 |
|
|
|||||
For its 2026 postage stamps, the U.S. Postal Service is going colorful and graphic. USPS gave a first look at some of the stamps set to be released next year, including the latest edition of its Love stamp, stamps commemorating the 250th anniversary of the U.S., and stamps depicting figures including a boxer, a martial artist and actor, and a pair of published poets. The stamps will be released on a rolling basis beginning in January and available at Post Office locations and online. This early preview of our 2026 stamp program underscores the Postal Services commitment to celebrating the artistry and storytelling that make stamps so special, Stamp Services director Lisa Bobb-Semple said in a statement. Each stamp is a small work of art an entryway into a larger story that connects people, places and moments in history.” [Image: USPS] Many of the stamps are bright or use typography in bold or creative ways. The 2026 Love stamps are a series of four illustrations of stylized red, white, and blue birds by illustrator James Yang that were inspired by midcentury U.S. design and Japanese children’s book illustrations, according to USPS. [Image: USPS] Stamps for Muhammad Ali designed by USPS art director Antonio Alcalá show an Associated Press photo of the boxer with his gloves up and his last name in big, all-caps, sans-serif type in red and black that evokes a boxing match promotional poster. [Image: USPS] A painting of Bruce Lee by artist Kam Mak shows the martial artist and actor against a yellow brushstroke background as he kicks the words “USA FOREVER” and “BRUCE LEE,” which were cleverly angled to look like he snapped them in two. [Image: USPS] For its “Figures of the American Revolution” stamps, multiple artists depict 25 people, from household name Founding Fathers like George Washington and Benjamin Franklin to lesser known figures as Deborah Sampson, the only woman to earn a military pension in the war after she dressed up like a man called Robert. The diverse selection of people were chosen to represent the Revolution as a collective effort, USPS says. “Its unusual to design a pane of stamps featuring 25 different portraits” USPS art director Ethel Kessler said in a statement. “But that number felt essential. How else could you begin to tell the story of the Revolutions complexity with fewer?” [Image: USPS] The typographic “Declaration of Independence” stamp also marks next year’s anniversary with “1776” written out in feather quill pens by typographer Juan Carlos Pagan. [Image: USPS] The “Lowriders” stamps pay homage to customized lowrider cars with photos by Philip Gordon and Humberto Beto Mendoza and gothic-style type paired with flourishes borrowed from lowrider paint jobs. Photographer David Schwartz contributed images for the “Route 66” stamps, which celebrate the 100th anniversary of the iconic highway. [Image: USPS] Other forthcoming stamps including “International Peace” showing an origami crane by Peace Crane Project founder Sue DiCicco, “Bald Eagle: Hatchling to Adult,” a pane of five stamps depicting the life of America’s national bird, and a stamp commemorating Colorado’s 150th anniversary. [Image: USPS] Writer Phillis Wheatley, who published what’s believed to be first book by a woman of African descent in the American Colonies, appears on the 49th Black Heritage stamp by artist Kerry James Marshall. Sarah Orne Jewett, a novelist and poet, appears on the 35th Literary Arts series by artist Mark Summers. Next year’s Lunar New Year stamp shows a horse mask by Sally Andersen-Bruce. [Image: USPS] USPS says more stamp announcements are forthcoming, and it’s also planning to rerelease an old stamp next year as part of its Stamp Encore Contest. [Image: USPS]
Category:
E-Commerce
Spend a few minutes on developer Twitter and youll run into it: vibe coding. With a name like that, it might sound like a passing internet trend, but its become a real, visible part of software culture. Its shorthand for letting AI generate code from simple language prompts instead of writing it manually. In many ways, its great. AI has lowered the barrier to entry for coding, and thats pulled in a wave of hobbyists, designers, and side-project tinkerers who might never have touched a codebase before. Tools like Warp, Cursor, and Claude Code uplevel even professional developers, making it possible to ship something working in hours instead of weeks. But heres the flip side: when AI can move faster than you can think, its easy to run straight past the