If you hear your organization talking about the Great ShakeOut, it has nothing to do with Taylor Swift or Florence and the Machine.
Instead, this international event promotes earthquake preparedness. Having a plan greatly improves outcomes and saves lives.
On October 16 at 10:16 a.m. local time, millions will be practicing how to properly drop, cover, and hold on. Lets take a look at the science behind earthquakes, the regions they impact, and how to participate in the Great ShakeOut.
What actually causes an earthquake?
The Earths outer layer is made up of seven major tectonic plates. Think of these as patches of a quilt that isnt stitched together perfectly.
The places where the plates meet are called plate boundaries. Some of these contain fault lines. The patches or plates move since they are not properly connected, which causes stress to build up at the borders.
When this reaches a boiling point, the pressure is released, causing the earth to shake.
Which regions have the greatest earthquake risk?
According to the United States Geological Survey, 81% of earthquakes take place along the Circum-Pacific seismic belt, which is located on the rim of the Pacific Ocean. This shaky area is also known as the Ring of Fire because of its plethora of volcanoes.Because of Japans advanced ability to detect earthquakes, it holds the dubious honor of having the most recorded quakes, though the USGS says its more likely that Indonesia experiences the most quakes annually by virtue of its larger size. The most catastrophic earthquakes have tended to occur in China, Iran, and Turkey.
How can I participate?
Organizations and individuals are welcome to participate in the Great ShakeOut. You can even make it a family affair.
Register at shakeout.org to make sure your efforts are counted. There, youll find resources such as a drill narration and discussion questions for a post-event debrief.
There are also steps to take to be ready for the big one. These include making sure furniture and decorative items are secure, having a disaster plan, and keeping emergency supply kits stocked and up to date with all the necessary items.
The whole idea of advertisingusing pictures and words to get people to buy stuff, or to do somethingis old indeed, with the first known example dating back almost 5,000 years to the heady days of Ancient Egypt.
The ads business changed a lot since we were writing notices on papyrus, but one thing thatuntil recentlyremained the same was that it was a deeply intentional business. The advertiser had to think about the language they used, the imagery they employed, the types of people they sought to reach, and how they would go about doing that.
Whether the advertiser was touting a weaving shop on the banks of the Nile during the days of the Pharaohs, or selling detergent or cigarettes through new mass media innovations like the television or the radio, that same thoughtfulness was a constant. Advertisers had to thinkand, by virtue of the fact they were forced to make decisions, they were in control of everything.
My biggest complaint with the digital ads ecosystem is it, by design, strips the ability of the advertiser to make some of those decisionsnot merely placement, but targeting, and with the emergence of dynamic creative and generative AI, messaging too.
In the process, weve turned advertising from a very deliberately engineered systemwhere the architect knows what each part of the process should dointo one thats, essentially, a black box.
And within this black box, theres little room for creativity.
The Process Is Creative
When we think about advertisingand, in particular, good advertisingwe always think about the messaging. Its true that some of the best campaigns in history have always used clever wordplay, or coy psychological tactics, to drive a point home.
The Pepsi Challenge, for example, started off as a series of in-person taste tests and culminated in a campaign that could confidently say (though some have identified flaws in the test itself) that Pepsi was Americas preferred cola. Not only did this directly undercut Pepsis main adversaryCoca-Cola, which easily had the most powerful brand perceptionbut it also allowed people to differentiate between products that people might otherwise think of as identical.
Messaging is important, but its far from the only creative part of the marketing process.
Take Subaru, for example. In the 1990s, it was a struggling car brandeclipsed not only by its Japanese rivals like Nissan and Toyota, but also by fierce domestic competition in the U.S. market.
Subaru hired a new advertising firm to turn its fortunes around, which ran a series of focus groups that asked why existing Subaru owners chose its vehicles, as opposed to those from one of its healthier rivals. That firm noticed that women dominated those focus groups, and many of those women identified as lesbian.
The company then launched a campaign that targeted both women and lesbiansitself a brave choice, considering the climate of the 1990s, which saw the passage of both the Defense of Marriage Act and Dont Ask, Dont Tell. To help it reach lesbian audiences, it hired Mulryan/Nash to create content specifically for the gay press, and to handle ad buying.
This campaign wasnt just pioneeringit also, arguably, helped revive Subarus fortunes, and the brand remains vibrant and relevant, especially in the U.S., where it sold over 667,000 cars in the 2024 calendar year.
The Subaru example is a potent one, not simply because it was so successful, but because it illustrates how each step of the processfrom identifying the customer, to determining where to reach them, to crafting the messagingrequired human thought and human creativity.
If were looking for a more contemporary example, Spotifys controversial Thanks 2016, its been weird springs to mind. Capitalizing on a year defined by seismic political shifts, celebrity deaths, and countless surreal moments to mention in the confines of this piece, Spotify tapped into its data, identifying equally surreal trends and turning them into highly relatable billboards positioned in prime urban locations.
These billboards featured pithy one-liners (for example, Dear person who played ‘Sorry’ 42 times on Valentines Day, what did you do?), with the text localized for target markets (Dear 3,749 people who streamed Its the end of the world as we know it the day of the Brexit vote, hang in there). It served as a reminder of how music isnt simply a form of entertainment, but a way in which we process events in our personal lives, as well as those happening within politics and culture.
