May 13, 20253 min readDisasters Displaced a Record 46 Million People Last YearIn 2024 disasters displaced a record number of people both globally and in the U.S. About 11 million U.S. residents had to relocate to another part of the country because of hurricanes, floods and wildfiresBy Chelsea Harvey & E&E News Aerial view of destroyed houses in Port St Lucie, Florida, after a tornado hit the area and caused severe damage as Hurricane Milton swept through Florida on October 11, 2024. Miguel J. Rodriguez Carrillo/AFP via Getty ImagesCLIMATEWIRE | Natural disasters and human conflict forced about 66 million people to flee within their own countries last year, and the United States led all nations with 11 million of these internal displacements, new research finds.That’s a record for the United States. And last year broke a global record, too, as 2024 saw the largest number of disaster-related internal displacements since experts began tracking the issue in 2008. About 46 million people worldwide had to relocate domestically last year because of natural disasters; another 20 million had to flee because of conflict or violence.Those are the latest statistics from the Internal Displacement Monitoring Centre, an international nongovernmental organization that keeps tabs on the number of people forced to flee within their countries each year because of natural disasters, extreme weather, conflict and violence.On supporting science journalismIf you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.IMDC publishes the previous year’s findings in an annual report each spring; its latest analysis was released Tuesday.The 46 million people forced to relocate domestically because of natural disasters last year represents a big jump from 2023, which saw 26 million people internally displaced because of events such as storms, floods, wildfires and droughts. In the United States, natural disasters were the sole reason 11 million residents had to relocate domestically.Cyclones triggered more than half of disaster-related movements worldwide last year. That includes Hurricanes Helene and Milton, which both caused tens of billions of dollars of damage in the United States. Floods accounted for another 42 percent of the disaster displacements, with events occurring on every continent last year.At least 8.3 million disaster-related movements occurred in the form of preemptive evacuations, the report found. Evacuations were reported in 53 of the 163 countries and territories reporting weather and disaster displacements in 2024. Many of these occurred in the United States; Hurricane Milton alone triggered around 6 million evacuations in Florida.Climate change is causing extreme weather events to intensify globally. At the same time, the report notes, other human factors are worsening the impact of these disasters, including inadequate infrastructure and urban expansion in disaster-prone areas.Low- and middle-income countries, on the whole, continue to be disproportionately affected by disaster displacements. And vulnerable communities within these countries tend to experience the worst outcomes.For instance, catastrophic flooding in the Brazilian state of Rio Grande do Sul last spring killed more than 180 people and triggered at least 775,000 displacements. Most occurred in the state capital of Porto Alegre, according to the IDMC report, and disproportionately affected Black and Indigenous communities.At the same time, conflict zones are increasingly overlapping with areas hit hard by natural disasters, many made worse by climate change. According to the report, the number of countries reporting both disaster and conflict displacements has tripled since 2009.The rise of these compounding crises points to the need for better preparation and resilience efforts, the report suggests.“Resolving displacement requires both immediate efforts to help people who have lost everything and investments to address underlying vulnerabilities, so people don't become displaced in the first place,” said IDMC Director Alexandra Bilak in a statement.More than 20 million people across 49 countries were internally displaced by conflict and violence last year, the report found. All but four of these countries reported disaster displacements as well.More than 5 million conflict displacements occurred in the Democratic Republic of the Congo, the most of any country worldwide. Heavy rainfall also triggered intense floods in the DRC last year, and many roads became blocked by water or by armed groups. The combination of these issues prevented aid organizations from assisting hundreds of thousands of displaced people, the report found.The report also notes that while some people are able to return home after disasters or conflict subside, many remain displaced for years. Around 65.8 million people were newly displaced in 2024 by both conflict and disasters. But millions of others already were living in displacement at the start of the year.At the end of 2024, about 83.4 million people were living in displacement worldwide. The vast majority, around 73.5 million, had been forced to move because of conflict — a 10 percent increase over the previous year.Still, disaster-related displacement is rising even faster. About 9.8 million people worldwide were living in displacement from disasters at the end of 2024 — a 29 percent increase over 2023.Reprinted from E&E News with permission from POLITICO, LLC. Copyright 2025. E&E News provides essential news for energy and environment professionals.

