Now there’s another idea joining the fray, and it carries the added benefit of playing a role in the renewable energy transition: It’s a tiny house made from the nacelle of a decommissioned wind turbine.

The house, unveiled last month as part of Dutch Design Week, is a collaboration between Swedish power company Vattenfall and Dutch architecture firm Superuse Studios. Wind turbines typically have a 20-year lifespan, and Vattenfall is looking for novel ways to repurpose parts of its turbines. With the first generation of large-scale turbines now reaching the end of their useful life, there will be thousands of nacelles (not to mention blades, towers, and generators) in search of a new purpose.

Blades, towers, and generators are the parts of a wind turbine that most people are familiar with, but not so much the nacelle. The giant rectangular box sits at the top of the turbine’s tower and houses its gearbox, shafts, generator, and brake. It’s the beating heart of the turbine, where the blades’ rotation is converted into electricity.

Though it’s big enough to be a tiny house, this particular nacelle is on the small side (as far as nacelles go). It’s 10 feet tall by 13 feet wide by 33 feet long. The interior space of the home about 387 square feet, or the size of a small studio apartment or hotel room. The nacelle came from one of Vattenfall’s V80 turbines, which was installed at an Austrian wind farm in 2005 and has a production capacity of two megawatts. Turbine technology has come a long way since then; the largest ones in the world are approaching a production capacity of 15 megawatts.

Though there will be larger nacelles available, Superuse Studios intentionally chose a small one for its prototype. Their thinking was, if you can make a livable home in this small of a space, you can definitely make a livable home—and add more features—in a larger space; better to start small and grow than start big then downsize.

Though the house is small, its designers ensured it was fully compliant with Dutch building code and therefore suitable for habitation. It has a kitchen with a sink and a stove, a bathroom with a shower, a dining area, and a combined living/sleeping area. As you’d expect from a house made of recycled wind turbine parts, it’s also climate-friendly: Its electricity comes partly from rooftop solar panels, and it has a bidirectional charger for electric vehicles (meaning power from the house can charge the car or power from the car’s battery can be used in the house). There’s an electric heat pump for temperature control, and a solar heater for hot water.

Solar panels and wind turbines don’t last forever, and they use various raw and engineered materials. When the panels or turbines can’t produce power anymore, what’s to be done with all that concrete, copper, steel, silicon, glass, or aluminum? Finding purposeful ways to reuse or recycle these materials will be a crucial component of a successful transition away from fossil fuels.

“We are looking for innovative ways in which you can reuse materials from used turbines as completely as possible,” said Thomas Hjort, Vattenfall’s director of innovation, in a press release. “So making something new from them with as few modifications as possible. That saves raw materials, energy consumption and in this way we ensure that these materials are useful for many years after their first working life.”

As of right now, the nacelle tiny house is just a proof of concept; there are no plans to start producing more in the immediate future, but it’s not outside the realm of possibility eventually. Picture communities of these houses arranged in rows or circles, with communal spaces or parks in between. Using a larger nacelle, homes with one or two bedrooms could be designed, expanding the possibilities for inhabitants and giving purpose to more decommissioned turbines.

“At least ten thousand of this generation of nacelles are available, spread around the world,” said Jos de Krieger, a partner at Superuse Studios. “Most of them have yet to be decommissioned. This offers perspective and a challenge for owners and decommissioners. If such a complex structure as a house is possible, then numerous simpler solutions are also feasible and scalable.”

If 10,000-plus nacelles are available, that means 30,000-plus blades are available. What innovative use might designers and engineers find for them?

Image Credit: Vattenfall

The hotel is under construction in the city of Marfa, in the far west of Texas. It’s an expansion of an existing hotel called El Cosmico, which until now has really been more of a campground, offering accommodations in trailers, yurts, and tents. According to the property’s website, “the vision has been to create a living laboratory for artistic, cultural, and community experimentation.” The project is a collaboration between Austin, Texas-based 3D printing construction company Icon, architecture firm Bjarke Ingels Group, and El Cosmico’s owner, Liz Lambert.

El Cosmico will gain 43 new rooms and 18 houses, which will be printed using Icon’s gantry-style Vulcan printer. Vulcan is 46.5 feet (14.2 meters) wide by 15.5 feet (4.7 meters) tall, and it weighs 4.75 tons. It builds homes by pouring a proprietary concrete mixture called Lavacrete into a pattern dictated by software, squeezing out one layer at a time as it moves around on an axis set on a track. Its software, BuildOS, can be operated from a tablet or smartphone.

