to miss on smaller screens: cleanup details. Not only did it take the person out, but it took their shadow out as well, he said. So now I can look at both side by side each other, and you can see the shadow that she had there is gone.
Being able to zoom in on before-and-after images simultaneously sounds minor. But for people who actually edit photos regularly, its the difference between trusting the result and hoping for the best.
[Photo: Emily Price]
This is very much not meant for everyone
The TriFold is not designed for everyone. Samsung isnt pretending otherwise. Gaiser is blunt about the intended audience. It is for your top productivity people, he says.
That philosophy shows up most clearly in DeX (short for desktop experience), Samsungs desktop-style interface. On the TriFold, DeX treats the device like a full monitor. You can resize windows freely, stack them, snap them into place, and even create multiple desktops that remember their layouts.
So if I’m consistently looking at news articles and Samsung apps because I’m working on a piece or whatever, I could set those up in their own desktop, Gaiser said. Even when I clear the memory and everything, it remembers that setup.
Gaiser has been using the TriFold as part of his own daily setup, and not always as the primary device.
The two key things that I’ve done with this personally, in the three months that I’ve had this device: I have just a portable stand that I put it on, wireless keyboard, mouse, use it like a PC, he said. Or in my hotel room, I had my PC and I had this set up as a second monitor.
The TriFold supports wired and wireless display output, including 4K when wired, making it less of a stretch to imagine it replacing a second screen for travel or temporary setups.
Built sturdier than it looks
Triple-folding phones raise obvious durability questions. Gaiser acknowledges the complexity.
Because we have two different hinges on here. You have two different pivot points, he said. The phone uses magnets to keep it shut, but also to give the third screen a gentle pop after you open the first, making it easier to lift.
Samsung also leaned heavily into materials, using ceramic glass fiber, a titanium lattice, and carbon fiber reinforcements to protect the folding display. Gaiser was candid in comparing it with competitors.
[Photo: Emily Price]
Power without cutting corners
Under the hood, the TriFold runs on a customized Snapdragon 8 Elite chip, includes a 200-megapixel camera, and uses a 5,600 milliamphour battery spread across its three panels. That complexity is invisible to the userwhich is the point. The phone lasts through a full day of heavy use and charges quickly enough not to feel precious.
Samsung also worked with Adobe to create a subscription-based Lightroom-specific app that behaves like its desktop counterpart, reinforcing the idea that this device is meant for people who actually produce things on their phones. The phone will come with a free trial.
How it stacks up against other foldables
Huawei Mate XT
Huawei was the first with the Mate XT, proving that trifold hardware was possible. Availability is limited, software support is complicated outside certain markets, and it feels more like a statement piece than an everyday device.
Concept triple-folds from other brands
Several manufacturers have shown trifold concepts at trade shows. Most trifold devices are still prototypes, and thats fine. Building one is hard. Making one that survives daily life, and the bumps that come with it, is even harder.
Samsungs advantage isnt that it folded a phone twice. Its that its spent years figuring out hinges, software behavior, durability testing, and what users actually tolerate. The TriFold feels like the result of that learning curve rather than a shortcut.
So who should even consider this?
Samsungs own answer is narrow. Gaiser calls the target audience the top 1% heavy users.
Productivity tools, multi-window users, your ultra-top users, he said. Its not for everyone.
That honesty helps. The Galaxy Z TriFold isnt trying to convince casual users to upgrade. Its aimed squarely at people who already push their devices hard and want fewer things in their bag. Its not flawless, not cheap, and not subtle. But its also the clearest signal yet that foldables are moving out of the experimental phase and into something more practical, even if only for a small slice of users.
2025 was a banner year for cryptocurrencies on many fronts. Global regulation eased. Stablecoins powered $46 trillion in annual transactions. And major shifts in U.S. government policy spurred wider adoption. But with that expansion came a notable bump in crypto fraud.
A new report from Chainalysis, a blockchain data platform based in New York City, estimates that $17 billion in crypto was stolen last year through fraud and scams. Impersonation scams, where criminals pretend to be trusted entities or use fake tokens or websites to trick victims into sending them crypto, were up a jaw-dropping 1,400% year over year.
And while it’s much too early to gather any conclusive data for 2026, the year got off to an inauspicious start. Earlier this month, the FBI warned about the use of Bitcoin ATMs, saying the devices are a magnet for scammers to convince people to send money (their entire life savings, in some cases) overseas. And just this week, the fintech firm Betterment confirmed that hackers had broken into its systems earlier this month and used the data to send a fraudulent crypto note to users, which funneled money to a wallet controlled by the attacker.
Meanwhile, former New York Mayor Eric Adams launched a new crypto token on Monday that he said would combat antisemitism and promote blockchain education. It quickly lost 81% of its value, bringing about accusations of a “rug pull” across the crypto community.