Author(s): Marie Originally published on Towards AI. Hallucinations in Healthcare LLMs: Why They Happen and How to Prevent Them Building Trustworthy Healthcare LLM Systems — Part 1 Image generated by the author using ChatGPT TL;DR LLM hallucinations: AI-generated outputs that sound convincing but contain factual errors or fabricated information — posing serious safety risks in healthcare settings. Three main types of hallucinations: factual errors (recommending antibiotics for viral infections), fabrications (inventing non-existent studies or guidelines), misinterpretations (drawing incorrect conclusions from real data). Root causes of hallucinations: probabilistic generation, training that rewards fluency over factual accuracy, lack of real-time verification, and stale or biased data. Mitigation approaches: Retrieval-Augmented Generation (RAG), domain-specific fine-tuning, advanced prompting, guardrails. This series: build a hallucination-resistant pipeline for infectious disease knowledge, starting with a PubMed Central corpus. Hallucinations in medical LLMs aren’t just bugs — they’re safety risks. This series walks through how to ground healthcare language models in real evidence, starting with infectious diseases. Introduction LLMs (large language models) are changing how we interact with medical knowledge — summarizing research, answering clinical questions, even offering second opinions. But they still hallucinate — and in medicine that’s a safety risk, not a quirk. In medical domains, trust is non-negotiable. A hallucinated answer about infectious disease management (e.g., wrong antibiotic, incorrect diagnostic criteria) can directly impact patient safety, so grounding models in verifiable evidence is mandatory. That’s why this blog series exists. This four-part series will show you how to build a hallucination-resistant workflow, step-by-step: Part 1 (this post): what hallucinations are, why they happen and how to build a domain-specific corpus using open access medical literature Part 2: Turn that corpus into a RAG pipeline Part 3: Add hallucination detection metrics Part 4: Put it all together and build a transparent interface to show users the evidence behind the LLM’s responses What Are Hallucinations in LLMs? Hallucinations are model-generated outputs that sound correct and coherent, but are not factually correct. They sound convincing but are often false, unverifiable or entirely made up. Why They Matter in Healthcare These errors can have serious implications, especially in clinical settings where they might lead to improper treatment recommendations. The wrong recommendation in clinical settings could have life or death consequences, which is why it is critical to mitigate these hallucinations by building transparent, evidence-based systems. Image generated by the author using ChatGPT Main Types of Hallucinations 1. Factual Errors Factual errors happen when LLMs make incorrect claims about verifiable facts. Using our infectious disease example, recommending antibiotics for influenza would be a type of factual error. 2. Fabrications Fabrications involve LLMs inventing non-existent entities or information. In the context of healthcare, for example, these could be fictional research studies, medical guidelines that don’t exist or made-up technical concepts. 3. Misinterpretations Misinterpretation happens when LLMs take real information but misrepresents or mis-contextualizes it. For example, a model might reference a study that exists, but draws the wrong conclusions Why LLMs hallucinate Large language models hallucinate because they don’t truly understand facts like humans do simply predict what words should come next based on patterns they’ve observed in their training data. When these AI systems encounter unfamiliar topics or ambiguous questions, they don’t have the ability to say “I don’t know” and instead generate confident-sounding but potentially incorrect responses. This tendency stems from several factors: Their training prioritizes fluent, human-like text over factual caution They lack real-time access to verified information sources They have no inherent understanding of truth versus fiction. Conflicting information in training data can push the model to average contradictory sources. The problem is compounded by limitations in training data that may contain outdated, biased, or inaccurate information, as well as the fundamental auto-regressive nature of how these models generate text one piece at a time. How Can We Address Hallucinations? There are various methods to mitigate or detect hallucinations. Mitigation Strategies Fine-tuning with Domain-Specific Data: The main reason for hallucination lies in knowledge gaps in the model’s training data. This approach helps by introducing domain specific knowledge and can be very powerful to create models that understand better the specialized medical terminology or various nuances in clinical text. Retrieval-Augmented Generation (RAG): This method allows the integration of external knowledge sources by retrieving relevant information before generating the answer. It helps by grounding the model outputs in verified external sources instead of relying only on the model’s training data. This is the method we will be focusing on in this series Other noteworthy strategies: advanced prompting methods like Chain-of-Thoughts or Few-Shot Learning can help mitigate hallucinations by guiding the model’s answer in the right direction. Rules-based guardrails that screen outputs before they reach users add another safety layer. Hallucination Detection Source-attribution scoring: This method compares the LLM answer to the retrieved documents to detect how much of the answer is grounded in the source. Beyond identifying hallucinations, it also allows to highlight the source behind the LLM answer, which helps building trust and transparency. Semantic Entropy Measurement: This method measures uncertainty about the meaning of generated responses and has been developed specifically to address the risk of hallucinations in critical areas involving patient safety for example Consistency-Based Methods: This method involves a self-consistency check, where hallucinations can be detected by prompting the model multiple times with the same query and comparing the outputs for consistency. Some interesting open-access publications to go a bit further: If