Again, this process required creative thinking at every levelfrom identifying the patterns within the data that would lead to the funniest trends, to choosing the most valuable locations to place the billboards.
I write all of this not because I believe that all digital advertisingwhere these decisions are outsourced, particularly to third-partiesis bad, but because I believe that the most effective and memorable campaigns are thoughtful ones.
The reason why I believe digital advertising is the enemy to advertising is because, by design, it strips us of the ability to use that creativity across all stages of the advertising process, from conceptualization to creating the final product.
Battling the Black Boxes
Last year was the 30th birthday of digital advertising. Its interesting to see how, as the internet grew and an adtech ecosystem emerged, the very nature of how this segment actually works changed.
Whereas at one point advertising deals were inked between companies, with money changing hands in exchange for prime placement for a set number of days, those manual transactions are now a thing of the past. Todays digital advertising mechanics are based on systems which the advertiser doesnt control or even understandand in the case of those which heavily rely on AI, even the developers dont have full insight into the factors behind each targeting and placement decision.
This opacity also allows the adtech provider or advertising network to act in ways that are contrary to the interests of the advertisereither by obfuscating data that could allow them to make more effective decisions, or by failing to protect said advertiser from, for example, click fraud.
Although digital ads allow a company to target and market at scaleand, arguably, with the economies of scale that wouldnt be otherwise possiblethe downside is, arguably, a degradation of the online experience for end users, profound concerns about user privacy, and an absence of transparency for those actually purchasing the ads.
Arguably, the biggest downsidefrom someone who cares profoundly about the intellectual and creative brilliance of the ads industryis that digital ads havent really produced something thats memorable, or has had any meaningful cultural impact.
Coca-Cola gave us Santa Clauss red outfit and the iconic flashing delivery trucks. Decades after tey commercials first aired, we still remember the Budweiser frogs croaking bud-wise-er, or its later ads that turned wazzup into a legitimate pop culture phenomenon (albeit a really irritating one).
And thats because creativity is like a muscle, and if you dont exercise itor dont have to exercise ititll wither away.
The Interborough Express linethe long-awaited light-rail link between Brooklyn and Queens in New York Cityhasnt broken ground yet. But on my computer screen, one part of the route is already operational.
A new simulation game called Subway Builder lets you design, build, and operate subway systems in 26 U.S. cities, from New York to Boston to San Francisco. The game uses real-life U.S. Census Bureau and employment data to map where residents and workers live, allowing you to simulate realistic passenger flows. Players must also contend with real-world constraints like tunnels, viaducts, existing foundations, and road layouts.
[Image: courtesy Colin Miller]
The goal is to design a subway network that gets the most people to their destinations as quickly as possible. But theres a deeper ambition: to spark more transit-minded thinking in a country that historically has underinvested in it.
I would secretly hope that maybe someone in power sees this and says Maybe we can build something like this, says Colin Miller, a software engineer and creator of the game.
[Image: courtesy Colin Miller]
Building a hyperrealistic transit game
Subway Builder launched on October 9 to much fanfare in the transit community. I’ve been playing Subway Builder for *checks notes* all night,” one user posted. I legitimately think this game is going to start a transit revolution in America, wrote another.
Over the years, many developers have tried to gamify transit design with offerings like MetroConnect, Brand New Subway, and Mini Metro. But few have attempted to make their simulations realistic enough to replicate real transit-planning challenges at the scale of cities like New York or Seattle.
To create Subway Builder, Miller drew on datasets from the U.S. Department of Education, the Federal Aviation Administration, and OpenStreetMap, among others. You can analyze demand statistics on a map of your chosen city; then, once you build a route, explore ridership station by station. One of the most satisfying features for me remains the constellation of red dots that represent riders commuting toward newly built stations and journeying across a network I just built.
[Image: courtesy Colin Miller]
The cost of building public transit in the U.S.
Subway Builder bills itself as hyperrealistic, but there are two key exceptions: politics and budgeting. Miller says he did not take into account the political will in any given U.S. city, nor did he calibrate the games budget to U.S. infrastructure costs. Instead, he used Spanish construction costs, which are among the lowest in the world. (Madrid, for example, tripled its metro network in just 12 years.) If I had it set to realistic American construction prices, it would have made the game unplayable because youd run out of money, he concedes.
Players can choose to play in sandbox mode, which comes with no budgetary constraints. Its the games normal mode that reveals a painful fact long criticized by expertsnamely that building transit in the U.S. is mind-bogglingly expensive. On average, domestic rail transit projects cost roughly twice as much per mile in the U.S. as they do in Europe or Canada, and as much as five times more in New York City. The relatively recent Second Avenue Subway expansion, for example, cost about $2.5 billion per mile. For reference, the Los Angeles Purple Line extension cost $800 million per mile, while Madrids extension was $320 million per mile.
When I played the game, I quickly learned that even $3 billion would get me only three lines and about 20 stations in Brooklyn. I also learned that building a subway route is just the beginning of a long road plagued by never-ending signal failures, broken-down trains, and overall operational costs. And considering that every dollar collected from fares helps fund new routes and buy new trains, I gained a bit more sympathy for the MTAs recent war on fare evasion.