Chainalysis warned in its report that this could be just the beginning of another year of new highs. “As we move into 2026, we expect further convergence of scam methodologies as scammers adopt multiple tactics and technologies simultaneously,” it wrote.
Early projections by Chainalysis indicate scammers in 2025 received at least $14 billion on-chain, a transaction that occurs directly on the blockchain (compared with a speedier and cheaper but riskier off-chain transaction). That’s a big jump from last year’s initial estimate at the same time of $9.9 billion.
Ultimately, the 2024 number settled at $12 billion following recalculations. The 2025 total is projected to come in above $17 billion, as more bogus wallet addresses are uncovered in the coming months. That would make last year’s rise in crypto scam losses the biggest since 2020 to 2021, when they doubled. Subsequent years have been fairly flat, hovering between $12 billion and $13 billion.
Scams were not only happening more frequently last year, the people perpetrating them were also pocketing more each time. The average scam payment in 2025 was $2,764, a 253% increase over 2024’s $782.
“The 2025 data reveal the extent to which cryptocurrency-enabled scams are becoming more sophisticated, organized, and efficient,” Chainalysis wrote. “There are no silver bullets to tackling such entrenched, industrial-scale scamming activity, and to be effective, a multipronged response is required.”
Impersonation scams were the biggest driver of losses. Not only were the number of those sorts of cons significantly higher, but the average amount people paid to the groups behind them was up 600%. Crime syndicates in East and Southeast Asia drove many of these, the report says, with forced labor compounds in Cambodia, Myanmar, and other regions forcing trafficking victims to operate the scamsthe most prolific of which was a phishing scam that targeted users of the E-ZPass toll collection system with a fake “outstanding toll.”
Artificial intelligence is becoming a weapon of crypto scammers as well. The technology’s ability to leverage large language models and deepfake technology makes the schemes more realistic. As a result, scams that used AI vendors to create on-chain links averaged a haul of $3.2 million, compared with $719,000 for those without.
While fraud was on the rise last year, there were some victories by law enforcement. Police in the U.K. recovered 61,000 in stolen Bitcoin. And TerraUS and Luna crypto developer Do Kwon, a Stanford graduate known by some as the cryptocurrency king, pleaded guilty in August to fraud charges stemming from the collapse of Terraform Labs, the Singapore-based firm he cofounded in 2018. Customers lost $40 billion in that fraud, a figure that exceeded the total losses of Sam Bankman-Fried’s FTX. Kwon was sentenced to 15 years in prison.
Should I take this project? Say yes to the new job offer? Stick with this plan or walk away? Every choice we make can feel huge. And every path has its own set of risks and rewards. There are always more questions for every life-changing decision. Sometimes the pros-and-cons lists feel more like busywork than progress. You check off the boxes, stare at the lists, and still end up confused, stuck in the same mental loop. Thats why I rely on the rule of 3 framework to make tough decisions. I hope it helps you clarify your life-changing choices.
How it works
Whenever youre stuck, force yourself to create three paths: B, C, and D.
Why not A? A is usually the default for most people. The thing youre already doing. The path of least resistance. It doesnt need your help. What you need are alternatives.
Then comes the second step, and this is where most people stop thinking too soon. Now, for each path, think through:
First-order effects
Second-order outcomes
And third-order consequences
And then, and this matters, choose the path with the most meaningful but least life-changing consequences.
Why the two-option path doesnt work
When you only have two options, your brain keeps going back and forth. Right vs wrong. Safe vs risky. Smart vs stupid. You stop being logical. Theres a term for it: binary bias or black-and-white thinking. We do it all the time. Two choices feel better. But they are not. Theyre restrictive and create a lot of unnecessary pressure.
Most decisions are not binary, and there are usually better answers waiting to be found if you do the analysis and involve the right people, Jamie Dimon, the CEO of JPMorgan Chase, says.
Three options open things up. Adding a third option reduces your emotional load and improves perceived control. You feel less trapped. And more capable. For example, if you are thinking about changing jobs. This is how it usually goes.
Option 1: Quit and leap.Option 2: Stay and suffer.
Now try the Rule of 3.
Path B: Quit and take a new role in a similar field.Path C: Stay for six months and skill up aggressively.Path D: Go part-time or freelance while testing something new.
Of course, none of these options is perfect. Thats why the next stage of the process is even more important: the consequences.
1st, 2nd and 3rd order effects
It simply means keep asking, and then what? First-order effects are immediate. What happens right away when you make the decision? Second-order effects come next. What does that lead to? Third-order effects are longer-term. Who do you become if this path continues?
I will now apply the effects to the job-changing example.
Path B: Quit and take a similar role.
First-order: New environment. Relief. You may stop dreading Mondays.
Second-order: You become more confident. Now, you know youre employable. You can actually change jobs.
Third-order: You might stay on the same path longer than you want.
