Last April, Facebook’s AI research lab (FAIR) announced and released as open source its BlenderBot social chat app. While the neophyte AI immediately proved far less prone to racist outbursts than previous attempts, BlenderBot was not without its shortcomings. For one, the system had the recollection capacity of a goldfish — any subject or data point the AI wasn’t initially trained simply didn’t exist in its online reality, as evidenced by the OG BB’s continued insistence that Tom Brady still plays for the New England Patriots. For another, due to its limited knowledge of current events, the system had a strong tendency to hallucinate knowledge, like a digital Dunning-Kruger effect. But the advancements BlenderBot 2.0 displays, which FAIR debuted on Friday, should make the AI far more sociable, knowledgeable, and capable.

While BlenderBot 1.0 could only maintain its memory for a single discussion, its successor can remember topics of conversation over the course of multiple talks that can take days, weeks or even months to complete thanks to the implementation of a long-term memory module. What’s more, the AI can actively update its knowledge base by searching the internet for the latest news and details on any subject that the user wishes to speak about.

“BlenderBot 2 queries the Bing API for search results based on a generated search query, and conditions its response on the top few results,” Kurt Shuster, Research Engineer at Facebook AI, told Engadget. “We rely on Bing to provide high quality search results.” As such, BlenderBot 2.0 is now capable of speaking coherently about breaking news and new media, not just the data it was trained upon.

FAIR

“BlenderBot 2 is limited only by what a powerful search engine can provide,” Jason Weston, Research Scientist at Facebook AI, added. So for example, if you are more interested in learning about Tom Yewcic (the Patriot’s combo QB/Punter from the 1962 season) than you are about Tom Brady, BB 2.0 has you covered. It’s the same with more scholarly subjects, like photosynthesis or redox reactions, Weston continued. So long as the information is available on the web, “there is no reason BlenderBot 2 cannot discuss this.”

By actively searching the internet for information, BlenderBot 2.0 can also reduce the instances in which it hallucinates knowledge. “Providing the system with more commonsense reasoning will allow BlenderBot to make sure it does not confuse subtle concepts,” Weston explained, “such as a movie director versus a producer or a pitching coach versus a hitting coach.”

FAIR

The only wrinkle really occurs when discussing non-english based media, such as Demon Slayer: Kimetsu no Yaiba. “It is reasonable to conclude Bing will surface information about it and BlenderBot 2 can use that information accordingly,” he said. “We currently focus on english-based search results, so non-english references may not be fully covered.” The system will, however, recognize that Demon Slayer is of interest to you and will be more likely to bring up manga-centric subjects in future discussions.

FAIR has taken multiple steps to ensure that BlenderBot does not become the next Tay. “BlenderBot 2 does not learn directly from user input, as Tay did,” Shuster said. “We have taken extensive safety steps to ensure that BlenderBot 2 can handle adversarial users. Specifically, we employ both baked-in and two-stage techniques. BlenderBot 2 can detect itself if the incoming context will result in an offensive response, and additional safety layers where a safety classifier can detect if either the user input or the bot's output is offensive. Each handles the response appropriately.”

And while the system is currently focused on chewing its way through the English language corpus, FAIR does see BlenderBot does eventually extend to other languages as well. “While not in our immediate plans, the goal of our team is to build a superhuman conversationalist,” Shuster said. “This kind of agent requires multilingual understanding.”

Recent internal benchmarking processes found that BlenderBot 2.0 outperformed its predecessor by 17 percent in its engagingness score and 55 percent in its use of previous conversation sessions according to human evaluators, per a Friday blog from FAIR. What’s more, BlenderBot's rate of knowledge hallucinations dropped from 9 percent (!) in BB 1.0 down to just 3 percent in the current iteration.

Looking ahead, “humans interacting with AI systems via discourse is the future of AI,” Weston asserted, “and ensuring that humans have an engaging, informative experience is critical to that future. BlenderBot 2 combines the engagingness of BlenderBot 1.0 with the knowledge capabilities of a system with access to the entire internet, so ostensibly we are on the right track.”