[Image: courtesy Colin Miller]
A tool for publication imagination
Much ink has already been spilled on the state of mass transit in the United States. Transit advocates such as Yonah Freemark have frequently lamented declining ridership and funding shortfalls in American cities. Others, like Brent Toderian, have emphasized the role of transit in shaping equitable, walkable urban environments. While public transit has recently blossomed in many U.S. cities, the system remains plagued by some of the worlds highest construction costs, red tape, political fragmentation, and a misguided adulation for the freedom that cars provide for the benefit of the few at the expense of the many.
Hayden Clarkin, a transport engineer and planner who recently published a Hitchhikers Guide to Building a Lot of Subways, argues the U.S. has the ability to build a world-class transit system but lacks the will.
Imagine what we could achieve if we built up our institutional capacity and if leaders spent as much political capital on transit as they do on expanding highways, he told me via email. The systems other G7 nations have enjoyed for decades are not beyond our reachthey are a choice we can and must make.
For Clarkin, games like Subway Builder arent just entertainment. He believes they could actually have real-world impact. This is a tool for public imagination, he says. Im genuinely excited for the day someone takes their in-game map to a city council meeting and says, Look at what we can achieve!
To announce its entrance into 5G home internet service, Mint Mobile found the real-life version of a new AI-generated actress, even if only in (nick)name.
Tilly Norwood is the name of a so-called AI actress launched by AI talent studio Xicoia. It also happens to be the name of a woman who stars alongside Ryan Reynolds in Mint Mobile’s new ad for its home internet service, which its branding “Minternet.”
“It’s hard to believe that Mint is launching 5G home internet. It’s also hard to believe that a real version of an AI actress is out there,” a Maximum Effort representative tells Fast Company. “And thanks to the incredible and somewhat disturbing stalking detective abilities of our team, we found her. Just outside of Dallas, Texas, just one day before filming the commercial. Luckily she responded to our random DMs and was happy to assure the world that both she and the internet are very real.”
The fake Norwood has inspired backlash and a Wikipedia page, and the labor union SAG-AFTRA refused to say in a statement that the AI character is an actual actor, instead stating its “a character generated by a computer program that was trained on the work of countless professional performerswithout permission or compensation.”
As it happens, the real NorwoodNatalie “Tilly” Norwoodis a real Mint Mobile customer.
In the commercial, Reynolds, a former Mint Mobile co-owner who still makes ads for the wireless service provider through Maximum Effort, a production company he cofounded, asks if Norwood is real and “not an AI-generated combination of actors.”
“I’m a combination of my parents,” the real Norwood says.
Mint Mobile’s parent company was acquired by T-Mobile in 2023 in a deal worth up to $1.35 billion, and its 5G home internet service shows the brand is broadening its ambitions beyond mobile. The brand says its home internet service will use T-Mobile’s 5G network, and Mint Mobile is offering it for as low as $30 a month for customers with a Mint Mobile phone plan who prepay for three months.
In an advertising landscape that could increasingly see more AI-generated ads, sticking to real people is a smart strategy. A 2024 YouGov poll of respondents from 17 markets around the world found 51% were uncomfortable with a brand creating a virtual ambassador (34% were comfortable with it; 15% didn’t know how they felt about it).
In other words, using a fake Tilly Norwood in your ad could turn away half your audience. Meanwhile, the real living, breathing Tilly Norwood appears to be anything but polarizing.
The J.M. Smucker Co. says it doesn’t have a problem with other companies selling their own prepackaged, crustless sandwiches like its own popular Smuckers Uncrustables. They just have to get their own design.
Uncrustables is on its way to becoming a $1 billion brand, so of course there will be knockoffs, but according to Smucker, a recent Trader Joe’s version of Crustless Peanut Butter & Strawberry Jam Sandwiches is a bit too blatant. The company is using the design of the Trader Joe’s product and packaging to prove its point in a new lawsuit.
Smucker accused the grocery store chain of “an obvious attempt to trade off of the fame and recognition” of Uncrustable’s protected design marks in a suit filed Monday in the U.S. District Court for the Northern District of Ohio. The round shape and crimped edges of Trader Joe’s crustless sandwiches, which it released in late summer, look too similar to Uncrustables, Smucker says.
[Photos: Smucker’s, Trader Joe’s]
“Smucker does not take issue with others in the marketplace selling prepackaged, frozen, thaw-and-eat crustless sandwiches,” attorneys for the Orrville, Ohio-based food and beverage manufacturer wrote in the suit. “But it cannot allow others to use Smucker’s valuable intellectual property to make such sales.”
Smucker, which reported annual net sales of $8.7 billion in the most recent quarter, says it has invested nearly $1 billion over 20 years to grow its brand of crustless sandwiches into the No. 1 frozen handheld brand in its category. It’s paid off even as Smucker’s snack brands like Hostess Twinkies and Ding Dongs, which have recently rebranded, struggled in a shifting snack food landscape.
CFO Tucker Marshall said on Smucker’s August earnings call that Uncrustables is a “growth brand” for the company, along with the pet food brands Meow Mix and Milk-Bone. Marshall said that “people who are consuming Uncrustables for the most part are athletes, families with kids,” and that the brand performs strongly at universities and convenience stores.
“We really haven’t seen any impact at all from the GLP-1,” Marshall added, referring to weight-loss medications that are driving a trend toward healthier, high-protein snacks. The importance of Smuckers Uncrustables in the companys portfolio helps underline the urgency of the lawsuit.