Now Path C: Stay and upgrade your skills
First-order: You may feel frustrated for a while. You will need a lot of discipline for this path.
Second-order: You will get leverage to open your options.
Third-order: You redefine yourself from stuck to building a career. You may become indispensable to your employer.
The mistake most people make
Most people pursue the best outcome. Thats a trap. The future is uncertain. Youre probably guessing what could work. Everyone is. Once you are done with the effects, choose the path with the least life-altering effects. The one that teaches you something. Keeps doors open. And doesnt completely make your life worse if youre wrong.
Its my risk psychology approach.
People regret irreversible decisions more than bad ones. We hate closing doors we didnt mean to close. Thats why picking the path that means a lot to you but wont burn bridges matters.
Make better decisions with the least panic.
This framework works when you are emotionally attached to the decision you are about to make. When youre stressed, your brain throws logic out of the window. The rule of 3 gets you back on the rational path. It takes you from reacting to responding to life. It helps you answer the most important question. Which future can I live with?
You can use this rule anywhere. Money decisions. Relationship decisions. Creative decisions. A big purchase. Even small ones. Do I say yes to this commitment? What are the effects, and what are my options? And what path can I live with and still function? Force the three paths.
Pursue the consequences in places most people ignore. Then, opt for the choice that makes life better without disrupting your entire life.
Use it to pick a path with tolerable unknowns
The rule of three doesnt remove uncertainty. Nothing does. Youre never picking certainty. Youre picking a path with tolerable unknowns. Good decisions come from better processes. The 3 rule takes away the emotional attachment that drains the life out of you.
Most of our hard decisions become unbearable because we want a perfect choice. The one that proves we are smart and avoids regret. So you panic. Or overthink. Some people let time decide for them.
Which is still a decision, by the way.
I use the rule of three to pick a direction.
Adjust where necessary. And keep moving. I want forward motion without self-destruction. You dont need to outsmart the future. Just stop putting so much pressure on yourself. Most choices dont need courage. They need structure. Three paths. Three consequences. It makes overthinking your options almost impossible.
AI is no longer just a cascade of algorithms trained on massive amounts of data. It has become a physical and infrastructural phenomenon, one whose future will be determined not by breakthroughs in benchmarks, but by the hard realities of power, geography, regulation, and the very nature of intelligence. Businesses that fail to see this will be blindsided.
Data centers were once the sterile backrooms of the internet: important, but invisible. Today, they are the beating heart of generative AI, the physical engines that make large language models (LLMs) possible. But what if these engines, and the models they power, are hitting limitations that cant be solved with more capital, more data centers, or more powerful chips?
In 2025 and into 2026, communities around the U.S. have been pushing back against new data center construction. In Springfield, Ohio; Loudoun County, Virginia and elsewhere, local residents and officials have balked at the idea of massive facilities drawing enormous amounts of electricity, disrupting neighborhoods, and straining already stretched electrical grids. These conflicts are not isolated. They are a signal, a structural friction point in the expansion of the AI economy.
At the same time, utilities are warning of a looming collision between AIs energy appetite and the cost of power infrastructure. Several states are considering higher utility rates for data-intensive operations, arguing that the massive energy consumption of AI data centers is reshaping the economics of electricity distribution, often at the expense of everyday consumers.
This friction between local resistance to data centers, the energy grids physical limits, and the political pressures on utilities is more than a planning dispute. It reveals a deeper truth: AIs most serious constraint is not algorithmic ingenuity, but physical reality.
When reality intrudes on the AI dream
For years, the dominant narrative in technology has been that more data and bigger models equal better intelligence. The logic has been seductive: scale up the training data, scale up compute power, and intelligence will emerge. But this logic assumes that three things are true:
Data can always be collected and processed at scale.
Data centers can be built wherever they are needed.
Language-based models can serve as proxies for understanding the world.
The first assumption is faltering. The second is meeting political and physical resistance. The third, that language alone can model reality, is quietly unraveling.
Large language models are trained on massive corpora of human text. But that text is not a transparent reflection of reality: It is a distillation of perceptions, biases, omissions, and misinterpretations filtered through the human use of language. Some of that is useful. Much of it is partial, anecdotal, or flat-out wrong. As these models grow, their training data becomes the lens through which they interpret the world. But that lens is inherently flawed.
This matters because language is not reality: It is a representation of individual and collective narratives. A language model learns the distribution of language, not the causal structure of events, not the physics of the world, not the sensory richness of lived experience. This limitation will come home to roost as AI is pushed into domains where contextual understanding of the world, not just text patterns, is essential for performance, safety, and real-world utility.
A structural crisis in the making
We are approaching a strange paradox: The very success of language-based AI is leading to its structural obsolescence.
As organizations invest billions in generative AI infrastructure, they are doing so on the assumption that bigger models, more parameters, and larger datasets will continue to yield better results. But that assumption is at odds with three emerging limits:
Energy and location constraints: As data centers face community resistance and grid limits, the expansion of AI compute capacity will slow, especially in regions without surplus power and strong planning systems.