you’re interesting in reading recent research on this topic, here’s a few research papers worth reading: Code Walkthrough: Downloading Medical Research from PubMed Central To reduce hallucinations in healthcare LLMs, grounding them in reliable medical literature is critical. Let’s start by building a corpus from one of the best sources available: PubMed Central (PMC). This script helps you automate the retrieval of open-access medical papers, making it easy to bootstrap a dataset tailored to your task (e.g., infectious diseases). Here’s how it works: 1. Setup and Environment import requestsimport xml.etree.ElementTree as ETimport jsonimport os, re, timefrom dotenv import load_dotenvload_dotenv()api_key = os.getenv("NCBI_API_KEY")email = os.getenv("EMAIL")base_url = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/" You’ll need to set your NCBI API key and email in a .env file. You can still call the NCBI API without an API key, but this unlocks higher rate limits and it is free 2. Search PMC Because we are interested in full texts to build our knowledge base, we should only download articles that are open access. To do so, we need to fetch the articles from PMC: # 1. Search PMCsearch_url = f"{base_url}esearch.fcgi"search_params = { "db": "pmc", "term": query, "retmax": max_results, "retmode": "json", "api_key": api_key, "email": email}print("Searching PMC...")search_resp = requests.get(search_url, params=search_params)search_resp.raise_for_status()ids = search_resp.json()["esearchresult"]["idlist"] This code queries PMC with your search terms (for example “infectious diseases”) and returns a list of document identifiers (PMCIDs). 3. Fetch and Parse Articles Now we can fetch the full texts using the PMCIDs: # 2. Batch fetchfetch_url = f"{base_url}efetch.fcgi"for i in range(0, len(ids), batch_size): = ids[i : i + batch_size] batch_ids fetch_params = { "db": "pmc", "id": ",".join(batch_ids), "retmode": "xml", "api_key": api_key, "email": email, } time.sleep(delay) r = requests.get(fetch_url, params=fetch_params) r.raise_for_status() Our response is an XML object, so the final step is to parse it and create a dictionary with the relevant sections: pmcid, title, abstract, full_text, publication_date, authors: root = ET.fromstring(r.content)for idx, article in enumerate(root.findall(".//article")): # Extract article details article_data = { "pmcid": f"PMC{batch_ids[idx]}", "title": "", "abstract": "", "full_text": "", "publication_date": "", "authors": [], } # Extract title title_elem = article.find(".//article-title") if title_elem is not None: article_data["title"] = "".join(title_elem.itertext()).strip() # Extract abstract abstract_parts = article.findall(".//abstract//p") if abstract_parts: article_data["abstract"] = " ".join( "".join(p.itertext()).strip() for p in abstract_parts ) # Extract publication date pub_date = article.find(".//pub-date") if pub_date is not None: year = pub_date.find("year") month = pub_date.find("month") day = pub_date.find("day") date_parts = [] if year is not None: date_parts.append(year.text) if month is not None: date_parts.append(month.text) if day is not None: date_parts.append(day.text) article_data["publication_date"] = "-".join(date_parts) # Extract authors author_elems = article.findall(".//contrib[@contrib-type='author']") for author_elem in author_elems: surname = author_elem.find(".//surname") given_names = author_elem.find(".//given-names") author = {} if surname is not None: author["surname"] = surname.text if given_names is not None: author["given_names"] = given_names.text if author: article_data["authors"].append(author) # Extract full text (combining all paragraphs) body = article.find(".//body") if body is not None: paragraphs = body.findall(".//p") article_data["full_text"] = " ".join( "".join(p.itertext()).strip() for p in paragraphs ) The data can then be saved into a jsonl that will be used in our next step — building our RAG system. Let’s be mindful of licensing restrictions: While open access literature allows anyone to access and read the content, it doesn’t mean the authors agreed to redistribution of their work. While this blog post and its content are intended for personal and educational use, if you decide to use this function to build a dataset that will be redistributed or commercialized, it is important to comply with the article’s license agreement. To do so, let’s define a function that will help us pull the license data from the downloaded article: def detect_cc_license(lic_elem): """ Inspect <license> … </license> for Creative Commons URLs or keywords and return a normalised string such as 'cc-by', 'cc-by-nc', 'cc0', or 'other'. """ if lic_elem is None: return "other" # 1) gather candidate strings: any ext-link href + full text candidates: list[str] = [] for link in lic_elem.findall(".//ext-link[@ext-link-type='uri']"): href = link.get("{http://www.w3.org/1999/xlink}href") or link.get("href") if href: candidates.append(href.lower()) candidates.append("".join(lic_elem.itertext()).lower()) # 2) search for CC patterns for text in candidates: if "creativecommons.org" not in text and "publicdomain" not in text: continue # order matters (most restrictive first) if re.search(r"by[-_]nc[-_]nd", text): return "cc-by-nc-nd" if re.search(r"by[-_]nc[-_]sa", text): return "cc-by-nc-sa" if re.search(r"by[-_]nc", text): return "cc-by-nc" if re.search(r"by[-_]sa", text): return "cc-by-sa" if "/by/" in text: return "cc-by" if "publicdomain/zero" in text or "cc0" in text or "public domain" in text: return "cc0" return "other" Here’s a short breakdown of what the licenses mean: Image uploaded by author Here’s the full function for PubMed download: def download_pmc_articles(query, ): max_results = 100, batch_size = 20, delay = 0.2, allowed_licenses = {"cc-by", "cc-by-sa", "cc0"}, out_file = "pmc_articles.jsonl" base_url = "https://eutils.ncbi.nlm.nih.gov/entrez/eutils/" # 1. Search PMC search_url = f"{base_url}esearch.fcgi" search_params = { "db": "pmc", "term": query, "retmax": max_results, "retmode": "json", "api_key": api_key, "email": email } print("Searching PMC...") search_resp = requests.get(search_url, params=search_params) search_resp.raise_for_status() ids = search_resp.json()["esearchresult"]["idlist"] # 2. Batch fetch fetch_url = f"{base_url}efetch.fcgi" skipped, saved = 0, 0 with open(out_file, "w") as f: for i in range(0, len(ids), batch_size): batch_ids = ids[i:i+batch_size] fetch_params = { "db": "pmc", "id": ",".join(batch_ids), "retmode": "xml", "api_key": api_key, "email": email } time.sleep(delay) r = requests.get(fetch_url, params=fetch_params) r.raise_for_status() root = ET.fromstring(r.content) for idx, article in enumerate(root.findall(".