In the suit, Smucker argues its trademarks for images like a “pie-like shape with distinct peripheral undulated crimping” as well as “a round crustless sandwich with a bite taken out showing filling on the inside” are being duped by Trader Joe’s without authorization. The suit extends to packaging concerns, as Smucker believes even the blue used for Trader Joe’s box of crustless PB&J sandwiches is strikingly similar to the blue used in the Uncrustables logo.
Smucker is seeking damages and demanding that Trader Joe’s destroy all the products, packaging, and promotional materials that use the current designs. Trader Joe’s did not respond to a request for comment.
There are other crustless sandwich brands that don’t use the Uncrustables-style circular shape and crimped design, like the square-shaped Jams and Walmart’s Great Value No Crust Sandwich. Chubby Snacks originally launched with circular sandwiches before getting hit with a cease-and-desist from Smucker. It pivoted in 2021 to a cloud-shaped sandwich.
Smucker’s suit follows a May lawsuit filed by Mondelez International against Aldi accusing the grocery store chain of duping the packaging of popular snack brands like Oreo and Nutter Butter. Aldi unveiled redesigned private-label packaging in September amid a wider industry trend toward upgrading generic branding that has spanned from Amazon to Walmart.
As lawsuits like those from Smucker and Mondelez show, with a rise in private-label competition, the big industry players are ready to protect their own branding, and with teeth if necessary.
I was interviewing for a job as a customer service agent with Anna. She had a low, pleasant voice and shed nailed the pronunciation of my namesomething few people do. I wanted to make a good impression except I had no idea what Anna was thinking because Anna couldn’t think. Anna wasnt technically a person. She was AI.
Not only is AI changing how we do our jobs, its also changing how we get jobs. This ranges from using AI to screen resumes, schedule interviews, even conduct them. According to a 2025 report, 20% of companies are using AI to interview candidates.
Even so, nothing can replace human recruiters, the folks whove deployed Anna into the wild stressed to me. After I spoke with her, I quickly understood why.
In this story, paid subscribers will learn:
What its like to actually go through a job interview with an AI agentand how to speak to them
Where companies should deploy AI interviewers that would benefit them and job seekers
AI Anna clocks in
Even though I wasnt really interviewing for a jobthis was all an exercise for this story, of courseI was still nervous.
I asked the team behind Anna to provide a job description so I could prepare, but outside of this experiment, I was sadly lacking in actual customer service experience. I also didnt know how AI Anna was going to react to awkward silences, panicked misdirection, or if shed be able to tell if I was lying. These worries are bad enough with a human. How would a computer program react?
I got on the phone and connected with Anna. She was pleasant, and frankly, sounded way more human than I was expecting. We exchanged greetings, and before long, I was in full-on job interview mode with an AI.
First up, she asked me to describe a time when I had to explain something complex over the phone clearly. I blanked. Finally, I described how journalism involves explaining complex ideas because youre asking questions. It sounded weak even to my own ears.
Sure enough, she was not impressed. Id like to explore a scenario thats more specific to the role were discussing, she replied firmly.
Fair point. I managed to dredge something up from a high school job. Mercifully, AI Anna accepted the answer and moved on.
Next, AI Anna wanted me to talk about a time when I had to problem solve for a customer. This, I could answer. I dove into my brief stint organizing a literary conference where writers paid to meet with agents. Occasionally agents went astray because they were hungover or running marathons and Id be left to find alternatives like rescheduling
Anna cut me off.
That sounds like a high-pressure situation. . . . Its great that you were able to come up with alternatives. Now Id like to switch topics for a moment.
Yikes. I wasnt ready to switch topics, but AI Anna was, and I couldnt tell why. Was my example off topic? Was I taking too long to answer her question? Before I could ask, Anna had already swept on to background checks.
I invented a criminal background and told AI Anna I had done some time in prison. She thanked me for being honest, and told me that she could not make any decisions. She said candidates with a criminal record would be considered on a case-by-case basis (something that would have to be verified by a human).
Then I wanted to know if Id be required to work overtime. She let me know Id be required to do overtime the first six months, but only one or two times a month. Needed, accurate information that couldnt just be googledgreat.
Honestly? While I found her transitions a bit jarring sometimes, she handled most questions with aplomb.
How we got here
AI Anna is the product of PSG Global Solutions, a staffing firm. Before deploying AI Anna in the market, the firm asked Brian Jabarian, a researcher at the University of Chicago Booth School of Business with doctorates in economics and philosophy, to study the AI Annas effectiveness. (Jabarian received no funding from PSG).
In a study released in September, Jabarian conducted an experiment where 70,000 applicants for a customer service job were randomly assigned a human interviewer, an interview with AI Anna, or the ability to choose between the two.
The results are surprising, and surprisingly promising for the candidates.
AI interviews resulted in a 12% increase in job offers, and a 17% increase in 30-day retention on the job. Moreover, when offered a choice, 78% of applicants chose to be interviewed by AI. Jabarian theorized this was because the AI was easier to schedule with: job applicants who needed a job quickly could book a call immediately.
Why the positive outcome data? Jabarian pointed out that, on average, an AI interviewer got through more required topics than human recruiters since they couldnt be distracted. (I mean, Anna did move at a brisk clip.)
AI leads to a more consistent interview experience, he said. It lets the candidate talk more, and has a 50% chance of covering 10 of 14 required topics compared to 25% for human interviewers.