Regulatory friction: States and countries will increasingly regulate electricity usage, data center emissions, and land use, placing new costs and hurdles on AI infrastructure.
Cognitive limitations of LLMs: Models that are trained only on text are hitting a ceiling on true understanding. The next real breakthroughs in AI will require models that learn from richer, multimodal interactions from real environments, sensory data and structured causal feedback, not just text corpora. Language alone will not unlock deeper machine understanding.
This is not a speculative concern. We see it in the inconsistencies of todays LLMs: confident in their errors, anchored in old data, and unable to reason about the physical or causal aspects of reality. These are not bugs: they are structural constraints.
Why this matters for business strategy
CEOs and leaders who continue to equate AI leadership with bigger models and more data center capacity are making a fundamental strategic error. The future of AI will not be defined by how much computing power you have, but by how well you integrate intelligence with the physical world.
Industries like robotics, autonomous vehicles, medical diagnosis, climate modeling, and industrial automation demand models that can reason about causality, sense environments, and learn from experience, not just from language patterns. The winners in these domains will be those who invest in hybrid systems that combine language with perception, embodiment, and grounded interaction.
Conclusion: reality bites back
The narrative that AI is an infinite frontier has been convenient for investors, journalists, and technologists alike. But like all powerful narratives, it eventually encounters the hard wall of reality. Data centers are running into political and energy limits. Language-only models are showing their boundaries. And the assumption that scale solves all problems is shaking at its foundations.
The next chapter of AI will not be about who builds the biggest model. It will be about who understands the world in all its physical, causal, and embodied complexity, and builds systems that are grounded in reality.
Innovation in AI will increasingly be measured not by the size of the data center or the number of parameters, but by how well machines perceive, interact with, and reason about the actual world.
In the world of social impact and sustainability, 2025s word of the year could have been headwinds. It became a euphemism for everything from political pressure and regulatory changes to economic uncertainty, AI disruption, and social upheaval.
But in many ways, headwinds is an understatement for what impact and sustainability leaders across the corporate and nonprofit sectors navigated in a year of budget cuts and evolving risk factors. For much of the past year, leaders across the corporate and nonprofit sectors have been recalibrating approaches to advancing their missions against these trends. In 2026, well start to see those new approaches in action.
Based on interviews with dozens of experts, here are five big shifts to pay attention to over the next year in social impact and sustainability.
1: Evolving the sustainability narrative
One of the most visible shifts to note is that social impact and sustainability are becoming much less, well, visible. For years, companies have been making bold commitments, setting lofty goals, and engaging in the kind of storytellingbut not always following through, a trend that finally led Merriam-Webster to add greenwashing to its dictionary in 2022.
2025 felt like a correction, as companies reacting to a changing landscape of risk and political attention ushered in a period of greenhushing, where companies were reluctant to talk about their sustainability initiatives. As Andrew Winston of Winston Eco-Strategies puts it, “The biggest issue in the U.S. is the very strong desire of leadership teams to keep their heads down and say nearly nothing about sustainability. The work seems to be mostly continuing, but it’s certainly not great for morale or moving at speed and scale if your bosses are telling you to hide out.”
Thats why 2026 is likely to bring another narrative correction that grounds sustainability storytelling in business performance and operational rigorwhich has always been where sustainability is heading. The best companies arent just making pledges, theyre building and executing solutions that scale, measure, and return value, says Dave Stangis at Apollo. Seeing capital, innovation, and outcomes align always gives me optimism.
2: Adopting a new leadership mindset
An organization laser-focused on delivering results also requires a laser focus from its leaders. As Alison Taylor of Ethical Systems notes, the rapid-fire disruption of 2025 made this focus hard to find: Many of sustainability’s core assumptions no longer apply, and there is a need for a reframe of the profession. The practitioners I talk to are struggling with terminology, legal risk, and threats to their roles. While it is true that much great work is going on behind the scenes, it is difficult for most leaders I speak to to maintain organizational momentum, simply because there is so much fire fighting to do.
2026 will bring new fires to fight, but the demand for results and focus will give rise to a new mindset for leaders. Kristen Titus of the Titus Group predicts that leaders will emerge from this period of uncertainty and paralysis with a renewed willingness to engage: Clients, customers, and employees are hungry for engagementand they’re craving moral leadership. Those that step forward with clarity and courage will help define the next chapter of impact and sustainability.
3: Aligning rapid response with long-term goals
One strategy that helps impact leaders maintain their focus involves finding ways to connect their communities immediate needs with long-term business strategy.
Uncertainty demands agility, as Laura Turner, VP and Head of Community Impact at TIAA points out: Most companies hold flexible funding that can be adapted for unexpected needs. When the government shutdown hit, TIAAs first-generation college student program pivoted quickly, redirecting funds to local food banks. That flexibility isn’t just nice to have anymore, it’s essential for navigating uncertainty.