guardrails. Weve already seen how that can go wrong, like with the recent Tea app breach, which shows even polished, fully tested code can hide critical vulnerabilities if humans dont review it thoroughly. Optimizing for speed over clarity lets AI produce something that works in the moment, but without understanding it, you cant know what might break later. This isnt just technical debt anymore; its a risk to customer trust. The instinctive reaction to solve this trade-off is to throw more tech at the problem: add automated scans, add a secure by default setting. Those things matter. But Id argue that failure in vibe coding doesnt start with tooling, it starts with leadership. If you dont lead your team through this new way of working, theyll either move too slow to benefit from AI or move so fast they start breaking things in ways a security checklist cant save you from. The real job is steering, not slowing down When we built agentic coding agent Warp 2.0, we put a simple mandate in place: Use Warp to build Warp. That means every coding task started with prompting an AI agent. Sometimes it nailed it in one shot; sometimes we had to drop back to manual coding. But the point wasnt dogma, it was to force us to learn, as a team, how to work in an agent-driven world. We learned quickly that more AI doesnt automatically mean better. AI can write a thousand lines of plausible-looking code before youve finished your coffee. Without structure, thats a recipe for brittle, unmaintainable systems. The real challenge was getting people to treat AI-generated code with the same discipline as code they wrote themselves. Thats a leadership problem. Its about setting cultural norms and making sure they stick. Three things leaders need to get right 1. Hold developers accountable The biggest mental trap is treating the AI as a second engineer who owns what it wrote. It doesnt. If someone contributes code to a project, they own that code. They need to understand it as deeply as if they typed it out line by line. AI wrote it should never be an excuse for a bug. Leaders cant just say this once; they have to model it. When you review code, ask questions that make it clear you expect comprehension, not just functionality: Why does this query take so long to run? What happens if the input is null? Thats how you set the standard that understanding is part of shipping. 2. Guide AI with specifics Using large, one-shot prompts is like cooking without tasting as you go: sometimes it works, but usually its a mess. AI is far more effective when you request small, testable changes and review them step by step. Its not just about quality, it also builds a feedback loop that helps your team get better at prompting over time. In practice, this means teaching your team to guide the AI like theyd mentor a junior engineer: explain the architecture, specify where tests should live, and review work in progress. You can even have the AI write tests as it goes as one way to force smaller, verifiable units of work. 3. Build the review culture now In AI workflows, teams move fastest when AI and humans work side by side, generating and reviewing in small steps. The first draft of a feature is the most important one to get eyes on. Have someone review AI-generated work early and focus on the big-picture questions first, like whether its secure, reliable, and solves the right problem. The leadership challenge is making reviews a priority without slowing anyone down. Have teams aim to give feedback in hours, not days, and encourage finding ways for work to keep moving while reviews happen. This builds momentum while creating a culture that values careful, early oversight over rushing to get something done. Guardrails only work if people use them Safety tools and checks can help catch mistakes, but they dont replace good habits. If a team prioritizes speed over care, AI guardrails just get in the way, and people will find ways around them. Thats why the core of leading in the AI era is cultural: you have to teach people how to integrate AI into their workflow without losing sight of the fundamentals. The teams that get this right will be able to take advantage of the speed AI enables without bleeding quality or trust. The ones that dont will move fast for a while, until they ship something that takes them down. Vibe coding isnt going away, and I think thats a good thing. So long as teams lead with people, not just technology, they will come out ahead and create better experiences for users along the way.