//article")): # Check license license = detect_cc_license(article.find(".//license")) if license not in allowed_licenses: skipped += 1 continue # skip disallowed license # Extract article details article_data = { "pmcid": f"PMC{batch_ids[idx]}", "title": "", "abstract": "", "full_text": "", "publication_date": "", "authors": [] } # Extract title title_elem = article.find(".//article-title") if title_elem is not None: article_data["title"] = "".join(title_elem.itertext()).strip() # Extract abstract abstract_parts = article.findall(".//abstract//p") if abstract_parts: article_data["abstract"] = " ".join("".join(p.itertext()).strip() for p in abstract_parts) # Extract publication date pub_date = article.find(".//pub-date") if pub_date is not None: year = pub_date.find("year") month = pub_date.find("month") day = pub_date.find("day") date_parts = [] if year is not None: date_parts.append(year.text) if month is not None: date_parts.append(month.text) if day is not None: date_parts.append(day.text) article_data["publication_date"] = "-".join(date_parts) # Extract authors author_elems = article.findall(".//contrib[@contrib-type='author']") for author_elem in author_elems: surname = author_elem.find(".//surname") given_names = author_elem.find(".//given-names") author = {} if surname is not None: author["surname"] = surname.text if given_names is not None: author["given_names"] = given_names.text if author: article_data["authors"].append(author) # Extract full text (combining all paragraphs) body = article.find(".//body") if body is not None: paragraphs = body.findall(".//p") article_data["full_text"] = " ".join("".join(p.itertext()).strip() for p in paragraphs) f.write(json.dumps(article_data) + "\n") saved += 1 print(f"Saved batch {i//batch_size + 1}") print(f"Downloaded {saved} articles to {out_file}, {skipped} articles removed by license filter") Now you can call your function with your query to create your corpus. For example: # Install packages if neededpip install python-dotenv requestsquery = 'bacterial pneumonia treatment'max_results = 500batch_size = 50download_pmc_articles(query, max_results, batch_size) And that’s it! Now all your articles are saved in a jsonl file and ready to be processed for RAG What’s Next: Preparing the Data for RAG In Part 2, we’ll take the domain-specific corpus you just built and use it to power a Retrieval-Augmented Generation (RAG) system — grounding your LLM in real evidence to reduce hallucinations and improve trust. Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor. Published via Towards AI

We may earn a commission from links on this page.When I bought my first robot vacuum, I thought my floors would thereafter be effortlessly spotless—endlessly patrolled by a 24/7 custodian, with absolutely no work required on my part. Today, there are never less than three robot vacuum cleaners installed in my small home at any given time. And yet I am definitely not doing nothing to keep my floors clean. While I’m certainly doing less daily upkeep, I would not say my floors are spotless, nor that I have been alleviated of all floor-related housework. Here’s what I wish I’d known about what robot vacuums can and can't do, and how much human effort they require. Roborock S8 MaxV Ultra Robot Vacuum and Mop $1,799.98 at Amazon Shop Now Shop Now $1,799.98 at Amazon Narwal Freo Z Ultra Robot Vacuum and Mop $1,499.99 at Amazon Shop Now Shop Now $1,499.99 at Amazon 3i S10 Ultra Robot Vacuum and Mop $1,399.99 at Amazon $1,899.99 Save $500.00 Shop Now Shop Now $1,399.99 at Amazon $1,899.99 Save $500.00 Saros 10 Robot Vacuum and Mop $1,599.99 at Amazon Shop Now Shop Now $1,599.99 at Amazon Saros 10R Robot Vacuum and Mop $1,618.98 at Amazon Shop Now Shop Now $1,618.98 at Amazon Narwal S20 Stick Mop $499.99 at Amazon Shop Now Shop Now $499.99 at Amazon SEE 3 MORE You'll have to rearrange your furniture The Freo Z navigating around an air purifier on the floor Credit: Amanda Blum When you hire someone to clean your home, you typically don’t move furniture around yourself to make their job easier,. For your robot vacuum, you might have to. While these robots have gotten much better at navigation, the less there is on the floor, the easier it will be for them to clean. Each table leg, chair and floor lamp is another obstacle to be worked around, and it’s worth knowing that most robots can’t get as close to table legs and floor lamps as you would be able to with a stick vacuum. Nor can they move a piece of furniture aside to get under it. As such, you will have to make some changes to your floor plan if you want to make the most of your new robot friend. More important than furniture relocation is the trend of having less on the floor, generally. If you like to let a laundry pile accumulate, have potted plants or a stack of books, you’ll end up relocating those items to give your robot better access. Corded items are particularly tricky, so the less cords on the floor, the better. Over time, I’ve learned to suspend the cords from the wall, and keep them wound as much as possible so nothing trails on the floor. I was recently quite impressed with the Roborock 10R, which relied solely on AI assisted cameras. The navigation skills of the robot were great, allowing it to venture into tight and low spaces that other robots hadn't. Robot vacuums can't clean near their docks the robot can't navigate around the Eureka tower, so debris accumulates around it Credit: Amanda Blum It's an unsurprising fact that robot vacuums can only clean where they can reach. Where they can’t reach, you’ll still see dirt and debris. All robot cleaners have a dock of some sort, and many now come with self emptying towers, which usually work exceptionally well at emptying the robots of dirt so you don’t have to do it manually. Given all of this automation, it's a bit ironic that the dirtiest place in your home might well be right around the tower or dock itself. As these towers are usually tucked into a corner or small space, detritus can easily accumulate around them, but the robot can’t move around the tower to clean it up. As the robot returns to the dock, it sweeps dirt with it, and it all eventually winds up at the foot of the tower.  