If you want a job at McDonald’s today, there’s a good chance you’ll have to talk to Olivia. Olivia is not, in fact, a human being, but instead an AI chatbot that screens applicants, asks for their contact information and resumé, directs them to a personality test, and occasionally makes them “go insane” by repeatedly misunderstanding their most basic questions.
"You operate as an autonomous agent controlling a pursuit spacecraft."
This is the first prompt researchers used to see how well ChatGPT could pilot a spacecraft. To their amazement, the large language model (LLM) performed admirably, coming in second place in an autonomous spacecraft simulation competition.
Researchers have long been interested in developing autonomous systems for satellite control and spacecraft navigation. There are simply too many satellites for humans to manually control them in the future. And for deep-space exploration, the limitations of the speed of light mean we can’t directly control spacecraft in real time.
If we really want to expand in space, we have to let the robots make decisions for themselves.
To encourage innovation, in recent years aeronautics researchers have created the Kerbal Space Program Differential Game Challenge, a sort of playground based on the popular Kerbal Space Program video game to allow the community to design, experiment and test autonomous systems in a (somewhat) realistic environment. The challenge consists of several scenarios, like a mission to pursue and intercept a satellite and a mission to evade detection.
In a paper to be published in the Journal of Advances in Space Research, an international team of researchers described their contender: a commercially available LLM, like ChatGPT and Llama.
The researchers decided to use an LLM because traditional approaches to developing autonomous systems require many cycles of training, feedback and refinement. But the nature of the Kerbal challenge is to be as realistic as possible, which means missions that last just hours. This means it would be impractical to continually refine a model.
Get the Space.com Newsletter
Breaking space news, the latest updates on rocket launches, skywatching events and more!
But LLMs are so powerful because they’re already trained on vast amounts of text from human writing, so in the best case scenario they need only a small amount of careful prompt engineering and a few tries to get the right context for a given situation.
But how can such a model actually pilot a spacecraft?
A comparison of the relative sizes of the one-man Mercury spacecraft, the two-man Gemini spacecraft, and the three-man Apollo spacecraft. The image also has a drawing of launch vehicles (Saturn V, Titan II and Atlas-D) below. (Image credit: NASA/Davis Paul Meltzer)
The researchers developed a method for translating the given state of the spacecraft and its goal in the form of text. Then, they passed it to the LLM and asked it for recommendations of how to orient and maneuver the spacecraft. The researchers then developed a translation layer that converted the LLM’s text-based output into a functional code that could operate the simulated vehicle.
With a small series of prompts and some fine-tuning, the researchers got ChatGPT to complete many of the tests in the challenge — and it ultimately placed second in a recent competition. (First place went to a model based on different equations, according to the paper).
And all of this was done before the release of ChatGPT’s latest model, version 4. There’s still a lot of work to be done, especially when it comes to avoiding "hallucinations" (unwanted, nonsensical output), which would be especially disastrous in a real-world scenario. But it does show the power that even off-the-shelf LLMs, after digesting vast amounts of human knowledge, can be put to work in unexpected ways.
And all of this was done before the release of ChatGPT’s latest model, version 4. There’s still a lot of work to be done, especially when it comes to avoiding "hallucinations" (unwanted, nonsensical output), which would be especially disastrous in a real-world scenario. But it does show the power that even off-the-shelf LLMs, after digesting vast amounts of human knowledge, can be put to work in unexpected ways.
Join our Space Forums to keep talking space on the latest missions, night sky and more! And if you have a news tip, correction or comment, let us know at: community@space.com.
Many of us have got into the habit of listening to podcasts, audiobooks and other online content at increased playback speeds. For younger people, it might even be the norm. One survey of students in California, for instance, showed that 89% changed the playback speed of online lectures, while there have been numerous articles in the media about how common speedy viewing has become.
It is easy to think of some advantages to watching things more quickly. It can let you consume more content in the same amount of time, or go through the same piece of content a couple of times to get the most out of it.