On May 5, 1961, Commander Alan Shepard piloted his Mercury Freedom 7 spacecraft to a soaring height of 116 miles above the planet's surface to become the first American to reach Earth's orbit. This past Sunday, Sir Richard Branson was escorted to an altitude of 50 miles aboard Virgin Galactic's SpaceShipTwo Unity. Somehow, these men are now both considered astronauts.

The discrepancy here lies in the fact that since the Space Era began, the world's exo-planetary powers have never really gotten around to formalizing where the Earth's atmosphere ends and where "space" — loosely defined as it is — begins. Even within the US federal bureaucracy, different agencies use different standards. What NASA mission control considers the edge of space is actually 26 miles farther out than where the NOAA and US Air Force demark the atmospheric boundary. So the next time you find yourself hurtling through the Mesosphere, keep a close eye on your altimeter if you want to earn the coveted Astronaut Badge.

Clubhouse, the oftcloned invite-only audio chat platform that recently expanded to Android, has since its inception lacked a seemingly fundamental feature present on virtually every other social media app: the ability to directly message other users. But no longer! On Wednesday, the company announced that the Clubhouse app will now support DMs in the form of the new Backchannel messaging feature.

Backchannel will function as both a private messaging system for both 1:1 interactions and group chats. With it, Clubhouse envisions users employing it to coordinate privately among a channel's co-hosts, take sensitive questions or privately thank participants outside of the general group discussion, or just audibly shoot the breeze with your buddies. 

The concept of space travel was so new to us that when President Kennedy issued his famous moonshot speech, not even NASA's top scientists were completely sure we could actually land on the lunar surface. Some thought any craft that set down there would simply sink into the moon's regolith like it was a massive, airless pit of quicksand! In his latest book, Across the Airless Wilds: The Lunar Rover and the Triumph of the Final Moon Landings, journalist and former Fulbright fellow, Earl Swift, examines the oft ignored Apollo 15, 16, and 17 missions, our last trips to the Moon's surface (at least until the Artemis project takes place). In the excerpt below, Swift takes the reader on a tour of the JPL's hyper-rigorous, tread-shredding lunar test course and the battle for rover supremacy waged there between GM and Bendix.

Custom House

From the book ACROSS THE AIRLESS WILDS: The Lunar Rover and the Triumph of the Final Moon Landings by Earl Swift. Copyright © 2021 by Earl Swift. From Custom House, a line of books from William Morrow/HarperCollins Publishers. Reprinted by permission.

All through 1962 and into 1963, both GM and Bendix kept an eye on the Surveyor program. Sure enough, come summer, the Jet Propulsion Laboratory laid out its requirements for a hundred-pound, remote-controlled rover that it wanted to stash aboard the landers. The vehicle would explore the lurrain up to a mile from the Surveyors, while its drivers back on Earth steered it with television eyes. The laboratory alerted companies planning to bid on the phase 1 design study—the normal first stage of any new hardware program—that they’d be expected to supply engineering models of their concepts. Proposals were due in seven weeks.

The short deadline weeded out the dilettantes. In October the two companies left standing—GM and Bendix—started work under contract. GM was ready with its six-wheeled design. Its Surveyor lunar roving vehicle was six feet long on eighteen-inch wheels and weighed ninety pounds—half the size and half again as heavy as its test bed, with a sure-footedness that was no less jaw-dropping. On Pavlics’s “lunarium” of rocks, craters, and slopes outside the Santa Barbara lab, it climbed forty-five-degree inclines, leapt twenty-inch crevasses, and bent its way up and over thirty-inch steps.

Bekker and Pavlics had been working on the idea for more than three years by then. Their main advancement this time: the wheels. Again, they were made of wire, but it was knotted into a wide mesh that resembled chain-link, and shaped into fat doughnuts. Like the team’s earlier wire tires, they deflected when they hit an obstacle and absorbed some of the bumps of cross-country travel. They worked with or without a fabric covering.

“We had a big program to try to come up with the wire material that would survive the vacuum environment on the moon,” John Calandro recalled. “Frank had devised a testing device that created the vacuum environment we needed.”