This Roomba can't navigate into the small space underneath this plant stand, so instead it just kicks debris into the space. Credit: Amanda Blum Only the latest robots (like the 3i S10 Ultra) are at all good at cleaning around the base of pieces of furniture and thresholds between rooms. With most robots, you'll usually still find a band of dust and grime against the wall and around any items that the robot can’t easily clean. And then there are all the spaces too narrow or too low for the robot to slide into. While a few robots, like the Narwal Freo Z, have a fluffy pad that’ll occasionally rub against the base board to dust it, they don't do any real cleaning, so you’ll need to clean your baseboards manually, too. Again, you can rearrange your space to allow the robot more autonomy by making sure any spaces you want to be cleaned allow clearance of at least 1.5 times the width of the robot, and enough vertical height that the robot can reach it.This is always going to be a bit of a losing game, however, which is why I still have a stick vacuum I use to quickly suck up debris from narrow spots the robot can’t reach, the corners and edges that the robot can get close enough to, and against the wall and thresholds. Robots are best at vacuuming up only certain types of debrisThe same way you might organize before a visit from a cleaning professional, you’ll find yourself pre-sweeping for your robot. While robots are reliable at getting up dust and small detritus like rice or cereal, they’re not great at large debris, which means anything larger than a standard dice cube. Pieces of mulch, pet toy fluff, kitchen scraps, bobby pins, and pocket change will all get caught in a robot's rollers and need to be fished out. Pre-sweeping is a way to avoid that annoying task by getting up larger items of debris and leaving the rest for the bot. (But once again allow me to praise the 3i S10 Ultra, which has excelled at vacuuming up larger debris than any other robot vacuum I'd ever tested.)Even if your robot vacuum has a mop, you'll need to mop manually This Roborock is starting to get up the muddy pawprints, but it takes many passes Credit: Amanda Blum Most robot vacuums are now combo vacuums and mops.Some models are great at mopping—sometimes even better than they are at vacuuming. The docking tower empties the mop of dirty water, fills it with clean water, and sanitizes everything afterward. With the combined functions of vacuuming and mopping your floors will surely be clean, right? Not necessarily. Robot mops are great at cleaning up surface spills. What they’re less good at is scrubbing away stains and compound floor messes like mud or spilled food. While most robots can mostly get muddy paw prints up after a number of passes, doing so requires a lot of water, battery life, and time, and I’ve yet to find a single robot mop that is good at getting grime out of tile or wood floors. The robot simply can’t put enough pressure on the mop to really scrub the floor. For these reasons, you’ll need to keep an old-fashioned mop around, and you should expect to have to give the floor a good, hardy mop every few days and use real elbow grease to get out grime. Once the floor is clean, repeated patrols by the robot mop can help stave off new grime deposits, but they’re inevitable, so don’t assume your mopping days are over. If mopping is your priority, I recommend highly Roborock models that have a singular vibrating mop pad, versus twin spinning mops. The Roborock S8 MaxV Ultra remains one of my top recommendations. Robot vacuums are slowWith most robot vacuums working with voice assistants like Google, Alexa, or Apple, its well within the realm of possibility to simply ask the robot to “clean up around the high chair” after dinner (presuming you’ve set up a zone around the high chair, and it is always in the same approximate spot). If you’ve got such a routine set up, the robot will begin whatever pre-deployment practices it has, whether that’s filling the water resivior or orienting or cleaning the mop head, and eventually will amble over to the intended spot and proceed to take its time cleaning. Though still not particularly fast in getting to the spot you pin, Roborock's Pin and Go feature allows you to quickly place a pin on your floor map and the robot will proceed immediately to that spot to clean. This feature is available on all premium Roborock models, including the latest Saros 10 and 10R. Roborock lets you put a pin on the map for the robot to proceed to Credit: Amanda Blum If you’ve got the time to spare, great. But if your kid spills a cup of orange juice on the floor, you might not be willing to wait the five to 10 minutes it will take the robot to get to the spot in question and start cleaning when you could grab a mop or towel and be done in under a minute. For that reason, I'm still happy to have a Narwal S20, a newer model stick mop I can quickly grab, mop as needed, and then put back on the dock, all without having to get my hands dirty.  Robot vacuums occasionally require human interventionRobot vacuums can range wildly in price (from $200 to $2,500) and quality, and they are getting more intelligent and feature-rich all the time. Robots that have AI on board can actually learn about your home and get better at cleaning it over time. Despite all of that, however, your robot will need your intervention more than you’d like. Depending on how good a model you own, you will likely need to routinely unclog the rollers (the most common robot vacuum maintenance task). Your robot will also get stuck—sometimes because there is a new obstacle for it to navigate around, like a waylaid toy or sock, or a loose cord under the couch the robot gets stuck on it. Sometimes, it will simply gets lost. The bottom line is, even the best robots can’t be left on their own for weeks at a time. Most will require a human around eventually to get them back to the dock, clear the rollers, unclog it any clogs in the mopping system, and/or change the water in the dock. If all of this hasn't scared you away from getting one, your goal should be to choose a robot that matches your home and needs, and that enough features that you won't be following it around every day, making sure it does its job.