Category:
E-Commerce
Want more housing market stories from Lance Lamberts ResiClub in your inbox? Subscribe to the ResiClub newsletter. Earlier this month, the National Association of Realtors (NAR) released its annual survey, which found that the median age of first-time U.S. homebuyers in 2025 climbed to 40. Thats up from 38 in 2024and far above the median age in 1992, when it was 28. At first glance, it appears that deteriorating housing affordabilitydriven by the Pandemic Housing Boom and the 2022 mortgage-rate shockhas pushed the age of first-time buyers higher. However, when you look across other data sources, including the Federal Reserve Bank of New York and the U.S. Census Bureau, you dont see the same spike. ResiClub dug deeper into the data to figure out whats really going on. According to the Federal Reserve Bank of New York, the average first-time homebuyer in 2024 was 36.3 years oldjust a little younger than NARs estimate of the median first-time homebuyer age of 38 in 2024. Initially, one might suspect the difference simply stems from the fact that the New York Fed reports an average while NAR reports a median. However, when you peel back the onion, youll see theres a large historical divergence between the two organizations figures. That raises the question: How did they each collect their data? The NAR data series is calculated by an annual survey. For this year’s survey, NAR mailed out a 120-question survey to 173,250 recent homebuyers. The recent homebuyers had to have purchased a primary residence home between July 2024 and June 2025. In total, 6,103 responses were received this year. The New York Fed doesnt collect its data by survey. Instead, its looking at credit report data, which it says has 5% of nationally representative individuals since 1999. Back in August, ResiClub emailed both NAR and the New York Fed to get their thoughts on the first-time homebuyer data divergence. Jessica Lautz, NARs deputy chief economist, told ResiClub on August 14: “The Federal Reserve is basing their data on credit data. The Profile of Home Buyers and Sellers [from NAR] is based on a survey of those who purchased a primary residence home in the last year. Thus, the NAR survey includes the 9% of first-time buyers who paid cash for their home and did not finance their purchase. Excluded are first-time buyers who purchased a vacation home or mom-and-pop rental as their first purchase. A trend which has popularized as young adults are unable to achieve homeownership in the expensive areas they may live in. The NAR data collection is mid 2023 to mid 2024 vs. a calendar year. Lastly, NAR uses medians as a measure of central tendency vs. average.” Donghoon Lee, an economic research adviser in microeconomics at the Federal Reserve Bank of New York, told ResiClub on August 13: “We cant tell you how the NAR annual survey was constructed, but in our previous blog, we wrote about a comparison between our data and NAR source. Here are some of the unsophisticated differences. New York Fed Consumer Credit Panel is a panel of credit report data where we follow 5% of nationally representative individuals since 1999, and is not a survey. We are not subject to any low response rate issue of the respondents. We identify first-time homebuyers when a mortgage account appears for the first time on the individuals credit reports. If a home purchase was made without a mortgage (such as a cash purchase) then we dont see thm, and not included in calculating the statistics. My takeaway? Im going to take this particular first-time homebuyer dataespecially the NAR serieswith a grain of salt going forward. Instead, Ill lean more on generational homeownership rates by age. And when you look at those figures, they clearly confirm that younger generations are entering homeownership more slowly than their older peers. ResiClub also messaged Apartment List to get its latest calculation. See below: Apartment Lists analysis shows that with each successive generation, homeownership rates take longer to ramp up. This pattern isnt unique to Gen Zbaby boomers were slower to reach key homeownership milestones than the Silent Generation, Gen X was slower than the boomers, millennials were slower than Gen X, and Gen Z is slower still. The fact that each generation takes a little longer to enter homeownershipduring both periods of good and poor housing affordabilitysuggests an underlying secular shift that isnt just driven by affordability. In my view, that secular shift largely comes down to lifestyle/cultural shifts. With each new generation, Americans are spending more years in school, marrying later, having children later (and having fewer kids), and ultimately buying homes later. I call this phenomenon lifestyle delays. Given how homeownership rates are calculated (the number of owner-occupied housing units divided by the total number of occupied housing units), its likely that the gradual slowdown in homeownership by generation is actually understating the true drop-off. In plain English, what do I mean? If someone in their twenties or thirties is still living with their parents, they technically arent counted as their own household and therefore arent included in the denominator. And when you look closely at the generational data (see the Apartment List analysis above), youll note that with each generation, Americans are taking longer to move out of their parents homes. If you adjust for this, the real homeownership rates by generationsomething John Burns Research and Consulting has analyzedyou find that the generational homeownership drop-off is indeed larger than the headline data suggests.