For many organizations, balancing immediate and long-term needs also means AI-proofing their impact strategy. Royal Bank of Canada, for example, is leveraging business expertise and resources around AI adoption to support nonprofit partners in keeping pace with innovation. There is a broad consensus that AI and digital innovation can drive the biggest economic transformation in a generation. And yet, at this very same moment, the non-profit sector faces unprecedented strain and ongoing barriers to funding and technical training. Without intentional support, the sector risks falling behind. said Kara Gustafson, President of the RBC Foundation USA.
4: Putting well-being first
All of this uncertainty and disruption has taken a toll on professionals in this space in 2025. In 2026, well-being will become a core function of impact strategyboth as a response to workforce and community needs.
Haviland Sharvit, Executive Director of Susan Crown Exchange (Susan Crown Exchange and TIAA, above, are clients of mine), predicts that more companies and nonprofits will meet the moment with an impact strategy focused on youth well-being in the age of AI: Rapid advances in technology and AI offer powerful opportunities for learning and connection. Yet impact leaders face rising youth mental health strain, widening digital inequities, advancements that have outpaced youth protections, and the erosion of real human connection. Well see a shift toward promoting and safeguarding youth wellbeing in an AI-driven world, more attention on responsible tech, and greater investment in human connection.
5: Investing in community
Amid all of this disruption, we asked leaders what gives them hope, and a common refrain emerged: we find hope in each other.
Community is, and will continue to be, everything. In real and virtual rooms all over the countryand across impact networks like Trellis, UN Global Compact, NationSwell and many moreleaders spent 2025 reflecting, commiserating, and charting a new course forward.
The last prediction Ill offer is one of my own: impact networks and convening spaces will grow rapidly in 2026, as new communities of practice emerge and existing communities grow. With all of the growth and learning 2026 has in store, finding safe spaces for reflection, knowledge sharing, and collaboration is a top priority for impact leaders.
Severance is the hit sci-fi show about office workers who sever their consciousnessslipping into another mode the moment they arrive at the office, then forgetting everything about their 9-to-5 as soon as they leave. The concept was inspired by the creators own monotonous desk job before he found success in television. Part of the shows appeal lies in how familiar the premise feels: a dull, repetitive workday that people cant wait to escape.
In the real world, employees dont have a mental switch to flip, but theyve found subtler, and potentially more insidious, ways to disengage. The latest trend, dubbed task-masking, has taken over Instagram and TikTok. Its all about looking busy without actually being productive: charging around the office with a laptop, pretending to be on an urgent call, or typing furiously with no real purpose. According to recent research, more than a third of U.K. workers admit to faking productivity.
Task-masking doesnt just waste timeit slows career growth and hurts company performance. Employees miss out on meaningful progress and promotion opportunities. Leaders lose engagement and confidence in their teams.
In short, task-masking is a problem no leader can afford to ignore. Here are some strategies to stop it.
Be clear on the companys values
Task-masking isnt born of laziness or lack of ambitionits a fear-based response to productivity pressure and always-on work cultures. Research from Workhuman found that strict time-tracking exacerbates the problem: When workers strongly agree they are expected to respond immediately to Slack, Teams, or other instant messages, the rate of fake productivity shoots up to 51%.
To free employees from the sense that their time card matters most, leaders should clarify what the company truly values. Face time or hours logged at a desk shouldnt be measures of successmeaningful productivity should. What that looks like will vary by organization, but at Jotform, for example, it means advancing projects and meeting reasonable deadlines. It also includes less-measurable but equally valuable behaviors like showing curiosity, supporting teammates, and helping create a more engaged work culture.
Leaders should also be explicit about what doesnt count: busywork, unnecessary meetings that could be handled asynchronously, and burning the midnight oil just to give the impression of busyness and commitment.
Break down projects into more manageable tasks
As AI and automation boost efficiency and productivity, theyve fundamentally transformed workloads. In many ways, thats a positive change. Employees can devote more time to meaningful, higher-impact work. For example, you can spend more time on strategizing and creative writing, and fewer hours sifting through your inbox and searching through meeting notes. But it also brings a challenge: When technology accelerates what you can accomplish in a day, leaders expectations often rise in tandem. The slope to burnout becomes slippery.
One of the best antidotes to that pressure, especially when facing large, intimidating projects that can leave employees feeling paralyzed or faking productivity, is to break them into smaller tasks. For starters, this helps people identify steps that can be automated, eliminated, or delegated. It also makes progress more tangible. Ticking off one item at a time, with restorative breaks in between, keeps momentum steady.
When a daunting to-do list is broken down into a sequence of manageable tasks, employees can work efficiently and stay on track toward deadlines without burning out.