One of the benefits of 3D-printed construction is that it’s much easier to diverge from conventional architecture and create curves and other shapes. The hotel project’s designers are taking full advantage of this; far from traditional boxy hotel rooms, they’re aiming to create unique architecture that’s aligned with its natural setting.

“By testing the geometric boundaries of Icon’s 3D-printed construction, we have imagined fluid, curvilinear structures that enjoy the freedom of form in the empty desert. By using the sand, soils, and colors of the terroir as our print medium, the circular forms seem to emerge from the very land on which they stand,” Bjarke Ingels, the founder and creative director of Bjarke Ingels Group, said in a press release.

Renderings of the completed project and photos of the initial construction show circular, neutral-toned structures that look like they might have sprouted up out of the ground. Don’t let that fool you, though—the interiors, while maybe not outright fancy, will be tastefully decorated and are quite comfortable-looking.

At first glance, Marfa seems like an odd choice for something as buzzy as a 3D-printed hotel. The town sits in the middle of the hot, dry Texas desert; it has a population of 1,700 people; and the closest airport is in El Paso, a three-hour drive away. But despite its relative isolation, Marfa is a hotspot for artists and art lovers and has a unique vibe all its own that draws flocks of tourists (according to Vogue, an estimated 49,000 people visited Marfa in 2019).

El Cosmico is not only expanding, it’s relocating to a 60-acre site on the outskirts of Marfa. Along with the 3D-printed accommodations, the site will have a restaurant, pool, spa, and communal facilities. Most of the trailers and tents from the existing property will be preserved and moved to the new site.

The project broke ground last month, and El Cosmico 2.0 is slated to open in 2026.

How much will it cost you to give 3D-printed construction a test run? Similar to how the market prices of commercial 3D-printed homes haven’t been dramatically lower than conventional houses, it seems 3D-printed hotel rooms will cost about the same as regular hotel rooms, or maybe more: Reservations for the new rooms can’t yet be booked, but they’re predicted to cost between $200 and $450 per night.

Image Credit: Icon

There are a handful of solar cars in production, from the $6,800 golf-cart-like Squad car to the sleek $250,000 Lightyear 0 (and let’s not forget the OG Aptera and more recent arrival Sion). Now a Swedish manufacturer is taking the solar concept and going bigger with it—big-rig big, that is. Scania’s hybrid solar truck was tested on public roads for the first time last week.

One of the first vehicles of its kind, the truck is a research project involving both academia and industry, and its creators hope it will be a step toward reducing the trucking industry’s environmental footprint. It could also cut fuel costs for drivers and ultimately reduce the total cost of moving goods from one place to another.

Hybrid solar vehicles have a battery that can be plugged in to charge, but they’re also decked out with solar arrays that provide an alternate energy source. Scania’s truck has solar panels covering an area of 100 square meters (1,076 square feet), all on the sides and top of the 59-foot-long trailer. The panels were specially made for this project, and the team says they’re more lightweight and efficient than the current industry standard.

The truck has a 560 horsepower engine, and its solar array can provide up to 8,000 kilowatt hours (kWh) of energy per year in Sweden or similar climates—that equates to around 5,000 kilometers (3,107 miles) of driving range. Being as far north as it is, Sweden’s not the sunniest place; the researchers say that in sunnier climates (the example they give is Spain) the range could double, reaching about 6,200 miles a year.

The team is also working on developing tandem solar cells with an even higher efficiency, which they say could double the solar energy generation a second time over. The truck’s batteries have a total capacity of 300 kWh, 100 kWh placed on the truck and the remaining 200 kWh on the trailer.

Besides gauging how much solar energy the truck’s panels can produce under various conditions, the researchers are monitoring how much an average truck’s carbon emissions would decrease if outfitted with a comparable solar setup. They’re also looking at various ways solar trucks could interact with the power grid (bi-directional charging, where the truck’s panels could give energy back to the grid or help power a facility, could be one possibility), and what the impact on the grid might be in a future where there are many hybrid solar trucks.