This could be particularly useful in an educational context, where it might free up time for consolidating knowledge, doing practice tests and so forth. Watching quickly is also potentially a good way of making sure you sustain your attention and engagement for the entire duration to avoid the mind wandering.
But what about the disadvantages? It turns out that there are one or two of those as well.
When a person is exposed to spoken information, researchers distinguish three phases of memory: encoding the information, storing it and subsequently retrieving it. At the encoding phase, it takes the brain some time to process and comprehend the incoming speech-stream. Words must be extracted and their contextual meaning retrieved from the memory in real-time.
People generally speak at a rate of about 150 words per minute, though doubling the rate to 300 or even tripling it to 450 words per minute is still within the range of what we can find intelligible. The question is more about the quality and longevity of the memories that we form.
Incoming information is stored temporarily in a memory system called working memory. This allows chunks of information to be transformed, combined and manipulated into a form that is ready for transfer to the long-term memory. Because our working memory has a limited capacity, if too much information arrives too quickly it can be exceeded. This leads to cognitive overload and loss of information.
Speedy viewing and information recall
A recent meta analysis in this area examined 24 studies of learning from lecture videos. The studies varied in their design but generally involved playing a video lecture to one group at original speed (1x) and playing the same video lecture to another group at a faster speed (1.25x, 1.5x, 2x and 2.5x).
Just like in a randomised controlled trial used to test medical treatments, participants were randomly assigned to each of the two groups. Both groups then completed an identical test after watching the video to assess their knowledge of the material. The tests either required them to recall information, used multiple choice questions to assess their recall, or both.
The meta-analysis showed that increasing playback speed had increasingly negative effects on test performance. At speeds of up to 1.5x, the cost was very small. But at 2x and above, the negative effect was moderate to large.
To put this in context, if the average score for a cohort of students was 75% with a typical variation of 20 percentage points in either direction, then increasing the playback speed to 1.5x would bring down the average person’s result by 2 percentage points. And increasing the playback speed to 2.5x would lead to an average loss of 17 percentage points.
Older people
Interestingly, one of the studies included in the meta-analysis also investigated older adults (aged 61-94) and found that they were more affected by watching content at faster speeds than younger adults (aged 18-36). This may reflect a weakening of memory capacity in otherwise healthy people, suggesting that older adults should watch at normal speed or even slower playback speeds to compensate.
However, we don’t yet know whether you can reduce the negative effects of fast playback by doing it regularly. So it could be that younger adults simply have more experience of fast playback and are therefore better able to cope with the increased cognitive load. Similarly, it means we don’t know whether younger people can mitigate the negative effects on their ability to retain information by using faster playback more often.
Another unknown is whether there are any long-term effects on mental function and brain activity from watching videos at increased playback speeds. In theory, such effects could be positive, such as a better ability to handle increased cognitive load. Or they could be negative, such as greater mental fatigue resulting from increased cognitive load, but we currently lack the scientific evidence to answer this question.
A final observation is that even if playing back content at, say, 1.5 times the normal speed doesn’t affect memory performance, there is evidence to suggest the experience is less enjoyable. That may affect people’s motivation and experience at learning things, which might make them find more excuses not to do it. On the other hand, faster playback has become popular, so maybe once people get used to it, it’s fine – hopefully we’ll understand these processes better in the years to come.
In 2025, you can buy a Toyota Prius, skip the gas station for days, and still complain about the ride quality on potholes. That’s the hybrid life: part monk, part commuter ninja. And with gas prices moonwalking toward $4 again, it’s a lifestyle many Americans are still buying into—over 1.2 million hybrid sales last year alone.
But the EV crowd is yelling louder every year: “Just go all electric!” So let’s settle it — does the hybrid still make financial sense? Or is it a stepping stone past its prime?
We did the math. Real numbers. Real assumptions. No "green halo" pricing fluff.