Category:
E-Commerce
It took the Equinox Groupthe parent company of luxury gym chain Equinox, Equinox hotels, and Soulcyclearound five years to recover from COVID. But the company has recovered, claiming that 2025 will be a record year from a profitability perspective. This year, it announced big plans for expansion. Harvey Spevak, executive chairman and managing partner of Equinox Group, tells us about the company’s plan to open 40 clubs in new markets, its expansion into the Middle East, and the real reason it ditched Kiehls for Grown Alchemist. In the next couple of years, Equinox plans to open 40 new clubs. What is driving this growth? We’ve always been a high growth company leading up to COVID. COVID was pretty rough. The way we thought about our strategy was to rebound the business and then get back to growth and transformation. We’re back to high growth from a financial perspective. From a growth perspective, there’s tremendous demand for the Equinox brand and the Equinox offering in existing markets, and also in lots of new markets. Over the last 12 months, we’ve built a pretty robust pipeline, like you’ve mentioned. That 40 locations is probably closer to 50 at this point. We’ve opened a bunch of locations in new markets this year. We opened our first in Philadelphia Rittenhouse, which is performing extremely well; better than we expected. We opened our first in Seattle in June, also exceeding our expectations. In the fall, we’re opening our first in San Diego. By the end of the year [well have opened], San Diego, West Loop of Chicago, another one in New York, another one in Los Angeles and Santa Monica. We’re excited about all of these. We’ll open five locations between now and by January. How do you decide which new markets to enter? Our real estate team is very knowledgeable about all the markets where we’ve always wanted to go. When I say the markets, it includes cities as well as zip codes, if not by block. That’s a big part of why our unit economics are so strong. But the world has changed in terms of where there is demand. While there continues to be great demand in some of our existing markets like Manhattan, Southern California, Northern California, there’s new markets where we didn’t exist before. I mentioned Philadelphia or Seattle, but then there’s also the South. We recently announced our first location in Nashville that’ll open next year. We think South Florida will become our third biggest market outside of New York and Los Angeles. And a lot of that’s because of the migration of people from New York and other markets. We’re opening our first location in West Palm Beach next month. We sell memberships in advance of opening, and the response has been absolutely outstanding. In West Palm Beach, we started selling memberships in Junethe worst time of the year it could be selling memberships, but we did that because we were going to open in November and we’re going to open with a member count that is higher than what we thought we would achieve in four years. Weve all read about the migration to Florida, but what do you think are the factors leading for places like Nashville to become new markets? These markets are evolving rapidly and there’s greater desire for health and wellness broadly. It’s become a bigger priority during and coming out of COVID than ever before. That’s a global phenomenon. Then you think about the Equinox offering and the authority the brand has around high-performance living; it’s just kind of a perfect storm. Are you opening more Equinox hotels? We have a pipeline on the hotel side of around 10 to 12 new hotels. The difference is the gestation period for a hotel is much longer, particularly when you’re building a new hotel as compared to a club. It takes years. Our next hotel is opening in the Red Sea in Saudi Arabia in spring of 2026. Why the Red Sea? Because there was demand for our brand, our hotel, our club, our entire offering. We also have branded residential there, which is another part of our offering now. But there was a lot of interest in bringing us to a resort location. It’s a beautiful setting. It’s some of the most magnificent water you’ve seen in the world. It’s an undeveloped piece of real estate in Saudi Arabia. So in partnership with the government there and the Sovereign Wealth Fund, they have built out a big area, a marina. Us and some other hotels are going to be operating there. Ours is designed by David Rockwell, and we’re excited about that. We’ve already announced another hotel in Saudi Arabia. There’s one unannounced that’s comingthere’s potentially one, if not two, more in the region. Then we’ve also announced one in Nashville. There’s one we’re working on