When fully geared up for a mission, the rover would be an electronic wonder, with subsystems supplied by RCA Astro-Electronics and by AC Electronics, a GM division in Milwaukee: it would have a stereo TV imaging rig, sophisticated navigation and control, and silver-zinc batteries recharged by solar panel. But Santa Barbara’s part of the job, the vehicle itself, was a study in doing more with less. The hardware was constantly “assessed to see if something simpler might be able to do the same job,” designer Norman J. James would remember. “‘The part that’s left off never breaks’ was an often-repeated phrase.”

Bendix took a radically different approach. Its SLRV was a squarish, two-part, articulated robot, with curving, shock-absorbing legs at its corners that ended in small caterpillar track assemblies. The tracks pitched independently to follow uneven ground. Its handlers steered it with commands to slow, speed up, or reverse the tracks on one side or the other, and the pivot linking the two halves did the rest. On the moon, it would be powered by a radioisotope thermal generator—a small nuclear device—hanging off the back, and bristle with scientific instruments and antennas. It weighed one hundred pounds.

Side by side with the GM model, the Bendix machine looked bulky and awkward, and those tiny tracks didn’t seem much of a match for Pavlics’s nearly spherical wire wheels. But Bendix was bullish on its design right up to the day in May 1964 when a panel from the U.S. Geological Survey, Caltech, and NASA took the two models to a volcanic field north of Flagstaff, Arizona, and turned them loose on the rugged Bonito Lava Flow. “We had one little section where they could really get into some pretty rough stuff,” the Geological Survey’s Jack McCauley recalled years later. “The GM vehicle was perfect. It got from point A to point B without any mishaps or turning over.

“The poor Bendix vehicle had tanklike treads that were made of some kind of rubber-type thing,” McCauley said. “The vehicle just started shredding the treads. In fact, when they finished halfway down the course, it had no treads left. So, the GM thing obviously got our blessing.”

General Motors had scored a decisive victory. Unfortunately, it didn’t add up to a rover on the moon. The “Rover Boys,” as that panel of testers came to be known, were mightily impressed with the six-wheeler, but its capabilities didn’t square with the Jet Propulsion Laboratory’s requirements: namely, to “go around and take pictures every ten meters, and also to use a penetrometer to see what the strength of the lunar soil was—and to do it in a preordained manner,” McCauley said. “Basically, just do a grid survey.” Bendix had produced too little rover for the mission; GM had produced too much. The Rover Boys reluctantly reported that neither rover matched the Surveyor program’s stated needs, and that was among the reasons that NASA scrubbed the rover component not long after.

By that time, JPL’s Ranger program had finally given NASA its first close looks at the moon. By design, they were fleeting glimpses: Ranger probes crashed into the lunar surface while taking high-resolution photos right up to the moment of impact. Conceived in 1959, the program had, at times, seemed another exercise in frustration. After Rangers 1 and 2 made two development test voyages in 1961, along came Rangers 3 through 6, all of which were busts. It wasn’t until July 1964, and Ranger 7, that the program literally hit pay dirt. As the spacecraft fell toward the moon, its cameras kicked on, and, for some seventeen minutes, it took and transmitted photographs of the approaching surface—4,316 images in all, some of them at a resolution hundreds of times greater than the best taken from Earth. The photos didn’t put to rest the fears inspired by Thomas Gold’s writings and lectures, but they did establish that the maria were smooth enough for a landing.

Volvo has made no secret of its plans to go green by 2030, first unveiling the XC40 Recharge SUV in 2019, then partnering with the Geely corporation to develop its performance EV Polestar line. And come early next year, the XC will be joined by a sleeker, curvier sibling dubbed the C40 Recharge. The C stands for coupe.

Understand, the XC40 and C40 are very much the same vehicle, at least under the hood. The two — along with the Polestar 2 — all share the same Compact Modular Architecture (CMA) platform that Volvo plans to build its future EV fleet on top of. As such, the C40 and XC40 offer literally identical performance profiles. They share a 78 kWh (75 kWh effective) battery pack which produces 408 HP and 486 lb-ft of torque across all four wheels, giving both cars a 0-60 time of 4.7 seconds, a 112 mph top speed and an estimated 210 miles of range.