AI Anna clocks out
Afterwards, I debriefed the interview with David Koch, PSGs chief transformation and innovation officer.
First, he showed me AI Annas backend: The platform had generated a recording of our conversation, a transcript, a summary of the call (including suggestions for next steps, like a follow up to discuss my criminal background), and an overall recommendation: AI Anna thought I was qualified (yay!) but merited human follow-up because of my criminal background.
AI Anna also recommended a follow-up because shed cut me off when I was talking about the literary conference. Koch explained my speaking cadence is a touch slower than average, and AI Anna is programmed to respond after a certain amount of time or else the flow of conversation can become jerky.
Koch noted that AI interviewing was better suited for some situations and not others. He recommended AI interviewing for high-volume hiring where theres a need to source candidates quickly for jobs that are seasonal and high turnover, like customer service agents or travel nurses. Koch also said AI interviewing is best suited for cases where theres less complexity, in which you dont need to sell a candidate on a role.
From my standpoint as a lay person, the technology behind AI Anna struck me as marvelous. She corralled me into staying on topic, and was capable of social niceties. She provided detailed answers to all my questions.
For recruiters, this could be life changing. Its not that AI Anna might replace them, per se (there were already things from the interview that a human colleague would have to address or follow up on). But recruiters could farm out tasks like screening calls to AI while they worked on more hih-level tasks.
However, this made me worry. If AI Anna existed to save companies time, what happens to candidates who get flagged for follow-up, even for something as simple as speaking slowlylet alone a criminal background? If there are more than enough qualified candidates to fill roles, I imagine a harried hiring manager would make offers to people who dont require follow-up. Exception cases that require more time, like me, might fall to the wayside.
The future: cold, but competent
After my conversation with AI Anna, I felt hollow. Typically, if an interview goes well, I have the high of having connected with someone who might make me feel valued, desired, and possibly in the mix for a new job. If it doesnt go well, I spend the next couple of days wallowing in self-pity and dissecting potential red flags.
AI Annas preprogrammed human-like intonation left me nothing to go on. Did she like me? Or was meh on me, but still think I was qualified?
I couldnt tell probably because AI Anna does not have emotions and did not care about me.
But how much does this matter? A Gallup study found that 44% of respondents said their interviews drove them to accept an offer or not. Ideally, candidates would be able to interview with their direct supervisor before getting a job in order to suss out personality matchbut for a screening interview, AI Annas value was undeniable.
She raised the floor for interview quality: Shes personable and she offers a consistent experience. Theres no need to worry about the mysterious intangible of chemistry. Jabarian also pointed out that AI interviews reduce gender discrimination by half.
Done right, AI interviewers could reduce bias and help qualified job candidates who may not perform well during interviews because they lack intangibles such as charisma.
Still, I missed talking to a human.
Generative AI is evolving along two distinct tracks: on one side, savvy users are building their own retrieval-augmented generation (RAG) pipelines, personal agents, or even small language models (SLMs) tailored to their contexts and data. On the other, the majority are content with LLMs out of the box: Open a page, type a query, copy the output, paste it elsewhere. That dividebetween builders and consumersis shaping not only how AI is used but also whether it delivers value at all.
The difference is not just individual skill. Its also organizational. Companies are discovering that there are two categories of AI use: the administrative (summarize a report, draft a memo, produce boilerplate code) and the strategic (deploy agentic systems to automate functions, replace SaaS applications, and transform workflows). The first is incremental. The second is disruptive. But right now, the second is mostly failing.
Why 95% of pilots fail
The Massachusetts Institute of Technology recently found that 95% of corporate GenAI pilots fail. The reason? Most organizations are avoiding friction: They want drop-in replacements that work seamlessly, without confronting the hard questions of data governance, integration, and control. This pattern is consistent with the Gartner Hype Cycle: an initial frenzy of expectations followed by disillusionment as the technology proves more complex, messy, and political than promised.
Why are so many projects failing? Because large language models from the big platforms are black boxes. Their training data is opaque, their biases unexplained, their outputs increasingly influenced by hidden incentives. Already, there are companies advertising SEO for GenAI algorithms or even Answer Engine Optimization, or AEO: optimizing content not for truth, but to game the invisible criteria of a models output. The natural endpoint is hallucinations and sponsored answers disguised as objectivity. How will you know if an LLM recommends a product because its correct, or because someone paid for it to be recommended?
For organizations, that lack of transparency is fatal. You cannot build mission-critical processes on systems whose reasoning is unknowable and whose answers may be monetized without disclosure.
From out of the box to personal assistant
The trajectory for savvy users is clear. They are moving from using LLMs as is toward building personal assistants: systems that know their context, remember their preferences, and integrate with their tools. That shift introduces a corporate headache known as shadow AI: employees bringing their own models and agents into the workplace, outside of ITs control.
I argued in a recent piece, BYOAI is a serious threat to your company, that shadow AI is the new shadow IT. What happens when a brilliant hire insists on working with her own model, fine-tuned to her workflow? Do you ban it (and risk losing talent) or do you integrate it (and lose control)? What happens when she leaves and takes her personal agent, trained on your companys data, with her? Who owns that knowledge?
Corporate governance was designed for shared software and centralized systems. It was not designed for employees walking around with semiautonomous digital companions trained on proprietary data.