the Caribbean. It’s remarkable who stays at our hotel now. It’s literally a who’s-who of Fortune 100 companies to people out of Hollywood or in the music industry. Why is the Middle East an attractive location for you to set up shop? It’s fascinating what’s been going on there over the last four or five years. And it varies by country and city, but there’s so much interest in health and wellness, hospitality, living a healthy lifestyle, and the Equinox brand. The brand awareness is very high because a lot of the 30 and 40 year olds in the Middle East have been educated here in the U.S., and they’re educated in a lot of cities where we have big portfolios of clubs. After billionaire real estate developer Stephen Ross, whose company Related is a majority owner of Equinox, held a fundraiser for then-republican candidate Donald Trump in August 2019, a lot of customers boycotted Equinox and Soulcycle clubs. Saudi Arabia doesnt have a great track record when it comes to respecting human rights. Do you worry about the reputational risk to your brand this might have? Fair question. When Stephen did the fundraiser back in summer of 2019, we definitely [were] given a very hard time by our members, our prospective members, what we call our alumni members. It was a time when cancel culture was a very big thing, and we definitely took our hits during that. But what I would also say is that, [it] is because people are so emotionally attached to the Equinox brand. We got a list of all of corporate America that at that time was donating to Donald Trump or the Republican Party. It’s a who’s-who of American household names. We dont donate to any party. Weve always been, in effect, Switzerland and we believe were an inclusive community. I dont mean that just by race or gender. Everybody can have their political views, everybody can have their religious views. That’s the magic of Equinox, that it’s such a diverse and inclusive community, both from our employee side and from our member side. What struck me was because people love us, they expect more from us. At CVS where you buy toothpaste, nobody cared that CVS was one of the biggest donors to Donald Trump at the time. Nobody was canceling them. Stephen Ross, who is obviously a significant investor and a partner, made a personal decision to do something with supporting Donald Trump. It was really from a business perspective more than anything, but that was his decision. It was unfortunate that it affected Equinox and SoulCycle at the time. As we move forward, we’re not going to weigh in on politics, we’re not going to weigh on religion, but peopl are allowed to have their personal points of view. That’s how America was built. It was an unfun time, but we’re always sensitive about who we partner with, big and small. When you talk about the Middle East, everybody is partnering with a lot of these countries, including Saudi Arabia. And Saudi Arabia is transforming and undergoing substantial reform. We’re not alone in deciding that it’s okay to do business there. Equinox is in the middle of transforming its wellness offeringslast year you launched Optimize, a $40,000-a-year health and longevity program that involves lab testing, personal training, and nutrition and sleep coaching. Tell me about your investments in that category. It’s soon going to be a $10 trillion global category. There’s no brand that’s better positioned than Equinox to take advantage of what we refer to as high-performance living. If you listen to the true respected longevity gurus, they align with this philosophy. We talk about it from a science perspective. How do you live a high-performance lifestyle? We talk about movement, which is working out and being active. We talk about nutritionI prefer the word nourishment because nutrition is medicinal. Nourishment is just eating well and eating in a way that gives you energy and fuel for the body. Then theres recovery and sleep is at the core of that, but there’s other aspects to recovery at this point. A lot of the biohacking comes into that. Then there is community. Nobody’s written that narrative more so in the community than we have. We introduced Optimize last year through a partnership with Function Health [which does lab testing]. We take biomarkers [from blood tests] and then you are assigned a concierge who quarterbacks your program with a team of specialists, your personal trainer that is referred to as a coach, your nutrition coach, your sleep coach. Then we retest to see how the progress is going. You also have programs to support members using GLP-1s. Has their increasing prevalence changed the business? I don’t think it’s had a dramatic effect on the business, but I