Volvo

And, like its predecessor, the C40 Recharge will do so at rates up to 150kW on an L3 DC charger, enabling it to refill its power cells from basically dead to 80 percent capacity in 40 minutes. The C40 Recharge can also accept power from Level 2 (220V) sources, requiring around 8 hours to fully charge. You can, technically, charge the C40 on a standard 110V outlet — Volvo includes an adapter for doing so standard — but the company is positioning that charging level as more of a trickle-charge, topping-off option than one for actually, fully refilling a completely depleted battery.

As a Volvo rep explained to Engadget on Wednesday, the company envisions drivers using Level 3 DC fast charging stations located along their commutes more as quick recharge points — stopping for 5-10 minutes at a time, grabbing a cup of coffee as they wait — while using an in-home L2 charger to fully replenish the charge overnight, like an drivable cell phone.

Volvo

On the outside, the C and XC are easily discernible. While the XC40 adheres to the classic tenets of SUV styling, the C actually stands around 3 inches shorter overall and features a broadly curved roofline that falls away into an upturned spoiler — resulting in the coupe designation. I for one am smitten with the styling, especially the Fjord Blue paint scheme, which mimics the color of Sweden's local waters, as well as the all-glass roof.

The interior is even more impressive. For one thing, you won’t find a speck of leather in there. The floor mats are produced from recycled water bottles, as are the startlingly realistic faux-suede seats. “It's a very practical, sustainable solution, trying to get us away from traditional luxuries,” Volvo’s design rep told Engadget. “I think our future of luxury is more about the simplicity of something. Not, how many layers of wood and how many buttons you can have, it's more about the experience.” One unique aspect of that experience are the highlight panels that run throughout the C40 Recharge’s cabin, which depict topographical features of a Swedish national park.

Volvo

The cabin itself is quite minimalist though you’ll find a host of storage spaces subtly placed around the front seats with slick holders for everything from travel mugs to credit cards. The dashboard consists of the front-and-center Android Auto infotainment system, a series of physical buttons and knobs controlling the audio playback, front and rear defrost, and hazards sit just below. While I personally am a fan of tactile controls, C40 drivers won’t have a whole much use for them on account of the ever-present Android Assistant. You’ll be able to control the stereo, make calls, send texts, adjust the climate controls and even turn on the heated steering wheel. The Assistant’s knack for locating and evaluating charging stations along your route should prove especially helpful to range-wary EV adopters, Volvo reps explained on Wednesday, by not only alerting drivers to where these stations are but also what kind of connections they offer and the status of the vehicle’s battery once it arrives.

Volvo has yet to officially announce its MSRP for the C40 Recharge so it’ll be interesting to see how it might compare to its expected competition, assuming the C40 ends up being priced roughly around $54,000 like its XC predecessor. For example the Model Y Long Range starts from $52,490 and gets a 100 miles-plus more distance using an equivalently sized battery pack. The Audi Q4 e-tron on the other hand manages to achieve the same range on a surprisingly tiny 52 kWh pack. To be fair though, its 0-60 is 9 seconds flat and they’re only for sale in Europe for the moment. Then you’ve got the ID.4 which starts at $40,000 and boasts 50 miles more range but, in my opinion at least, doesn't offer quite the same level of refinement that I saw in the C40 Recharge.

Volvo

The C40 is expected to hit US streets in the first quarter of 2022 but it will not be available for sale through Volvo dealerships. You’ll be able to see them at the dealership, sure, as well as test drive them, pick yours up from there and get it serviced there if you buy one. However the purchase process itself happens exclusively online. You can reserve one today for $500 at the Volvo website.

This is but the second step in Volvo’s efforts to transition to EVs. The company plans to release a new electric model every year until 2025 as part of its larger goal of becoming completely carbon neutral by 2040. Rumored to be coming next: a fully-electric XC90 Recharge.

Volvo has made no secret of its plans to go green by 2030, first unveiling the XC40 Recharge SUV in 2019, then partnering with the Geely corporation to develop its performance EV Polestar line. And come early next year, the XC will be joined by a sleeker, curvier sibling dubbed the C40 Recharge. The C stands for coupe.