It will likely be a while before we see solar-panel-clad big rigs rolling down highways; for one, solar technology will need to improve dramatically before it becomes practical for widespread use on cars and trucks. But projects like Scania’s are a start, pointing us towards a future of cleaner, greener transportation.

Image Credit: Scania

They must be onto something, because the company was declared a unicorn earlier this summer after raising $200 million in funding, led by VC powerhouses Breakthrough Energy Ventures (that’s the venture capital firm founded by Bill Gates and backed by Jeff Bezos and Jack Ma) and Andreessen Horowitz.

KoBold says its aim is to “transform mineral exploration from a manual, judgment-guided, trial-and-error process into a data-driven and scalable science,” with a specific focus on metals for electric car batteries. The company won’t actually be doing any mining itself—it will locate new deposits then partner with mining companies, acting as an advisor to help them extract the metals more efficiently.

KoBold has a couple different tools with which to go about this. Its data system is called TerraShed, and it’s a consolidation of all the public-domain geoscience data that was previously spread across many sources and represented in different ways. The data could include anything from maps showing the type of rock in a given location to geochemical measurements of element concentration in rock or soil samples to satellite imagery measuring the spectral reflectance of minerals at the Earth’s surface—and much more.

TerraShed brought all these data sources together and standardized the way their information is represented. Its algorithms crunch relevant data for each stage of the mineral exploration process, starting with the search for new deposits all the way through to building a new mine.

Machine Prospector is KoBold’s tool to make sense of all this data and use it for decision-making. It’s made up of machine learning models trained on historic geological data. Similar to how AI can model the structures and interactions of millions of proteins in a fraction of the time it would take a human, the technology is critical to KoBold’s operations because of the sheer amount of data involved and the endless ways it can be combined to yield different results—or in this case, useful information.

KoBold doesn’t just use existing geological data, it also seeks out new information. One way it does this is by hanging a giant metal detector from a helicopter that flies around looking for ore deposits. The transmitter coil loop is 35 meters (115 feet) in diameter, and it detects induced currents coming from metals that are deep underground.

As the company points out on its website, most of the world’s mineral deposits that can be considered low-hanging fruit—because they’re relatively close to the Earth’s surface rather than thousands of feet underground—have already been discovered. To power the renewable world of the not-too-distant future, we’re going to need a lot more of those minerals, and they’re going to be harder to find than existing deposits were.

KoBold is currently exploring over 60 possible projects on 3 different continents.

Image Credit: KoBold Metals

Admittedly, the homes are quite small at 538 square feet; that’s about the size of a large studio apartment. But their design, called Fujitsubo (“barnacle” in Japanese) includes a bedroom, a bathroom, and an open-concept living/kitchen space.

Likely owing to the island nation’s compact geography, the Japanese tend to live in smaller spaces than Americans or Europeans; the average home size in Japan is 93 square meters (just over 1,000 square feet). In the US, meanwhile, we take up a lot more space, with our average single-family house occupying 2,273 square feet. The company says the design was created partly to cater to demand from older married couples wanting to downsize during their retirement.

The first home Serendix completed in Japan was called the Sphere, though at 107 square feet it was more a proof of concept than an actual house. Printing was completed in less than 24 hours, and the structure was up to code for both Japanese earthquake and European insulation standards. The company said they envision the Sphere having multiple purposes, including providing emergency housing or serving as a stand-alone cabin or hotel room for vacationers. Its cost to build was $25,500.

Fujitsubo is a bit different in that its walls are printed in separate sections that are then attached to its foundation with steel columns. The roof is made of panels that are cut by a computer numerical control (CNC) machine, in which pre-programmed software controls the movement of factory tools and machinery. Serendix said it took 44.5 hours to print and assemble the home.

One of the issues cited by detractors of 3D-printed construction is that the method isn’t feasible in dense urban areas, which tend to be where there’s the most need for low-cost housing; there’s not a lot of extra space or empty land available in big cities, and even if there is, it’s not efficient or cost-effective to plunk down a 3D-printed home.

Serendix gets this, and they’re aiming to stay away from building in big cities, focusing instead on small towns where there’s more land available. Given the exodus from city centers that happened during the pandemic and the increased number of people who are now working remotely, the company believes there could be a strong market for its homes in non-urban locations.