Make psychological safety a priority
If task-masking is rooted in fear, a quick fix wont eliminate it. Economic downturns, global pandemics, and rapid technological change have all contributed to a heightened sense of workplace anxiety, especially among the younger generations. More than one-third (37%) of Gen Z workers fear losing their jobsmore than any other generationaccording to research from Edelmans Gen Z Lab. Creating an environment where psychological safety is a priority can help assuage career-related fears and the pressure to appear productive all the time. When employees feel safe admitting theyre stuck or uncertain, theyre less likely to mask their struggles with performative busyness.
At Jotform, we have multiple channels where employees can voice their concerns, ranging from all-hands meetings and dedicated chat threads to a general management open-door policy. I make a point to share the challenges Im facing, too, in hopes that my candor will encourage others to speak openly about their own doubts and setbacks.
Ultimately, leaders must be explicit about the resources available to support employees and model the transparency they want to see. A bit of vulnerability from the top can help promote psychological safety throughout an organization. Employees shouldnt fear work so much that they want to escape itthrough severance or through task-masking.
While traveling to Riyadh for the Fortune Global Forum, FII9, and the Global Health Exhibition, I witnessed something that should be a wake-up call for health systems everywhere. Saudi Arabia is already operating the kind of connected, AI-enabled healthcare infrastructure many countries are still debating how to build. At FII9, the conversation was unmistakable. Global innovation momentum is shifting toward the Middle East, and nowhere more than Saudi Arabia, where national digital platforms like Sehhaty already give millions of residents unified access to their health data. At the Global Health Exhibition, I saw population-level analytics, AI-powered diagnostics, multiomic initiatives, and interoperable infrastructure deployed at a speed and scale that would take years in other countries. It made something clear: Healthcare does not have a data problem. We have a connection problem.
LIFESTYLE DRIVES OUTCOMES, BUT REMAINS CLINICALLY INVISIBLE
Studies show that lifestyle and environmental factors account for more than 80% of health outcomes. A healthy lifestyle can prevent the vast majority of chronic diseases, including heart disease and diabetes. And yet, the data that reflects how people actually live, how they eat, move, sleep, and manage stress, remains largely absent from clinical care.
In the United States, for example, healthcare is not suffering from a data shortage. Its drowning in data. Every day, people generate powerful information through wearables, continuous glucose monitors, fitness and sleep apps, and smart rings. As of 2023, nearly one in three Americans use a wearable to track their health, according to a Health Information National Trends Survey. These tools capture meaningful lifestyle signals that directly affect clinical outcomes. Yet almost none of this data reaches the exam room. It remains siloed on consumer platforms, invisible to clinicians, and unusable in medical decision making.
This disconnect has consequences. Preventive opportunities are missed. Chronic conditions go unmanaged. Healthcare remains reactive instead of proactive. Clinicians rely on structured snapshots like lab results and prescriptions, important, but incomplete, because they capture what happens in the clinic rather than daily life.
AI AS THE BRIDGE BETWEEN LIFESTYLE DATA AND CLINICAL CARE
Its not that clinicians are uninterested in wearable data. Many are eager. Remote patient monitoring has grown rapidly, with a 1,300% increase in related procedures between 2019 and 2022. However, the friction points are real. Data security is a concern. Device accuracy varies. Practices often lack the IT infrastructure to onboard new tools, train staff, and integrate multiple data streams. Most importantly, clinicians are overwhelmed. More raw data is not the solution.
This is why healthcare systems need a bridge that makes lifestyle data usable, reliable, and safe in clinical settings. That bridge is clinical-grade AI. When lifestyle and longitudinal behavioral datasets are used at inference time via retrieval, AIs outputs are grounded in real-world signals rather than abstract reasoning alone, distilling only the most relevant insights for the point of care. The goal is not another dashboard, but meaningful signals embedded within existing workflows that reduce burden rather than increase it. With the right tools, AI also empowers patients. Personalized, real-time guidance rooted in their own physiology helps them understand their data, make better decisions, and stay aligned with their care plan.
A NATIONAL DIGITAL HEALTH INFRASTRUCTURE
Countries such as Saudi Arabia are demonstrating what happens when clinical and systems-level data come together. The unified national health platform Sehhaty serves as an access point for millions of residents and offers integrated services far beyond scheduling. These include secure medical records, online prescriptions, lab results, vaccination history, teleconsultations, and remote monitoring. The app reportedly contains 31 million unified health files, representing nearly 88% of the population, and 140 million online prescriptions.
At the center of this transformation is the Seha Virtual Hospital, which delivers remote specialist care across 224 hospitals and dozens of specialties, including critical-care consults and AI-driven diagnostics. Investments in genomics, proteomics, metabolomics, and advanced AI at institutions such as King Faisal Specialist Hospital rival those of some of the best programs in the United States. The result is a coordinated model of nationwide digital health integration, something long envisioned but not yet achieved.