SaaS under siege
At the same time, companies are beginning to glimpse what comes next: agents that do not just sit alongside software as a service (SaaS); they replace it. With enterprise resource planning systems, you work for the software. With agents, the software works for you.
Some companies are already testing the waters. Salesforce is reinventing itself through its Einstein 1 platform, effectively repositioning customer relationship management, or CRM, around agentic workflows. Klarna has announced it will shut down many SaaS providers and replace them with AI. Their first attempt may not succeed, but the direction is unmistakable: Agents are on a collision course with the subscription SaaS model.
The key question is whether companies will build these platforms on black boxes they cannot control, or on open, auditable systems. Because the more strategic the use case, the higher the cost of opacity.
Open source as the real answer
This is why open source matters. If your future platform is an agent that automates workflows, manages sensitive data, and substitutes for your SaaS stack, can you really afford to outsource it to a system you cannot inspect?
China provides a telling example. Despite being restricted from importing the most advanced chips, Chinese AI companies, under government pressure, have moved aggressively toward open-source models. The results are striking: They are catching up faster than many expected, precisely because the ecosystem is transparent, collaborative, and auditable. Open source has become their work-around for hardware limits, and also their engine of progress.
For Western companies, the lesson is clear. Open source is not just about philosophy. Its about sovereignty, reliability, and trust.
The role of hybrid clouds
Of course, there is still the question of where the data lives. Are companies comfortable uploading their proprietary knowledge into someone elses black-box cloud? For many, the answer will increasingly be no. This is where hybrid cloud architectures become essential: They allow organizations to balance scale with governance, keeping sensitive workloads in environments they control while still accessing broader compute resources when needed.
Hybrid approaches are not a panacea, but they are a pragmatic middle ground. They make it possible to experiment with agents, RAGs, and SLMs without surrendering your crown jewels to a black box.
The way forward
Generative AI is splitting in two directions. For the unsophisticated, it will remain a copy-and-paste tool: useful, incremental, but hardly transformative. For the sophisticated, its becoming a personal assistant. And for organizations, potentially, a full substitute for traditional software.
But if companies want to make that leap from administrative uses to strategic ones, they must abandon the fantasy that black-box LLMs will carry them there. They wont. The future of corporate AI belongs to those who insist on transparency, auditability, and sovereignty, which means building on open-source, not proprietary, opacity.
Anything else is just renting intelligence you dont control while your competitors are busy building agents that work for them, not for someone elses business model.
Below, Scott Anthony shares five key insights from his new book, Epic Disruptions: 11 Innovations That Shaped Our Modern World.
Scott is a clinical professor of strategy at the Tuck School of Business at Dartmouth College. His research and teaching focus on the adaptive challenges of disruptive change. Previously, he spent over 20 years at Innosight, a growth strategy consultancy founded by Harvard Business School professor (and father of the idea of disruptive innovation) Clayton Christensen.
Whats the big idea?
In 1620, Sir Francis Bacon wrote that there were three technologies for which it was possible to draw a clear line before and after: the printing press, the compass, and gunpowder. Those three technologies that changed the world stretched over 1,600 years.
Today, it feels like theres a big disruptive development every 1,600 seconds. Autonomous vehicles . . . augmented reality . . . artificial intelligence . . . additive manufacturing. And those are just the ones that begin with A.
How do we make sense of a world where change is truly the only constant? Understanding how disruptive innovation and epic change happens allows us to see the world more clearly.
1. Disruptive innovators transform the world.
Florence Nightingale was a nurse. You might have a visual of The Lady with the Lamp, and thats part of Florences story, but there is so much more.
Shocked by her experience in the Scutari hospital during the Crimean War, she developed a series of analyses, brilliantly visualized in polar area charts that showed the power of prevention and proper hygiene in hospitals. She wrote books explaining the essence of nursing that anyone could buy and read, and set up schools to train nurses.
What she did was disruptive innovation. Nightingale enabled a broader population to improve health standards and living conditions, focusing on prevention rather than treatment. Many of the things that we take for granted today, such as modern sewage systems or having light and fresh air during recovery, trace back to Nightingales work.
Disruptive innovators transform existing markets and create new ones by making the complicated simple and the expensive affordable. They open markets to broader populations that historically lacked wealth or specialized skills. They literally change the world.
2. Every story of disruptive innovation has heroes.
In the year 1437, Johannes Gutenberg was working on something in Strasbourg. No, it was not the printing pressat least, not yet. He was part of a team working on a trinket: a mirror that could capture the essence of the Holy Spirit during a planned pilgrimage in 1439.
Well, that pilgrimage was called off because of an outbreak of the Bubonic Plague. That was bad for many people, but good for the world, because Gutenberg and his team went in a different direction. They met someone named Conrad Saspatch, who had an innovative wooden press. In 1440, they combined that with a range of other things to create a working version of the printing press.
If you have an idea that you think could be disruptive, you need to find people who will support you.
To commercialize it, they needed customers, scale, and funding. They found a merchant named Johann Fust who gave them the capital to build their business. Fust ultimately sued them and took control of the technology, but thats not the primary point here. The point is that every story of disruption has a protagonist, but it is always accompanied by multiple people involved. Every story has heroes, and that word is plural.
So, if you have an idea that you think could be disruptive, you need to find people who will support you. If youre in an organization thats seeking to have more disruptions, you need to make sure the environment supports those innovators who are going to do the work.