A few summers ago, when the culture was moving through micro trends as fast as they could be Instagrammed, Cottage Core was born. The trend, which came out of a Covid-influenced romanticism for living close to nature (but not ruffing it, à la gorpcore, fashion’s cousin trend), inspired an infusion of chintz and whicker-filled interiors, and, of course, lush English-style gardens.Flash forward to 2025. All those cottage gardens planted in early 2020—rustic, sophisticated, chic—are at their peak. And there really is something to an outdoor space that merges with the indoors, is there not? On a warm summer evening, when the bougainvillea is in bloom, and the grass is a bit damp, what could be more appealing than a home built to nestle into a fantastical garden.Here, we’ve collected some of our favorite cottage gardens. They range from fairy house gardens to campground landscape, historical (Anne Hathaway’s famed cottage that has inspired Shakespeare devotees the world over) to contemporary compound gardens in the woods.While we may have a specific notion of a cottage garden, but they are—and should be—as unique as the people who tend them. One lesson for planting your own? A small space is an asset rather than a limitation.Below, you’ll find 15 of our favorite cottage gardens from Marin County California to Stratford-Upon-Avon, England.William Jess LairdThis Amagansett cottage was literally designed for “summertime snoozes.” It’s also a good reminder that a delicate slate garden pathway can take you far. Designer Melissa Lee noted how “unexpected” the whole place felt, surrounded by the many mansions of the Hamptons. As she rightfully notes, the charm is in the surprise. A suggestion of mystery always adds to a cottage! Think The Secret Garden or the unexpectedly expansive Weasley Family home.Noe DewittVines climb up this 1920s English Art and Crafts style cottage in the Hamptons. The elegant and eclectic cottage was re-designed by Nick Olsen to emphasize outdoor living with comfy couches, a tiled patio and a pool.William James LairdThis pink cottage kitchen looks out over a garden in Litchfield County, Connecticut. Designer Clive Lonstein’s work is vibrant and unexpected, particularly for a modest Connecticut cottage built in the late 1800s. In a way though, the bright colors all throughout the house are a reflection of the original design for the house. The architect, Ehrick Rossiter was known for his own whimsy, and even included a turret in this design. This cottage is a great reminder to leave the door open all summer long.Stephen Kent JohnsonThis former fishing shack in Provincetown proves that a sprawling garden can fit into a small space. From Windex yellow fox gloves to arching lavender, this is a bucolic slice of heaven. A classic shingled home, complete with flower boxes and a white picket fence, it has a deeply cottage-core sequence backstory. It was used as an artist studio for William Maynard until his death in 2016, and when it was sold, prospective buyers were asked to write why they wanted to live there.Rachael SmithWe love an indoor / outdoor cottage garden. Ideally, you have a branch that grows through a window, like this one in Suzie de Rohan Willner’s English Country Garden. It is a charming marriage of dynamics: English and French, contemporary and historical, and, of course just as eclectic as a cottage should be. Willner notes, “The whole house is a collection of things from each period of my and my husband’s lives. I love to pick up bibs and bobs and it all comes together very happily.”Chronicle / Alamy Stock PhotoKate Middleton’s Adelaide Cottage conjures images of an Arthurian fantasy. The Wales family made this their Windsor home since 2023. Built in 1831 for Queen Adelaide (the German-born wife of William IV, who was the Uncle of Queen Victoria). It went through a transformative renovation in 2015 which left the historical decorations in tact. Fun fact: the Wales family pay market rent for their use of the home.Photo 12//Getty ImagesThe poet, actor, and playwright Anne Hathaway’s famed cottage and accompanying garden must have inspired her husband’s plays (that would be Shakespeare). This might be what comes to mind when you think of a cottage garden. Now open to the public daily, it was originally built more than 500 years ago, and is the site of Hathaway’s own birth in 1556.CostcoThis Costco (yes, Costco!) shed turned cottage is an ideal backdrop for your cottage garden fantasy. If you’re feeling very DIY this year, start here. Priced at $6,499, it measures 12’ x 24’ feet, a perfect amount of space for your own summer hide away or gardening shed.Richard PowersThis glass house is a reminder that a cottage garden doesn’t have to follow a prescribed style. The Amagansett cottage, originally built in 1960, is a marvel of mid-century design, an aesthetic reflected in the mod-furniture choices. Again, we love a stylistic mix in an updated cottage. Japanese Maple Trees complete the woodsy vibe.© David Hockney, Photo By Jonathan WilkinsonDavid Hockney illustrated his own cottage garden during the Pandemic. His drawing is illustrative of the benefits of an English garden: a bit wild, extremely lush, and more green than anything else. If we could, we’d jump right into this scene like Mary Poppins on a rainy day.John M. Hall for ELLE DecorHere’s a rule of thumb: trust Ina Garten. This cottage-like structure, on the grounds of the East Hampton home Garten shares with her husband Jeffrey, is perennially perfect. Note, too, the green and purple color scheme here. This is perhaps the dream cottage garden and something of a childhood playhouse. It has just enough space for a cozy chat and is a reminder that you can build your own little cottage on a very small plot of land.Photo 12//Getty ImagesMarie Antoinette’s Hamlet on the grounds of Versailles still sets the standard for the cottage garden with a thatched roof, hedges, and roses straight out of a fairy tale. During the former French Queen’s reign, her hamlet was used as a faux farm house, where she and her young daughter, Princess Marie Thérèse, would dress as idealized versions of French peasant farmers and milk cows. The interior, though, of this modest cottage, is appropriately grand with silk furnishings and canopy beds.Douglas FriedmanA garden that proves succulents and cottages are a match made in heaven. This one, in Marin County, California, adds a bit of desert flair. On the other side of this cottage is a water way and a perfect little dock for launching paddle boards. We love how the greens liven up this side of the house and create a completely different, almost modern desert-like, aesthetic. As with any great cottage garden, there is a distinctly transportive factor.Michael CliffordA light wood sauna and cold plunge on the grounds of Jenni Kayne’s Hudson Valley farmhouse are hidden behind shrubbery for a sense of privacy against a wide open landscape. We love the idea of adding a spa-like ambiance to a cottage garden as well as finding inventive ways to use the space. This is exactly where we want to be in the summer!Getty ImagesThis is sort of cheating, but Bunny Williams is a necessary inclusion! Williams’s Oak Spring Garden in Upperville, Virginia continues to inspire garden and cottage enthusiasts the world over. Rather than one cottage, the grounds of Williams’s large estate feature a guest cottage and a basket house, both of which are charming in the extreme.Dorothy ScarboroughDorothy Scarborough (she/her) is the assistant to the Editor in Chief of Town & Country and Elle Decor. 