TO BUILD THE FUTURE OF HEALTH, CONNECT THE DOTS
The United States has the devices and data, but has long lacked the infrastructure and incentives to connect them meaningfully. That is beginning to shift. Recent CMS initiatives signal recognition that prevention, lifestyle data, and technology-enabled care must play a larger role in how health outcomes are measured and reimbursed. Initiatives such as MAHA ELEVATE and the CMS ACCESS Model reflect a growing shift toward prevention-first, lifestyle-driven care. MAHA ELEVATE supports Medicare pilot programs that test whether whole-person, lifestyle-based care can improve health outcomes and lower costs, while ACCESS helps bring these approaches to scale through new care delivery and outcome-based payment models. In parallel, CMS Aligned Networks is focusing on improving interoperability and coordination across the healthcare ecosystem, creating standards and incentives that allow data to move safely between patients, providers, and care teams. The opportunity is to ensure that lifestyle data are treated as essential clinical information and that AI translates complexity into actionable insight at the point of care.
The most valuable health data we possess is already being captured on our wrists, in our pockets, and throughout our daily routines. The challenge is no longer collection. Its connection. To close the gap, we must treat lifestyle data as essential clinical information and not a consumer novelty. Interoperable systems must allow this information to move securely to the right stakeholders, with AI surfacing timely, relevant signals that support decision making without adding friction for clinicians or patients.
Only then can healthcare move from fragmented snapshots to continuous understanding, from episodic and reactive care to a model that anticipates risk, promotes healthy behaviors, and supports the whole person. The future of healthcare is already taking shape in places such as Riyadh, where vision, infrastructure, and execution are aligned. Other countres, including the United States, can get there too, but only if we connect the dots.
Noosheen Hashemi is founder and CEO of January AI.
Scott Adams, the creator of the uber-popular and satirical comic strip Dilbert, has died. He passed away on January 13, after announcing his diagnosis of metastatic prostate cancer last spring. He was 68.
On Tuesday morning, the cartoonists former wife, Shelly Miles, shared the news of his death during a livestream on X. Miles read from a statement that Adams had prepared himself for the occasion.
I had an amazing life, the statement said. I gave it everything I had. If you got any benefits from my work, I’m asking you to pay it forward as best you can. That is the legacy I want. Be useful. And please know I loved you all to the very end.
Dilbert was created in 1989, and it broke new ground, offering a refreshing and pointed critique of white-collar work life. It became known for its ever-relatable digs about the drudgery of office culture and insufferable bosses, long preceding relatable movies and TV shows like Office Space and The Office, which featured similar dismal (and hilarious) views of work culture years later.In its heyday, the comic strip appeared in over 2,000 newspapers worldwide, with an estimated readership of more than 150 million. Adams’s strip amassed such popularity that he was named the 1997 recipient of the National Cartoonist Society’s Reuben Award. That same year, Dilbert (the character) became the first fictional person to make Time magazine’s list of the most influential Americans.However, while Dilbert became one of the most popular cartoons of all time, Adams battled deep controversy in his later years. In 2023, hundreds of newspapers dropped the classic comic after Adams made racist comments on his podcast, saying that it no longer makes any sense as a white citizen of America to try to help Black citizens anymore. He also described Black people as a “hate group.”Adams said his statements were taken out of context. Still, the incident, and its aftermath, effectively ended Dilbert‘s wide syndication in newspapers. Other comics weighed in, too. “He’s not being canceled. He’s experiencing the consequences of expressing his views,” Bill Holbrook, creator of the strip On the Fastrack, told The Associated Press at the time. “I am in full support of him saying anything he wants to, but then he has to own the consequences of saying them.”Regardless of Adamss troubling personal views and complicated legacy, Dilbert has played a large role in the conversation around work life. Experts say that his cartoons’ outspoken critiques of bosses and work life, which were perhaps ahead of their time, can’t be rolled back.
Phil Lohmeyer, a cartoonist, animator, and middle school design teacher from Connecticut, tells Fast Company that hes confident the kind of office critiques made popular by Dilbert will live on because they are so universal. Dilbert wasn’t as much about the characters, even though the characters themselves became famous. It was more about the annoyance of middle management, he notes.
Lohmeyer says that the idea truly resonated with office workers, who posted the comics in their cubicles in the 90s, or emailed them to coworkers. While younger generations might not be well-versed in Dilbert, the teacher still sees the ideas show up in his middle school classroom. The kids make fun of the rules, schedules, and more, he says. They use comic strip humor to question the system, kind of how Adams was doing years ago.
While so much has changed in offices and classrooms alike, Lohmeyer says that feeling seen in your role will forever be relevant. Adams turned work issues into cartoon gags, making the previously invisible finally visible.