definitely think the needs and wants of certain members have changed. So early on when GLP-1s exploded, there was lots of press out there asking, does this mean gyms are dead? But what the world has learned through lots of science and research and the different experts is if you use that as a kickstart, but then compliment it and move towards living a different lifestyle, meaning all the stuff around being active, strength training, eating wellthose are the people that are getting the greatest results from GLP-1. We saw it as an opportunity to create a special GLP-1 training program. I think what’s definitely part of the GLP-1 effect is that people realize strength training is more important than ever before. Youve mentioned Optimize, your GLP-1 program, and Equinox Living. There’s another chain, Life Time, that has similar offerings. How do you view that competition? There’s some similarities in the programming, but the offerings are very different. They are much more of a rural to suburban play, and we’re much more of an urban play. We’re much more luxury, high touch service oriented and more boutique-oriented versus. But there is some overlap programmatically. I think that at the end of the day though, what I would say is if you want the Equinox experience, you’re not going to get it at Life Time. If they want to follow a lead on some aspects, that’s up to them, that’s flattering. I have a lot of respect for what they’ve been able to accomplish, but what we’re doing is very unique and our community loves what we do. So I don’t really view it as competition; I just view it as someone else operating in the category. And I think there’s plenty of room for both of us and others. Why did you ditch Kiehl’s for Grown Alchemist products in your locker rooms? I am going to be too transparent. I’m going to get yelled at by my team afterwards. Kiehls was a very successful relationship for a very long timesince 2009. But through all the changes of their leadership, which seemed to be very frequent, the product was, in our mind, a little stale, and they weren’t innovating around it. And there were some things in the product, which in the world of getting clean were not good ingredients, and they were reluctant to change [them]. So we said, this is not the right product for our members. Despite what they’ve said in the market, we decided to end the relationship. Interestingly, I’m going to point outand this is where I’m really in troubleis that other brand that you mentioned, that started in Minnesota they’ve picked up our sloppy seconds [and started offering Kiehls in their locker rooms]. Nobody likes sloppy seconds. Thats certainly not us. So we ended the relationship. We decided to go another direction. There’s no doubt that the other direction caused an uproar. Clean is tricky, and Grown Alchemist is clean, but it comes with issues. It doesn’t suds as much. So anyway, to make a long story short in this regard. I would just say stay tuned for more announcements in the not too distant future. Were your customers asking for clean products? There was some of that, but also we were looking to say, what’s the future here versus what’s yesterday? We felt that Kiehl’s was yesterday, and we wanted something more progressive. And so we went in direction. There’s no doubt a lot of our members love the direction we went in. And other members were likeas you saw on Redditwhat are you doing? This is ridiculous. You’re cheapening out. It’s actually more expensive. Just so we’re clear, it was not us cheapening out. [But] we have some things coming in the not too distant future. Whats your own workout routine? I’ve practiced what I preach. I’ve had the same coach for 20 years who’s been amazing. I strength train with him three times a week. Then separately, I love to sprint. I probably sprint more than I should, but I sprint probably four times a week and do cardio like six times a week. I eat really clean although I believe that pizza’s a separate food group because that is my kryptonite, that’s my weakness. I’m a big believer in the sleep side. I do all the biomarker stuff that I mentioned early on to inform this, but I make sure I get my sleep regardless of what’s happening. Which Equinox club do you visit the most? Thats like asking which is your favorite child. I have twins, and I often joke about when I’m with one or the other, I tell them you’re my favorite. And then I tell the other one, she’s my favorite. You’re going to be paying a lot of psychiatry bills down the line. Probably, so far it’s worked out okay. But I like so many clubs for different reasons. Because my offices are at Hudson Yards, I use the Equinox at Hudson Yards most frequently.