Understand, the XC40 and C40 are very much the same vehicle, at least under the hood. The two — along with the Polestar 2 — all share the same Compact Modular Architecture (CMA) platform that Volvo plans to build its future EV fleet on top of. As such, the C40 and XC40 offer literally identical performance profiles. They share a 78 kWh (75 kWh effective) battery pack which produces 408 HP and 486 lb-ft of torque across all four wheels, giving both cars a 0-60 time of 4.7 seconds, a 112 mph top speed and an estimated 210 miles of range.

Volvo

And, like its predecessor, the C40 Recharge will do so at rates up to 150kW on an L3 DC charger, enabling it to refill its power cells from basically dead to 80 percent capacity in 40 minutes. The C40 Recharge can also accept power from Level 2 (220V) sources, requiring around 8 hours to fully charge. You can, technically, charge the C40 on a standard 110V outlet — Volvo includes an adapter for doing so standard — but the company is positioning that charging level as more of a trickle-charge, topping-off option than one for actually, fully refilling a completely depleted battery.

As a Volvo rep explained to Engadget on Wednesday, the company envisions drivers using Level 3 DC fast charging stations located along their commutes more as quick recharge points — stopping for 5-10 minutes at a time, grabbing a cup of coffee as they wait — while using an in-home L2 charger to fully replenish the charge overnight, like an drivable cell phone.

Volvo

On the outside, the C and XC are easily discernible. While the XC40 adheres to the classic tenets of SUV styling, the C actually stands around 3 inches shorter overall and features a broadly curved roofline that falls away into an upturned spoiler — resulting in the coupe designation. I for one am smitten with the styling, especially the Fjord Blue paint scheme, which mimics the color of Sweden's local waters, as well as the all-glass roof.

The interior is even more impressive. For one thing, you won’t find a speck of leather in there. The floor mats are produced from recycled water bottles, as are the startlingly realistic faux-suede seats. “It's a very practical, sustainable solution, trying to get us away from traditional luxuries,” Volvo’s design rep told Engadget. “I think our future of luxury is more about the simplicity of something. Not, how many layers of wood and how many buttons you can have, it's more about the experience.” One unique aspect of that experience are the highlight panels that run throughout the C40 Recharge’s cabin, which depict topographical features of a Swedish national park.

Volvo

The cabin itself is quite minimalist though you’ll find a host of storage spaces subtly placed around the front seats with slick holders for everything from travel mugs to credit cards. The dashboard consists of the front-and-center Android Auto infotainment system, a series of physical buttons and knobs controlling the audio playback, front and rear defrost, and hazards sit just below. While I personally am a fan of tactile controls, C40 drivers won’t have a whole much use for them on account of the ever-present Android Assistant. You’ll be able to control the stereo, make calls, send texts, adjust the climate controls and even turn on the heated steering wheel. The Assistant’s knack for locating and evaluating charging stations along your route should prove especially helpful to range-wary EV adopters, Volvo reps explained on Wednesday, by not only alerting drivers to where these stations are but also what kind of connections they offer and the status of the vehicle’s battery once it arrives.

Volvo has yet to officially announce its MSRP for the C40 Recharge so it’ll be interesting to see how it might compare to its expected competition, assuming the C40 ends up being priced roughly around $54,000 like its XC predecessor. For example the Model Y Long Range starts from $52,490 and gets a 100 miles-plus more distance using an equivalently sized battery pack. The Audi Q4 e-tron on the other hand manages to achieve the same range on a surprisingly tiny 52 kWh pack. To be fair though, its 0-60 is 9 seconds flat and they’re only for sale in Europe for the moment. Then you’ve got the ID.4 which starts at $40,000 and boasts 50 miles more range but, in my opinion at least, doesn't offer quite the same level of refinement that I saw in the C40 Recharge.

Volvo

The C40 is expected to hit US streets in the first quarter of 2022 but it will not be available for sale through Volvo dealerships. You’ll be able to see them at the dealership, sure, as well as test drive them, pick yours up from there and get it serviced there if you buy one. However the purchase process itself happens exclusively online. You can reserve one today for $500 at the Volvo website.

This is but the second step in Volvo’s efforts to transition to EVs. The company plans to release a new electric model every year until 2025 as part of its larger goal of becoming completely carbon neutral by 2040. Rumored to be coming next: a fully-electric XC90 Recharge.