Olfactory Design: Spaces That Awaken MemoriesSave this picture!Mi Pan Bakery / Concéntrico. Image © Apertura ArquitectónicaWith each of the more than 23,000 breaths we take per day, air travels through the respiratory system into the lungs, where vital gas exchange occurs: oxygen is absorbed, and carbon dioxide is expelled. This involuntary and essential act also triggers another, less visible but deeply impactful process—our perception of smell. As air passes through the nasal cavity, odor molecules come into contact with olfactory receptors located in the olfactory epithelium. These receptors send signals directly to the olfactory bulb, which is part of the limbic system—the area of the brain linked to memory and emotion. Far from being a secondary sense, smell acts as a direct bridge between the environment and our deepest emotional responses. Scents carry a unique power to evoke vivid memories, elicit comfort or immediate aversion, and influence our emotional state almost instantaneously. Save this picture!This ability of scents to shape spaces, atmospheres, and memories is the foundation of what is known as olfactory design. The primary goal is to provoke emotional responses in physical environments—a sensory strategy gaining traction in architecture, retail, hospitality, and even institutional settings. It goes beyond simply adding fragrance to a space; it involves the intentional selection of scents that align with a place's identity, elicit desired sensations in users, and often reinforce a brand's message or a project's purpose.This aligns with the concept of atmosphere as defined by German philosopher Gernot Böhme, who describes it as a spatial quality perceived sensorially—a diffuse presence that surrounds and affects those within a space. In this context, smell plays a crucial role: unlike vision, which requires focus and direction, olfaction acts diffusely and immersively. It fills the environment in an almost invisible but deeply tangible way, capable of altering mood, triggering memories, and preparing the body for alertness, relaxation, or stimulation. As highlighted in the article Touch It, Smell It, Feel It: Architecture for the Senses, contemporary architecture is increasingly recognizing that space should be experienced with the entire body—not just the eyes.Save this picture!Smell as an Architectural ToolOlfactory design is closely related to sensory design, a field that incorporates all five senses—sight, hearing, touch, taste, and smell—into spatial creation. Within this universe, multisensory marketing explores how coordinated sensory stimuli shape consumer perception, create emotional bonds, and influence decision-making. This sensory dimension is also addressed by architect and theorist Juhani Pallasmaa, who in The Eyes of the Skin, critiques the dominance of vision in contemporary architecture and advocates for a design approach that engages all senses. According to him, smell is a fundamental part of spatial experience, capable of evoking memories and emotions with unparalleled intensity. "Architecture evokes memories through smell more than through shape or color," he states.Save this picture!Save this picture!Classic examples illustrate this power: the scent of popcorn in cinemas evokes pleasure and encourages consumption; the smell of fresh bread in supermarkets conveys warmth and increases dwell time; or the universally recognized "new car smell." Scientific studies have also shown that aromas significantly influence human behavior by activating brain areas related to memory, emotion, and decision-making. Pleasant fragrances can extend time spent in commercial spaces, increase purchase intent, and even alter subjective perceptions of temperature, spaciousness, or cleanliness. In healthcare settings, the use of essential oils such as lavender and bergamot has been linked to reduced anxiety and stress relief in preoperative patients, as shown in randomized clinical trials. In offices, subtle fragrances have been correlated with increased concentration, productivity, and general well-being. These findings reinforce the potential of olfactory design as a strategic sensory tool for creating more welcoming, functional, and emotionally resonant environments.In an increasingly overstimulated retail environment, the sense of smell is emerging as a powerful and underutilized channel. Scent branding—the development of exclusive scents to reinforce a brand's identity—aims not only to differentiate but also to create lasting emotional connections with users. Some brands develop custom fragrances that become integral to the experience. Luxury hotels spray their exclusive scents on linens and curtains, creating a memory-laden atmosphere for guests. Airlines infuse specific aromas in cabins to reduce flight stress and enhance comfort. Fashion houses go further, registering proprietary scents as part of their sensory branding toolkit, akin to logos, slogans, or jingles.Save this picture!Save this picture!Technology, Immersion, and ResponsibilityTechnology has enabled increasingly smart and personalized use of scents. Digital diffusers allow for scheduled scent release based on time, crowd density, or room temperature—enhancing both efficiency and impact. Presence sensors, automation systems, and remote control via apps are becoming common in commercial and hospitality spaces. Olfactory design is also entering digital and immersive territories. Virtual reality, sensory exhibitions, and artistic installations have started incorporating scent as a narrative layer. In multisensory experiences, scent is used to intensify immersion, trigger targeted emotions, or signal spatial and temporal transitions.Save this picture!As olfactory design evolves, so do the challenges around sustainability and health. The composition of scents—natural or synthetic—is a critical concern. Natural aromas derived from flowers, woods, and spices are often more sustainable and less aggressive, though they can be expensive and less durable. Synthetic compounds, widely used in industry, offer consistency and control but raise concerns regarding toxicity, environmental impact, and allergic reactions.Save this picture!Save this picture!For these reasons, several countries have enacted regulations limiting the use of volatile organic compounds (VOCs)—substances commonly found in synthetic fragrances—in enclosed environments. These regulations are particularly strict in high-traffic or sensitive locations such as hospitals, schools, and public buildings, where indoor air quality is crucial to occupant health. Olfactory design, in this context, must go beyond the creation of immersive experiences. It must also take into account diverse user profiles, including those with chemical sensitivities or respiratory conditions, and meet technical standards for air quality, safety, and accessibility.Evoking memories, stirring emotions, or reinforcing identity—all of this can be achieved through a single breath. To incorporate smell into the creative process is more than simply adding a fragrance; it is an opportunity to expand architecture beyond shape, color, and light—touching what is most intimate in human experience. Image gallerySee allShow less About this authorEduardo SouzaAuthor••• Cite: Eduardo Souza. "Olfactory Design: Spaces That Awaken Memories" 13 May 2025. ArchDaily. Accessed . <https://www.archdaily.com/1029952/olfactory-design-spaces-that-awaken-memories&gt ISSN 0719-8884Save世界上最受欢迎的建筑网站现已推出你的母语版本!想浏览ArchDaily中国吗?是否 You've started following your first account!Did you know?You'll now receive updates based on what you follow! Personalize your stream and start following your favorite authors, offices and users.Go to my stream