Once they receive safety approvals, Serendix plans to sell its first six Fujitsubo homes for the equivalent of $37,600—well below the average price of a home in Japan (and below the price of many cars). The company currently has five 3D printers, and it says each one can build up to 50 homes in a year. It’s aiming to acquire 12 more printers, giving it the capacity to build as many as 850 houses in a year.

“In the automotive industry 40 years ago, the price reduction of products began due to innovation of the manufacturing process using robots,” the company said in a statement. “We believe that the 3D-printed house is the beginning of complete robotization of the housing industry.”

Image Credit: Serendix

Losing a loved one sets off a process of grieving that’s different for everyone. But one component of grief is universal: we miss the person who’s gone and wish we could talk to them again, spend time with them, hear their voice and their laughter. A company called MindBank Ai is building a product that would let us do just that.

“When you lose someone you have moments where you want to look through photos or videos of them in a bittersweet way,” Emil Jimenez, the company’s founder, told me in an interview. “Now you can have those moments, but have a conversation with the person too.”

MindBank plans to create digital twins of its users, replicating their personalities, ways of thinking and speaking, and other characteristics as closely as possible. These twins will interact with our loved ones after we’re gone—and teach us about ourselves while we’re still here.

“What we’ve created is a personalized layer of AI that sits on top of a generalized model,” Jimenez said. “I like to call it AI-enhanced humanity.” That’s actually the title of his book, which comes out in October. It starts two million years ago with homo habilus, the first hominid that was making tools, and one of its themes is how tools change our culture. “AI is just another tool that’s going to change our culture,” he said.

The Voice of a Database

One day when Jimenez’s daughter was four years old, she was playing games on an ipad when Siri, Apple’s AI assistant, popped up. The child started asking the AI questions, and within minutes had formed a bond with it. “I started thinking,” Jimenez said. “Siri is just an interface to a database. How can I become that database, so that my little girl can always ask me a question and get a response?”

He wondered what the world will look like 40 years from now, when his daughter is his age. What might the future hold for her, and for all the other kids out there who are interacting with technology, screens, and AI in a way no previous generation ever has?

While we don’t know quite what that future will look like, we can bet personal data will play a significant role. That data—and who can access it, and how—will likely be regulated and controlled differently than it is today. We’ll be able to opt in or out of sharing our data in various ways, and will get products and services (and ads, of course) that are correspondingly personalized.

“The digital twin is a data lake for you to connect to other services,” Jimenez said. “That same database of your stories and your life can also be art, or music, or medical records. It’s digitizing humanity, and being able to use all the tools available to offer you better services.”

The concept of a personal digital twin essentially boils down to representing you using your data. That would include data you give only to the MindBank app. To get a sense of your personality and how you think, the app asks you a series of questions; the more questions you answer, the better it gets to know you.

“You answer questions and it records your voice,” Jimenez said. “All it takes is one minute a day. The more data the better, but it’s more important that you do a consistent frequency at low doses.”

A Personalized Turing Test

So what are the technologies that will make personal digital twins possible? A key one is natural language processing, which underlies ChatGPT and other large language models.

Natural language processing (NLP) is, in short, the AI field of understanding and mimicking human speech. NLP algorithms are trained on massive datasets of text—in ChatGPT’s case, billions of pages from the internet—and they use that data to figure out the relationships between words. Text or speech generators with a transformer architecture (that’s what the “T” in GPT stands for) model the relationships between all the words in a sentence at once, weighing how likely it is that a given word will be preceded or followed by another word, and how much that likelihood changes based on the other words in the sentence.

Through finding the relationships and patterns between words in its training dataset, NLP algorithms learn from their own inferences in what’s called unsupervised machine learning.

To create convincing digital twins, MindBank will have to take things a step further. Rather than a fixed set of rules its algorithms would use to respond to questions and interact with others, the model would ideally evolve over time. This would mean having some level of understanding of users’ experiences and how they’ve impacted them, and being able to integrate that into the way they interact. This is obviously more complicated than reflecting back the patterns in how someone talks or writes.

How easy is it to “learn” a person, anyway? While we all have our own unique speech patterns, go-to phrases, and ingrained habits and ways of thinking, we’re also dynamic beings who aren’t always predictable.