Moderna has injected its mRNA-derived vaccine for the seasonal flu into a human volunteer for the first time as part of a Phase 1/2 clinical study, the company announced on Wednesday. 

This is a very early test for the new vaccine technology, geared primarily towards building a baseline understanding of the treatment's "safety, reactogenicity and immunogenicity," according to a Moderna release. mRNA-1010, as the vaccine has been dubbed, is designed to be effective against the four most common strains of the virus including, A H1N1, H3N2, influenza B Yamagata and influenza B Victoria. According to the World Health Organization, these strains cause between 3 and 5 million severe cases of flu every year, resulting in as many as 650,000 flu-related respiratory deaths annually. In the US alone, roughly 8 percent of the population comes down with the flu every winter. The company hopes this vaccine will prove more potent than the current 40 - 60 percent efficacy rate of conventional flu vaccines.  

“We are pleased to have begun this Phase 1/2 study of mRNA-1010, our first mRNA seasonal flu vaccine candidate to enter the clinic. We expect that our seasonal influenza vaccine candidates will be an important component of our future combination respiratory vaccines,” Stéphane Bancel, Moderna's CEO said. “Respiratory combination vaccines are an important pillar of our overall mRNA vaccine strategy. We believe that the advantages of mRNA vaccines include the ability to combine different antigens to protect against multiple viruses and the ability to rapidly respond to the evolution of respiratory viruses, such as influenza, SARS-CoV-2 and RSV. Our vision is to develop an mRNA combination vaccine so that people can get one shot each fall for high efficacy protection against the most problematic respiratory viruses."

This vaccine has been generated using the same genomic techniques the company utilized to develop its COVID-19 treatment in 2020. The technique works by exploiting the human body's own cells to reproduce snippets of viral DNA to instigate an immune response and prime the body against future infection. Since this method doesn't require the entire virus (either weakened or dead) but rather just a birt of its genetic code, mRNA vaccines could be applied to any number of deadly modern diseases including malaria, TB — even cancer.  

COVID vaccination efforts continue to gain momentum both in the US and around the world but that doesn’t mean we’re out of the pandemic woods just yet. Regular testing remains an important factor in helping slow the spread of the disease but has typically required a trip to your doctor or local clinic. Luckily, that’s no longer the case. In recent months, the FDA has approved a number of COVID tests which can be administered in the comfort of your own home and return results in a matter of minutes rather than days.

If you’ve taken an in-person COVID test in the last year, your nasal swab sample was likely diagnosed using a real time reverse transcription–polymerase chain reaction (real time RT–PCR), which is among the most accurate and widely used lab-based methods for detecting viral pathogens such as Zika, Ebola and coronavirus.

The coronavirus only contains single strand RNA for genetic material which, unlike DNA, cannot be independently reproduced without the assistance of a host. As such, the virus must infect and repurpose healthy cells in order to make more of itself. RT–PCR mimics this process by first converting any coronavirus RNA present in a given sample into DNA — hence “reverse transcription” — then creating billions of copies of the genetic material and marking them with a fluorescent dye for identification.

The process is an offshoot of the more general PCR method, which is used to detect pathogens whose genetic material comes in DNA form and therefore doesn’t need amplification. This process is sensitive and highly accurate, albeit time-consuming, but does enable pathologists to detect a coronavirus infection in its early stages since only a minute amount of initial RNA is required. While the RT–PCR technique poses a low chance of outside contamination, its capability is limited in that it can only spot the coronavirus when it’s currently present in a sample. This method cannot tell if someone has been previously infected.

Some at-home tests rely on a similar process called isothermal amplification. Like PCR, isothermal amplification generates numerous copies of the coronavirus’ genetic material to aid in detection. Though IA is not as sensitive as the lab-based PCR method, it’s more accurate than other at-home tests which look for antigens — bits of coronavirus proteins that provoke the body’s immune response.

For example, an antigen-based BinaxNOW test correctly detects the virus only about 64 percent of the time in symptomatic people. That figure drops to just 35 percent in asymptomatics. Because antigen tests don’t include a reverse transcription phase, they are faster and less expensive to perform than PCR- and IA-based methods but are less accurate and return higher rates of false negatives — especially among people who have only recently been exposed.