3. Disruptive innovation is predictably unpredictable.
In 1947, a trio of researchers at Bell Labs developed a breakthrough that would change the world: the transistor. Their goal was to create a technology that would replace vacuum tubes in communications networks. That happened, but the path to get there was unexpected.
The transistor was an imperfect product in its early days. It had the benefits of being small, rugged, and not giving off heat, but it was also unreliable. You would have to redesign a system if you were going to use it. It wasnt good enough to plug into communications networks.
The first commercial market was in hearing aids. In 1952, the Sonotone 1010 featured a transistor. The fact that the transistor doesnt give off heat was a huge benefit for people wearing battery packs on their waists. The fact that its rugged was incredibly beneficial. The limitations just didnt matter. A couple of years later, 95 percent of hearing aids were powered by transistors, and the market had exploded.
This is a very predictable pattern. You never know exactly where disruptive innovation is going to start. Generally, however, you know it will be in a place that values it despite its limitations. That place is typically on the fringe of an existing market or in a completely new setting.
Around the same time that Sonotone was taking license to the transistor technology, chef Julia Child was dealing with a surprising setback. When we think of disruptive innovations, we dont think of chefs, but Child changed the world of cooking, making it much easier for people to cook great French dishes in their own homes.
Pull back and watch the full movie to understand disruptive change.
In 1951, the French chef failed her final exam at Le Cordon Bleu. That same year, she met Simca Beck and Louisette Berthold. The two were working on a book that would bring French recipes to an American audience. They asked Julia to join the team and bring her voice to the project. She agreed.
Mastering the Art of French Cooking came out 10 years later. Success was not a straight line. There were three different publishers and one near-death experience in November 1959, in which, at the very last minute, publisher number two said this book cannot be published.
This is predictable. Every story of disruptive innovation has twists and turns and fumbles and false steps and things that look and feel like failures. You cannot predict the specifics. You can, however, predict they will happen. What separates success from failure is not how good the original idea was. Its how the disruptive innovator deals with the journey.
When youre trying to understand disruption, focus on patterns like this. Recognize that a single moment can deceive you. Pull back and watch the full movie to understand disruptive change. Julia Child ultimately passed her test at Le Cordon Bleu and, in my opinion, her chocolate mousse recipe is perfection.
4. Disruption casts a shadow.
Disruption is very good for some, but it can be less good for others. Particularly in the middle of a disruptive change, there can be a lot of messiness.
Back in the 1920s, Henry Ford was obsessed with his visionto create a car for the great multitude. In 1908, he rolled out the Model T. It cost $890, or about $30,000 in todays terms. By 1924, the assembly line and lower employee turnover, facilitated by better wages, allowed Ford to dramatically decrease the cost to $260, or approximately $5,000 in todays terms. Sales of automobiles took off.
This was good for some, but less good for others. Cities were designed for people, not for cars. There were no traffic signals. There were no rules and norms governing who could do what, and sadly, people were getting hit, injured, and sometimes killed.
Two sides broke out. The motorists said, The problem here are the pedestrians. Were going to brand them as jaywalkers. Jay being slang for a country bumpkin who wasnt very educated. They had Boy Scouts hand out cards in cities, telling people to cross at designated areas.
This was good for some, but less good for others.
The pedestrians fought back. They sought to brand the motorists as flivverboobs. Flivver was slang at the time for cars, and boob . . . well, thats still pretty universal.
You know who won the battle. In 1924, a New York traffic warden said, We now know about 80 percent of incidents are caused by jaywalkers. By the late 1920s, the word flivverboob had basically disappeared.
Disruption always casts a shadow. The middle can be very messy. You have to understand it, or it will swallow you.
5. Success with disruption requires patient perseverance.
People talk about the accelerating pace of change, but we forget that when we see a big breakthrough, theres often been decades of work before it.
For example, in 2022 OpenAI introduced ChatGPT. It became the fastest technology in history to get to 100 million users. But by some dimensions, that technology was 67 years old, tracing back to a 1956 conference at Dartmouth College where the term AI was coined.
Around the same time as that conference, a chemist at Corning, Don Stookey, made a surprising discovery. He accidentally set his kiln to a temperature that was way too hot. He expected a gooey mess, but instead he discovered the first synthetic glass ceramic.
Corning commercialized this in a line of kitchenware and, in parallel, launched Project Muscle to make the material clear. The result was something 14 times stronger than normal glass. But Corning couldnt make it thin. They thought a possible market could be automobile windshields, but tests with crash test dummies showed that the head would not survive a collision with the glass because it was that strong. In 1971, after $300 million investment in todays terms, Corning put the project on ice.
In 2007, Steve Jobs was getting ready to launch the iPhone. He picked up the prototype, and its plastic screen just didnt appeal to his aesthetic sense. He wanted glass. He knew Corning had provided an innovative screen for Motorolas RAZR phone. Even though Corning shut down the project, they continued experimenting and exploring, and ultimately made the glass thinner. They called it Gorilla Glass.
Steve Jobs came to Cornings headquarters, talked to CEO Wendell Weeks, and said, I want this, I want it at scale, and I want it fast. Weeks said, Great, but we cant do it at scale and we cant do it fast. Steve Jobs turned on his reality distortion field and, without blinking, said, Yes, you can. You can do it. And Corning did. By 2024, eight billion devices had screens with Gorilla Glass. When it comes to disruption, you must be comfortable being uncomfortable because it almost always takes a lot longer than you think.