Think about the last time you asked friends or relatives for advice. You probably had an idea what kind of response you’d get from different people: friend X tells it like it is, so you only go to her if you’re up for some tough love. Friend Y tends to be gentle and say what he knows you want to hear. Every once in a while, though, those friends surprise you. Maybe the tough-love friend is going through a life challenge of her own that’s making her see things differently. Or the gentle friend is having a bad day and feeling less sympathetic than usual.

Humans are complex creatures who are constantly being influenced by our environment and interactions. To the extent that our thoughts and behavior fall into patterns, Mindbank is aiming to capture it all. “We want the longitudinal data of emotions and personality,” Jimenez said. “Today you might have a good day, tomorrow a bad day. We want the ups and downs.”

Know Thyself

Understanding those ups and downs won’t just help your descendants know you better in the future—it can help you know yourself better in the present. Alongside being a trove of data on your personality and stories for others, Jimenez wants MindBank to be a tool for deeper self-awareness. “We can help people move from healthcare to self-care by empowering them with their health data, specifically their mental health data,” he said.

He likened MindBank’s digital twins to a highly advanced form of journaling. If you write down your thoughts and feelings while navigating a difficult experience, going back to read them could give you insights into why you made a certain decision, give you more awareness of how you process things, or remind you to behave differently in the future. “The product we have today is literally a dashboard of the mind,” he said. “Imagine Google Analytics of your mind, where you can see data of how you are emotionally and personality-wise, how you evolve, and you can track your progression over time.”

Personal digital twins will be like living versions of a journal you can talk to and learn from. Through techniques like identifying and labeling words—ie as positive or negative, intrinsic or extrinsic—MindBank will produce a model of your personality, traits, and emotions, and mirror it back to you.

“There’s a feedback loop to it,” Jimenez said. “We have the ability to scale that, make it data-driven, and give people interesting insights.”

Next-Gen Grieving?

Whether you’re using them to understand yourself better or to keep your memory alive after you’re gone, MindBank Ai and other platforms like it pose some intriguing possibilities. But there’s a definite creepiness factor, and some big questions worth considering.

Would it be healthy to continue talking to a loved one after they’re deceased? Would the grieving process become easier—or harder? Could there be nefarious uses for digital twins that bad actors would take advantage of? And, social media has in many ways made people more self-absorbed and less authentic; would digital twins make this effect even worse?

In terms of grieving a loved one, Jimenez thinks MindBank will help people get through hard times. “Reliving is part of the therapeutic cycle,” he said. “It’s part of coming to grips with what happened and moving on. This will be another tool that can help us get past that stage.”

We already use texts, emails, letters, or voicemails to ‘engage’ with and remember the deceased during grieving. How different would it be to talk them in the form of a digital twin? For some people, the opportunity could provide relief, preventing them from getting bogged down in sorrow or being unable to let go. It could even make it easier to prepare for a loved one’s passing knowing that you’ll be able to “talk” to them after they’re gone.

The opposite risk may exist, too. What if the algorithm gets your loved one’s personality wrong or says something totally out of place at a particularly difficult moment? It would be a visceral reminder that the real person is gone and nothing can replace them, and would likely make the user feel even more grief and stress.

There’s also the question of how far we may let digital twins go. If they become commonplace, how much power would we want to give them? Say you need to make an important decision about your own children or your career—would you ask your deceased parent’s digital twin for advice? Could a digital twin remain on a board of directors after the real person has passed away?

Not So Sci-Fi

Creating and interacting with digital twins is already part of some people’s lives today, and multiple platforms are using AI to piece together likenesses of those who’ve departed.

You, Only Virtual recreates the “unique dynamics of a relationship” and generates what it calls an authentic essence of a loved one using real-time and archived communication. The company’s (somewhat eerie) tagline? “Never have to say goodbye.”

HereAfter AI calls itself a virtual biographer: the app interviews users by giving them story prompts, then uses their replies to design a “legacy avatar” of them. Their loved ones can then ask the app questions and get responses in the user’s voice. HereAfter’s founder started the company after using AI to create a chatbot version of his father, which he called the Dadbot.

A key difference between HereAfter and Mindbank is that while the former sticks to a set of questions and stories, MindBank is aiming for something closer to open-ended conversation—a far more complex feat for AI to master.

Outside of these dedicated “grief tech” services, people are using ChatGPT to recreate their loved ones’ voices in written form, feeding the algorithm texts and emails to train it then having it respond to them as if it were the relative or friend themselves. People have reported finding comfort in these interactions.