The antigen method provides more of a general estimate as to how contagious you are, Dr. Gigi Gronvall of Johns Hopkins University told the NYT. “If you test positive on that, you really need to isolate,” she said.

So, if you are asymptomatic, a PCR or similar molecular-based test should be sufficient. If you suddenly find that you can’t smell or taste anything, pick up either a PCR or the most sensitive antigen-based test you can find. If you test positive, regardless of the type of test you use, isolate yourself immediately and call a doctor.

So far, only a handful of at-home tests have earned Emergency Use Authorization (EUA) from the FDA. EUAs are otherwise-unapproved treatments, tests or medical countermeasures that have been allowed to sidestep the formal (and lengthy) FDA approval process in response to life-threatening health crises like the one we are currently facing.

The Ellume COVID-19 Home Test is the first at-home test available without a prescription. It is antigen-based and takes 15 minutes to return a result. You’ll have to swab your nose then drop the sample in a desktop analyzer along with some processing liquid. Once the device does its thing, the results are transmitted via Bluetooth to your smartphone. That data is also shared through a secure, HIPAA compliant cloud connection to health authorities to aid in outbreak mapping. The test is expected to cost around $30 though the company has yet to announce where and when it might become available. The Ellume test displayed 96 percent accuracy in people ages 2 and up during its US clinical trial.

The Biden administration recently announced a $231.8 million deal with the Australian company to purchase 8.5 million units of the test. Ellume “will be delivering 100,000 tests per month from the Australian manufacturing facility until the U.S. facility is built,” a company representative told NPR. “At full capacity, the U.S facility will be able to produce up to 19 million tests per month. The 8.5 million tests for the US government is a portion of the overall manufacturing." However, not all health professionals are excited about this public-private partnership, with one going so far as to characterize the effort as “a spit in the ocean.”

The Cue OTC Test will also soon be available over the counter but unlike the Ellume, it uses a nucleic acid amplification test (similar to PCR) for improved accuracy and generates results in 20 minutes. Per a recent press release, “in prospective studies to evaluate the use of the Cue OTC Test, the results were 97.4% agreement for positive cases and 99.1% agreement for negative cases compared to the results from a highly sensitive EUA PCR laboratory-based test.” The company has not yet announced pricing or availability.

Abbott's BinaxNOW COVID-19 Ag Card Home Test isn’t just a mouthful to say, it’s a whole process. After answering a series of screening questions via the eMed digital health website, Abbot will deliver a test to your home. You’ll then be connected with a “telehealth professional” who will guide you through the sample collection process. Once you drop your sample into the analyzer, you should get your results in about 15 minutes through the Navica smartphone app. The Abbot test costs $25 and is only available with a prescription.

The Lucira COVID-19 All-In-One Test Kit costs $50 and, like the Cue, uses molecular tech. As with BinaxNOW, it requires a prescription so your doctor will have to order it on your behalf. The Lucira test is intended for use by people ages 14 and up, it returns results in 11 to 30 minutes.

If you don’t mind waiting a bit for your results, Amazon is selling Dxterity mail-in tests. Each one costs $110 and requires you post it to the company’s LA-area lab for diagnosis. On the plus side, this test is saliva-based so you won’t have to impale your nasal cavity with a Q-Tip to collect your sample. Results will be available within 24 - 72 hours through the company’s web portal.

Similarly, the CRL Rapid Response COVID-19 Saliva Test can be purchased from the Walgreens website for $119 and returns results in a scant 24 - 48 hours. If you test positive, the company will have a telemedicine rep reach out to discuss the diagnosis and potential next steps.

Though all of the fully at-home tests showed 90 percent-plus accuracy during their clinical trials, those figures can fluctuate when regular folks perform the tests themselves, so don’t treat a negative result as some license to get out and mingle. "When manufacturers are preparing data to submit to FDA, they are conducting studies under very specific, highly controlled conditions that optimize the performance of the test,” Kelly Wroblewski, director of infectious disease programs at the Association of Public Health Laboratories, told the AARP.