This article originally appeared in Next Big Idea Club magazine and is reprinted with permission.
A majority of Gen Z workers are turning to AI chatbots during the workday for personal reasons, including mental health support, with 40% saying they talk to AI for at least an hour every day, according to a new Resume.org survey.
Many Gen Zers entered hybrid or remote jobs where casual mentorship or watercooler chats never formed, so AI fills that relational void,” said Kara Dennison, Resume.orgs head of career advising. “It listens, it responds thoughtfully, and it never criticizes.”
She added: That creates a sense of psychological safety thats often missing in corporate hierarchies. Its about connection, control, and immediacy. Theyre using AI the way earlier generations used coffee breaks or hallway chats: to decompress, problem-solve, or feel understood.
While older generations might describe ChatGPT as a tool, 47% of Gen Z say it feels far more personal: 25% of Gen Z describe ChatGPT, Copilot, and other AI bots as their therapist or coach, a friend, or coworker, while 34% admit to confiding in AI chatbots about things theyve never told another person. Some 16% say they frequently discuss personal topics such as mental health or relationships with AI, while 33% say they do so occasionally.
Resume.org’s survey collected data from 1,000 full-time U.S. Gen Z workers ages 18 to 28 who used an AI chatbot such as ChatGPT or Copilot in the past week.
Gen Z may be using ChatGPT for therapy, but mental health experts say it comes with risks. “Using a general-purpose chatbot as a therapist compromises the fundamental elements of safe care: clinical oversight, legal confidentiality, and a dependable route to human intervention,” Gijo Mathew, chief product officer at Spring Health, a global mental health platform for employers and health plans, told Fast Company.
“This can introduce significant risks, particularly in multi-turn, emotionally charged discussions,” Mathew continued. “Most chatbots and large language models (LLMs) were not designed for mental health support and may overlook warning signs or offer articulate yet clinically unsound advice.”
According to the survey, 43% of Gen Z workers spend at least 30 minutes per day using ChatGPT or a similar AI chatbot; 13% use it for one to two hours a day; 6% for two to four hours a day; and 5% for more than four hours a day.
When it comes to dealing with stress and well-being on the job, 38% of Gen Z are turning to AI to take breaks, and 33% to talk through work-related stress or frustrations. That’s time that could be spent interacting with other humans.
The findings also come at a time when 89% of corporate workers say they faced at least one mental health challenge in the past year.
Every office has that coworker that turns up to a meeting coughing and sniffling while proudly proclaiming they have never once taken a sick day in their career. (If there isnt one, maybe its you.)
But as one viral TikTok makes clear, those attitudes towards taking sick days may be changingjust as sick days themselves are changing, as some think being sick isnt a real excuse to not work in the WFH era.
The skitwhich has more than 2.3 million viewssees popular TikTok creator Delaney Rowe adopting the role of that coworker, turning up to a meeting with a hospital tag still on wrist, oh-so bravely battling through the workday while simultaneously making it everyone elses business.
Person you work with who thinks theyre a hero for ‘powering through’ while sick, she wrote.
The days of powering through are now gone, as nearly a third of Americans say theyd rather you didnt show up to work if youre feeling sick, according to a new Talker Research survey of 2,000 people in the US. Many in the TikTok posts comments agreed.
I get so mad when people risk ME getting sick, one wrote. Get away from me.
Another joked: This is me but I’m just soft-launching calling out the next few days.
Others have even offered scripts for how to successfully call in sick without guilt or fear.
Take your sick days. Those days are for you and theyre not just for when youre sick, one TikTok creator advised. Theyre for when you just want a day to lay down all day and watch movies and eat food. You can do that. Theyre for when youre feeling a little bit off and you just dont want to deal with it today.
The workplace is a minefield of unwritten rules which workers have long abided by. Not taking sick days, even when allotted by an employer, is one.
However, 31% of those surveyed by Talker Research say theres no longer a badge of honor or admirable quality to employees turning up to work ill. Just a quarter of Americans strongly believe it would impress bosses or superiors.
As one Reddit thread put it: Never taking sick days is not a flex some people think it is. One commenter went on to describe a coworker who point blank refuses to take sick days, writing: All of this goes unnoticed by management. No one gives a damn. No one is asking him to do it, no one is patting him on the back.
They added: Then he proceeds to get frustrated with the rest of us that we don’t do the same.
The COVID-19 pandemic permanently changed workers’ attitudes to sick leave, making clear the importance of staying home to avoid infecting coworkers. At the same time, Gen Zers entered the workforce in droves, championing mental health days, worklife balance, and the importance of boundaries at work.
But some workers may still feel compelled to show up and put in face time with the boss. That translates to remote work, as well; a green active light on Slack or Teams communicates availability. But whether its in person or online, working while sick sets a bad example for the rest of the staff, chipping away at work-life boundaries that are already blurrier than ever.
Working while sick can also lead to presenteeism: working while sick, but since youre sick, youre less productive. Presenteeism is bad for business, especially when it risks infecting an entire office with a cold or flu picked up over the weekend.
And after witnessing layoff after layoff, todays employees may be more inclined to take that R&R thats available to themrather than give their all to a job that deems them disposable.