Common personal care and beauty products like lotions, soaps, shampoos, eyeliner, and even eyelash glue can contain formaldehyde or preservatives that release formaldehyde—a known carcinogen that has been linked to cancer. And Black and Latina women could be at particular risk. Formaldehyde is a preservative (it’s a key ingredient of embalming fluid) and so it’s sometimes added to beauty products as a way to extend their shelf life and inhibit the growth of bacteria or mold. Formaldehyde-releasing preservatives are seen as an alternative to formaldehyde, but these chemical compounds do the same thing: they extend shelf life while slowly releasing formaldehyde into the product over time (just how much depends on multiple factors, but studies suggest longer storage times and higher temperatures lead to more formaldehyde released).  That formaldehyde could then be absorbed by the skin, and even though the amount may be small, experts say low levels of formaldehyde still pose health risks. Personal care products are often used frequently, so repeated exposures could add up. These chemicals have already been found in hair-straightening products, which are predominantly used by Black women. A new study, recently published in the journal Environmental Science & Technology Letters, found that this risk extends beyond chemical hair relaxers to all sorts of beauty products: lotions, shower gels, face creams, shampoo and conditioners, hair oils, eyeliner, eyelash glue, and so on.  In that study, researchers asked a group of Black and Latina women in Los Angeles about their use of personal care products over a week. More than half reported using items that contain formaldehyde-releasing preservatives—and many of those products are ones that the participants used daily, or multiple times in a week. Finding formaldehyde in beauty products For the study, 64 Black and Latina women were tasked with tracking all of their beauty product use, logging the information in an app developed by the Silent Spring Institute, a research organization focused on the environmental causes of breast cancer. (Silent Spring chemists authored the study, and it was part of a larger research effort between Silent Spring, Occidental College Black Women for Wellness, and Columbia University.)  That app also asked them to take a photo of each ingredient label, which allowed the researchers to analyze the ingredient lists for formaldehyde and formaldehyde-releasing preservatives. Formaldehyde-releasing preservatives often go by complex chemical names like 1,3-dimethylol-5,5-dimethylhydantoin, also called DMDM hydantoin or DMDMH, meaning they don’t actually appear as “formaldehyde” on ingredient lists.  Fifty-three percent of participants said they used at least one product with formaldehyde releasers on its ingredient label, and DMDMH was the most common. Of the items that contained any formaldehyde-releasing preservatives, DMDMH was in 47% of skincare and 58% of hair products.  The fact that these toxic chemicals are in so many products highlights the health risks women face, particularly Black and Latina women. One woman in the study used three products with formaldehyde-releasing preservatives: a leave-in conditioner, rinse-off conditioner, and a body wash. Some women used these products multiple times a day, like hand soap or lotion. Over a five day period, 20 study participants used lotions with formaldehyde-releasing preservatives for a total of 76 times. One eyelash glue even specifically listed formaldehyde as an ingredient. The preservatives were also found in hair gels, oils, curl creams, and edge controls, predominantly used by Black women. The study didn’t list specific brands or product names containing formaldehyde-releasing preservatives, though it did note that 12 such lotions were from Bath & Body Works. “While this study does not specify which of body lotions its participants were using, we rigorously test formulas for all our personal care and home fragrance products, including FRPs to meet regulatory and safety standards,” a Bath & Body Works spokesperson said in a statement. Protecting consumers from formaldehyde releasers Formaldehyde exposure is linked to adverse health effects, including increased risk of multiple types of cancer. Researchers say there’s been a growing concern about formaldehyde-releasing preservatives, and how personal care products that contain them could pose a risk to women’s health, particularly Black and Latina women. Previous studies have connected the use of hair relaxers to an increased risk of uterine cancer in Black women.  Others say these formaldehyde-releasing chemicals aren’t a concern. Unilever, for example, has a web page about how it doesn’t use formaldehyde as an ingredient but does use “formaldehyde donors” like DMDMH. It says they’re safe to use, per the U.S. Cosmetic Ingredient Review Expert Panel and Europe’s Scientific Community on Consumer Safety. Still, in Europe, products with formaldehyde are more regulated. The European Union has banned formaldehyde in cosmetics, and requires any cosmetics with formaldehyde releasers above a 0.001% concentration to have a warning label.  The U.S. currently doesn’t ban formaldehyde in cosmetics (a federal ban on formaldehyde and formaldehyde releasers in chemical hair straighteners was considered back in 2023, and is currently stalled after President Donald Trump paused all federal regulations). At least 10 states, including California and Oregon, have enacted or considered laws to regulate formaldehyde in cosmetics—either by banning it, or requiring warning labels for formaldehyde releasers. There has been a drop in products containing formaldehyde in California after the state’s Safe Cosmetics Program began in 2007, but experts say even beyond warning labels, banning formaldehyde releasers completely across the country would be the best-case scenario to reduce risks. The researchers suggest people avoid products containing DMDMH. Silent Spring has resources for how people can avoid formaldehyde releasers, including by noting the other chemical names for such preservatives that may appear on ingredient lists. Black Women for Wellness also has resources for consumers concerned about chemical exposure. “We’re trying to do the right thing,” Janette Robinson Flint, executive director of Black Women for Wellness, says in a statement. “But there needs to be more government oversight. We shouldn’t have to be chemists to figure out what kinds of products will make us sick.”