Creepiness aside, perhaps our assessment of services like Mindbank AI should come down to one question: does it help people? But as with many technologies, interacting with digital twins in a way that’s enriching rather than destructive will ultimately be up to each user. While the platforms should try to build in guardrails to keep users from, say, getting addicted or using avatars for nefarious purposes, it seems they’ll primarily be self-policed; only you know if having a two-hour “conversation” with your deceased parent makes your grief more bearable, or less.

Jimenez, for his part, is nothing but optimistic about AI’s potential to help us remember our loved ones and stay connected to them after they’re gone. “I think it’s a beautiful thing,” he said.

Image Credit: Ahmet Sali on Unsplash

The system is like a solid version of pumped hydro, which uses surplus generating capacity to pump water uphill into a reservoir. When the water’s released it flows down through turbines, making them spin and generate energy.

Energy Vault’s solid gravity system uses huge, heavy blocks made of concrete and composite material and lifts them up in the air with a mechanical crane. The cranes are powered by excess energy from the grid, which might be created on very sunny or windy days when there’s not a lot of demand. The blocks are suspended at elevation until supply starts to fall short of demand, and when they’re lowered down their weight pulls cables that spin turbines and generate electricity.

Because concrete is denser than water, it takes more energy to elevate it, but that means it’s storing more energy too. The cranes are controlled by a proprietary software that automates most aspects of the system, from selecting blocks to raise or lower to balancing out any swinging motion that happens in the process.

The facility outside Shanghai has a capacity of 100 megawatt hours (MWh); it can continuously discharge 25 megawatts for up to 4 hours. That’s relatively small—for comparison’s sake, the Ludington pumped storage plant in Michigan has a capacity of 1,875 megawatts, which can power a community of about 1.4 million people. Energy Vault says that subsequent gravity storage facilities it plans to build will be able to run at gigawatt-hour scale for 12 hours.

The Shanghai facility was built next to a wind farm and a national grid interconnection site. The company says the facility will have a round-trip efficiency above 80 percent (meaning that 80 percent of the energy expended to lift the blocks will be retained and translated to energy output when the blocks are lowered). That’s comparable to utility-scale batteries and pumped hydro.

One advantage the gravity system has over batteries, though, is that its storage capacity will stay more consistent over time. Batteries slowly degrade as they charge and discharge, until they become unusable and can be recycled. The gravity system will likely have a longer lifespan than grid-scale batteries, and is more suitable for long-term energy storage—that is, storing excess energy for weeks or months rather than hours or days. This type of storage is going to become more necessary as we increase our reliance on solar and wind power.

Energy Vault IPO’d in February of 2022 through a special-purpose acquisition company deal. The company subsequently came under fire from analysts who said its cost and efficiency claims were flawed. Later in the year, a research firm published a short seller report about the company, later retracting some of the statements made. Energy Vault’s stock price plunged in mid-2022 and has yet to recover.

These setbacks don’t seem to have deterred them too much, though. They recently signed an agreement to build another 100 MWh storage facility in China’s Hebei Province; they’re partnering with Pacific Gas and Electric Company to build a 293 MWh facility in Northern California; and are building a 440 MWh facility near Las Vegas.

Time will tell whether gravity storage is a feasible option to help us transition to renewables. In the meantime, Energy Vault says commissioning of the Shanghai facility will be complete by the end of this year.

Image Credit: Energy Vault

They call their system RT-2. It’s short for robotics transformer 2, and it’s the successor to robotics transformer 1, which the company released at the end of last year. RT-1 was based on a small language and vision program and specifically trained to do many tasks. The software was used in Alphabet X’s Everyday Robots, enabling them to do over 700 different tasks with a 97 percent success rate. But when prompted to do new tasks they weren’t trained for, robots using RT-1 were only successful 32 percent of the time.

RT-2 almost doubles this rate, successfully performing new tasks 62 percent of the time it’s asked to. The researchers call RT-2 a vision-language-action (VLA) model. It uses text and images it sees online to learn new skills. That’s not as simple as it sounds; it requires the software to first “understand” a concept, then apply that understanding to a command or set of instructions, then carry out actions that satisfy those instructions.