Iranian demonstrators’ ability to get details of bloody nationwide protests out to the world has been given a strong boost, with SpaceX’s Starlink satellite internet service dropping its fees to allow more people to circumvent the Tehran government’s strongest attempt ever to prevent information from spilling outside its borders, activists said Wednesday.
The move by the American aerospace company run by Elon Musk follows the complete shutdown of telecommunications and internet access to Iran’s 85 million people on Jan. 8, as protests expanded over the Islamic Republic’s faltering economy and the collapse of its currency.
SpaceX has not officially announced the decision and did not respond to a request for comment, but activists told The Associated Press that Starlink has been available for free to anyone in Iran with the receivers since Tuesday.
Starlink has been crucial, said Mehdi Yahyanejad, an Iranian whose nonprofit Net Freedom Pioneers has helped smuggle units into Iran, pointing to video that emerged Sunday showing rows of bodies at a forensic medical center near Tehran.
That showed a few hundred bodies on the ground, that came out because of Starlink,” he said in an interview from Los Angeles. “I think that those videos from the center pretty much changed everyone’s understanding of what’s happening because they saw it with their own eyes.
Since the outbreak of demonstrations Dec. 28, the death toll has risen to more than 2,500 people, primarily protesters but also security personnel, according to the U.S.-based Human Rights Activists News Agency.
Starlink is banned in Iran by telecommunication regulations, as the country never authorized the importation, sale, or use of the devices. Activists fear they could be accused of helping the U.S. or Israel by using Starlink and charged with espionage, which can carry the death penalty.
Cat-and-mouse as authorities hunt for Starlink devices
The first units were smuggled into Iran in 2022 during protests over the country’s mandatory headscarf law, after Musk got the Biden administration to exempt the Starlink service from Iran sanctions.
Since then, more than 50,000 units are estimated to have been sneaked in, with people going to great lengths to conceal them, using virtual private networks while on the system to hide IP addresses and taking other precautions, said Ahmad Ahmadian, the executive director of Holistic Resilience, a Los Angeles-based organization that was responsible for getting some of the first Starlink units into Iran.
Starlink is a global internet network that relies on some 10,000 satellites orbiting Earth. Subscribers need to have equipment, including an antenna that requires a line of sight to the satellite, so must be deployed in the open, where it could be spotted by authorities. Many Iranians disguise them as solar panels, Ahmadian said.
After efforts to shut down communications during the 12-day war with Israel in June proved to be not terribly effective, Iranian security services have taken more extreme tactics now to jam Starlink’s radio signals and GPS systems, Ahmadian said in a phone interview. After Holistic Resilience passed on reports to SpaceX, Ahmadian said, the company pushed a firmware update that helped circumvent the new countermeasures.
Security services also rely on informers to tell them who might be using Starlink, and search internet and social media traffic for signs it has been used. There have been reports they have raided apartments with satellite dishes.
There has always been a cat-and-mouse game, said Ahmadian, who fled Iran in 2012 after serving time in prison for student activism. The government is using every tool in its toolbox.
Still, Ahmadian noted that the government jamming attempts had only been effective in certain urban areas, suggesting that security services lack the resources to block Starlink more broadly.
A free Starlink could increase the flow of information out of Iran
Iran did begin to allow people to call out internationally on Tuesday via mobile phones, but calls from outside the country into Iran remain blocked.
Compared to protests in 2019, when lesser measures by the government were able to effectively stifle information reaching the rest of the world for more than a week, Ahmadian said the proliferation of Starlink has made it impossible to prevent communications. He said the flow could increase now that the service has been made free.
This time around they really shut it down, even fixed landlines were not working, he said. But despite this, the information was coming out, and it also shows how distributed this community of Starlink users is in the country.
Musk has made Starlink free for use during several natural disasters, and Ukraine has relied heavily on the service since Russia’s full-scale invasion in 2022. It was initially funded by SpaceX and later through an American government contract.
Musk had raised concerns over the power of such a system being in the hands of one person, after he refused to extend Ukraine’s Starlink coverage to support a planned Ukrainian counterattack in Russian-occupied Crimea.
As a proponent of Starlink for Iran, Ahmadian said the Crimea decision was a wake-up call for him, but that he couldn’t see any reason why Musk might be inclined to act similarly in Iran.
Looking at the political Elon, I think he would have more interest … in a free Iran as a new market, he said.
Julia Voo, who heads the International Institute for Strategic Studies’ Cyber Power and Future Conflict Program in Singapore, said there is a risk in becoming reliant on one company as a lifeline, as it creates a single point of failure, though currently there are no comparable alternatives.
China has been exploring ways to hunt and destroy Starlink satellites, and Voo said the more effective Starlink proves itself at penetrating government-mandated terrestrial blackouts, the more states will be observing.
It’s just going to result in more efforts to broaden controls over various ways of communication, for those in Iran and everywhere else watching, she said.
David Rising, Associated Press
Associated Press writers Jon Gambrell and Melanie Lidman contributed to this report.