Category:
E-Commerce
Youve just finished a strenuous hike to the top of a mountain. Youre exhausted but elated. The view of the city below is gorgeous, and you want to capture the moment on camera. But its already quite dark, and youre not sure youll get a good shot. Fortunately, your phone has an AI-powered night mode that can take stunning photos even after sunset. Heres something you might not know: That night mode may have been trained on synthetic nighttime images, computer-generated scenes that were never actually photographed. As artificial intelligence researchers exhaust the supply of real data on the web and in digitized archives, they are increasingly turning to synthetic data, artificially generated examples that mimic real ones. But that creates a paradox. In science, making up data is a cardinal sin. Fake data and misinformation are already undermining trust in information online. So how can synthetic data possibly be good? Is it just a polite euphemism for deception? As a machine learning researcher, I think the answer lies in intent and transparency. Synthetic data is generally not created to manipulate results or mislead people. In fact, ethics may require AI companies to use synthetic data: Releasing real human face images, for example, can violate privacy, whereas synthetic faces can offer similar benefit with formal privacy guarantees. There are other reasons that help explain the growing use of synthetic data in training AI models. Some things are so scarce or rare that they are barely represented in real data. Rather than letting these gaps become an Achilles heel, researchers can simulate those situations instead. Another motivation is that collecting real data can be costly or even risky. Imagine collecting data for a self-driving car during storms or on unpaved roads. It is often much more efficient, and far safer, to generate such data virtually. Heres a quick take on what synthetic data is and why researchers and developers use it. How synthetic data is made Training an AI model requires large amounts of data. Like students and athletes, the more an AI is trained, the better its performance tends to be. Researchers have known for a long time that if data is in short supply, they can use a technique known as data augmentation. For example, a given image can be rotated or scaled to yield additional training data. Synthetic data is data augmentation on steroids. Instead of making small alterations to existing images, researchers create entirely new ones. But how do researchers create synthetic data? There are two main approaches. The first approach relies on rule-based or physics-based models. For example, the laws of optics can be used to simulate how a scene would appear given the positions and orientations of objects within it. The second approach uses generative AI to produce data. Modern generative models are trained on vast amounts of data and can now create remarkably realistic text, audio, images, and videos. Generative AI offers a flexible way to produce large and diverse datasets. Both approaches share a common principle: If data does not come directly from the real world, it must come from a realistic model of the world. Downsides and dangers It is also important to remember that while synthetic data can be useful, it is not a panacea. Synthetic data is only as reliable as the models of reality it comes from, and even the best scientific or generative models have weaknesses. Researchers have to be careful about potential biases and inaccuracies in the data they produce. For example, researchers may simulate the home-insurance ecosystem to help detect fraud, but those simulations could embed unfair assumptions about neighborhoods or property types. The benefits of such data must be weighed against risks to fairness and equity. Its also important to maintain a clear distinction between models and simulations on one hand and the real world on the other. Synthetic data is invaluable for training and testing AI systems, but when an AI model is deployed in the real world, its performance and safety should be proved with real, not simulated, data for both technical and ethical reasons. Future research on synthetic data in AI is likely to face many challenges. Some are ethical, some are scientific, and others are engineering problems. As synthetic data becomes more realistic, it will be more useful for training AI, but it will also be easier to misuse. For example, increasingly realistic synthetic images can be used to create convincing deepfake videos. I believe that researchers and AI companies should keep clear records to show which data is synthetic and why it was created. Clearly disclosing which parts of the training data are real and which are synthetic is a key aspect of responsibly producing AI models. Californias law, Generative artificial intelligence: training data transparency, set to take effect on January 1, 2026, requires AI developers to disclose if they used synthetic data in training their models. Researchers should also study how mistakes in simulations or models can lead to bad data. Careful work will help keep synthetic data transparent, trustworthy, and reliable. Keeping it real Most AI systems learn by finding patterns in data. Researchers can improve their ability to do this by adding synthtic data. But AI has no sense of what is real or true. The desire to stay in touch with reality and to seek truth belongs to people, not machines. Human judgment and oversight in the use of synthetic data will remain essential for the future. The next time you use a cool AI feature on your smartphone, think about whether synthetic data might have played a role. Our AIs may learn from synthetic data, but reality remains the ultimate source of our knowledge and the final judge of our creations. Ambuj Tewari is a professor of statistics at the University of Michigan. This article is republished from The Conversation under a Creative Commons license. Read the original article.
Category:
E-Commerce