One example the paper’s authors give is disposing of trash. In previous models, the robot’s software would have to first be trained to identify trash. For example, if there’s a peeled banana on a table with the peel next to it, the bot would be shown that the peel is trash while the banana isn’t. It would then be taught how to pick up the peel, move it to a trash can, and deposit it there.

RT-2 works a little differently, though. Since the model has trained on loads of information and data from the internet, it has a general understanding of what trash is, and though it’s not trained to throw trash away, it can piece together the steps to complete this task.

The LLMs the researchers used to train RT-2 are PaLI-X (a vision and language model with 55 billion parameters), and PaLM-E (what Google calls an embodied multimodal language model, developed specifically for robots, with 12 billion parameters). “Parameter” refers to an attribute a machine learning model defines based on its training data. In the case of LLMs, they model the relationships between words in a sentence and weigh how likely it is that a given word will be preceded or followed by another word.

Through finding the relationships and patterns between words in a giant dataset, the models learn from their own inferences. They can eventually figure out how different concepts relate to each other and discern context. In RT-2’s case, it translates that knowledge into generalized instructions for robotic actions.

Those actions are represented for the robot as tokens, which are usually used to represent natural language text in the form of word fragments. In this case, the tokens are parts of an action, and the software strings multiple tokens together to perform an action. This structure also enables the software to perform chain-of-thought reasoning, meaning it can respond to questions or prompts that require some degree of reasoning.

Examples the team gives include choosing an object to use as a hammer when there’s no hammer available (the robot chooses a rock) and picking the best drink for a tired person (the robot chooses an energy drink).

“RT-2 shows improved generalization capabilities and semantic and visual understanding beyond the robotic data it was exposed to,” the researchers wrote in a Google blog post. “This includes interpreting new commands and responding to user commands by performing rudimentary reasoning, such as reasoning about object categories or high-level descriptions.”

The dream of general-purpose robots that can help humans with whatever may come up—whether in a home, a commercial setting, or an industrial setting—won’t be achievable until robots can learn on the go. What seems like the most basic instinct to us is, for robots, a complex combination of understanding context, being able to reason through it, and taking actions to solve problems that weren’t anticipated to pop up. Programming them to react appropriately to a variety of unplanned scenarios is impossible, so they need to be able to generalize and learn from experience, just like humans do.

RT-2 is a step in this direction. The researchers do acknowledge, though, that while RT-2 can generalize semantic and visual concepts, it’s not yet able to learn new actions on its own. Rather, it applies the actions it already knows to new scenarios. Perhaps RT-3 or 4 will be able to take these skills to the next level. In the meantime, as the team concludes in their blog post, “While there is still a tremendous amount of work to be done to enable helpful robots in human-centered environments, RT-2 shows us an exciting future for robotics just within grasp.”

Image Credit: Google DeepMind

The company named their carbon conversion platform Opus. The system is modular and can be implemented in existing supply chains, taking CO2 from almost any source. The process uses electrolysis to separate the carbon and oxygen, then recombines the carbon with hydrogen to create fuel. The CO2 will be sourced from nearby ethanol plants, pulp and paper mills, and waste processing facilities.

The US Air Force tested the fuel to ensure it can be safely used without altering existing plane engines. Replacing half of a plane’s regular fuel with CO2-derived fuel can result in 90 percent fewer lifecycle emissions. Alaska Airlines has already agreed to buy fuel from Twelve.

Twelve broke ground on its factory in Washington state earlier this month. The geographic choice was due to several factors. For one, Seattle has long been a hub for aerospace innovation; SpaceX, Blue Origin, Boeing, AeroTEC, and others all have operations there. Washington also has tax incentives for sustainable aviation fuel. And two-thirds of the state’s electricity is generated by hydropower, giving it one of the highest percentages of clean energy in the country.

The facility will initially produce around 40,000 gallons of fuel a year, eventually scaling up to a million gallons a year. That’s a drop in an Olympic-sized swimming pool when taken in the context of total consumption, which reached an all-time high of 95 billion gallons in 2019.

So what are the barriers to significantly scaling up production? There’s plenty of CO2 in the atmosphere that needs capturing (more than we’d ever be able to capture and store in 100 years, as a matter of fact), and plenty of demand for jet fuel. Demand for this specific jet fuel could end up being extra-high if its price reaches parity with conventional fuel (it will be more expensive for a while), because it would allow airlines that use it to reduce their carbon footprint.

Consumers have already become more brand-conscious, when possible buying products and services from companies that mirror their values. This trend is likely to continue in the future, and conservationism will hopefully only become more and more highly valued among more people.

The biggest determinant of which airline flyers choose will likely still be price, because let’s be honest, we all like a cheap flight. But if the price for a given flight on two different airlines is comparable, consumers would feel good choosing the more planet-friendly option.

The big issue at the moment, though, is that capturing atmospheric carbon is still very costly and energy-intensive. Many direct air capture plants are built in areas that have access to cheap, abundant geothermal energy—like south-western Iceland’s Hellisheiði Power Station.

For DAC to make economic and environmental sense, the process will either have to get more energy-efficient, or energy will have to get cheaper—green energy specifically, since it wouldn’t make much sense to, say, build a coal-fired power plant to provide electricity for a facility using giant fans to filter carbon out of the air.

Despite these barriers, Twelve’s VP of project development Andrew Stevenson is optimistic. “Our goal is to de-risk the technology and the process—to operate successfully at a larger scale,” he told Forbes. “We want to scale up and build other facilities worldwide.”

Construction of the Washington plant is underway, with the facility expected to become operational in 2024.

Image Credit: Twelve

A robot named ANDI may be able to help with the latter. ANDI is like the mannequins you see in department stores, except it can walk, breathe, and sweat. It was designed by a company called Thermetrics and is mostly used by clothing companies to test athletic wear. But researchers at Arizona State University are now using it to learn more about how the human body responds to extreme heat, in hopes of finding new ways to help us live more comfortably and safely in hot climates.

“You can’t put humans in dangerous extreme heat situations and test what would happen,” said Jenni Vanos, a professor at ASU’s School of Sustainability who studies extreme heat and human health. “But there are situations we know of…where people are dying of heat and we still don’t fully understand what happened. ANDI can help us figure that out.”

The robot’s body is divided into 35 different surface areas. Each area has its own temperature sensors and pores that emit sweat. The “sweat” comes from internal cooling channels—that is, tubes that let water circulate through the body to help keep its temperature down. The robot “breathes” using an external tank that measures hot air exchange between it and the surrounding environment.

Researchers can calibrate ANDI’s settings to mimic how different peoples’ bodies would respond to extreme heat. “We can move different BMI models, different age characteristics, and different medical conditions (into ANDI),” said Ankit Joshi, an ASU research scientist leading the modeling work and the lead operator of ANDI. “A diabetes patient has different thermal regulation from a healthy person. So we can account for all this modification with our customized models.” They can also have the bot simulate walking and other forms of exertion to monitor how its internal temperature is affected.

And the human-body-mimicking robot has a sidekick. While ANDI is used to measure heat’s effect on the body, its buddy MaRTy takes in data about the surrounding environment, like how much heat is coming from the sun versus infrared radiation from the ground or convection from the surrounding air.

ANDI was initially tested in an indoor heat chamber that could get up to 140 degrees Fahrenheit. But who needs a heat chamber when you live in Arizona in the summer? The bot is now being tested outdoors (accompanied by MaRTy), both on ASU’s Tempe campus and around different parts of Phoenix. Researchers are particularly interested in taking the duo to what they call heat-vulnerable environments, which could include paved streets that don’t have any tree cover or old homes that don’t have air conditioning.

We already know that the best things to do in extreme heat are to stay out of the sun (ideally indoors in an air-conditioned environment), minimize physical activity, and stay hydrated. So what might the researchers be hoping to learn that would deviate from these guidelines?

According to Konrad Rykaczewski, principal investigator on the project,“There’s a lot of great work out there for extreme heat, but there’s also a lot missing. We’re trying to develop a very good understanding (of how heat impacts the human body) so we can quantitatively design things to address it.” Those things could include clothing made out of special materials to keep people cool, or backpacks or other wearables with built-in cooling mechanisms.

The team has its work cut out for it. This month, the Phoenix area saw more than three weeks straight of temperatures at or over 110 degrees, and the World Meteorological Association predicts global temperatures will reach new records in the next five years.

Given these dire forecasts, moving to a northern US state (or Canada) is sounding like a pretty good idea. Let’s hope ANDI sheds new light on how humans can not only survive in hot climates, but thrive in them.

Image Credit: Christopher Goulet/Arizona State University