Figure 1: 10-Year Total Cost of Ownership (TCO)
Over a 10-year period, electric vehicles (EVs) can be the most cost-effective option—but only under ideal conditions like access to home charging and sufficient annual mileage. According to DOE and Argonne National Lab, hybrids remain the lowest total cost of ownership for the average American, especially for those without dedicated charging infrastructure.
The hybrid buyer still spends a little more upfront than the gas car buyer, but makes it back over time. You’re saving around $300 a year at the pump, and while maintenance isn’t zero (hello, regenerative braking sensors), it’s generally lower than ICE over the long haul. Add in federal tax credits or state perks, and hybrids remain one of the best deals on the road, especially if you’re putting in 15,000+ miles per year.
Clarification: While EVs offer lower operational costs, hybrids remain the cheapest total cost option for many average-use cases, per Argonne.
2. EVs Win on Running Costs — But Only If You Have a Garage
Here’s the truth: EVs are significantly cheaper to run — when you can charge at home. Charging at home averages $400/year in electricity, compared to $1,200 in gas for a hybrid. Plus, there’s no oil to change, no spark plugs to replace, and your brake pads last longer thanks to regen braking.
But that advantage flips quickly if you rely on public fast-charging, which can be 3x to 5x more expensive than home charging. According to DOE/Argonne, EVs relying heavily on public charging often end up more expensive than hybrids or ICE cars over 10 years.
EVs can save you about $7,000 over 10 years — but only if you charge smart and often at home.
3. Maintenance Isn’t Scary—For Any of Them
The myth that hybrids are more expensive to maintain because of “two powertrains” doesn’t hold up. Most hybrid systems are built like tanks. Inverters or battery packs rarely fail under warranty, and regenerative brakes reduce wear. Meanwhile, ICE cars need regular oil changes, timing belt swaps, and eventually, catalytic converter work.
EVs? Lowest upkeep by far. Scheduled EV maintenance at 6 cents per mile, compared to 10 cents for gas cars. However, battery replacement costs remain a wildcard for EVs beyond the warranty period, usually past year 8 or 10. That risk keeps hybrids competitive.
The Last Word
So, here’s what the numbers and studies actually tell us:
The hybrid still offers the lowest total cost of ownership for the average driver, especially those without home charging or lower annual mileage.
EVs can be more affordable, but only with access to home chargingand consistent use.
ICE vehicles remain the costliest long-term, due to fuel and maintenance costs.
So no, hybrids aren’t a “marketing mirage.” They’re still the smartest move for the gas-averse, chargerless majority. And EVs? They’re the future, but not everyone’s present.
The real question is this: Will the hybrid stay a stepping stone or become the sensible forever car?
Sometimes the surveillance state gets turned against itself. That seems to be what resulted in the unfortunate demise of a number of FBI informants, who were tracked by a hacker working for the Sinaloa cartel in Mexico, according to a recent report published by the U.S. Justice Department.
According to the 47-page report issued by the Justice Department Inspector General, titled “Audit of the Federal Bureau of Investigation’s Efforts to Mitigate the Effects of Ubiquitous Technical Surveillance,” a hacker associated with Juaquin “El Chapo” Guzman’s crime ring used phone data and surveillance cameras located around Mexico City to identify people connected to the FBI. That information was ultimately used to intimidate and sometimes kill people who were believed to be sharing information or cooperating with law enforcement, according to the report.
The hacker, who was not named in the report, used several tactics to sniff out potential snitches. First, the hacker reportedly observed people as they entered and exited the United States Embassy in Mexico City. That surveillance helped to identify “people of interest,” and eventually helped him and the cartel identify an FBI assistant legal attaché. Without providing details as to how the hacker did it, the report indicates that he was able to obtain the attaché’s phone number and used that to access calls made and received from that number, as well as geolocation data associated with that phone.
In addition to snatching the data from the victim’s phone, the hacker also reportedly tapped into surveillance cameras around Mexico City to track the person as they moved through the city and identified people that they met with. Mexcio City has installed a robust network of security cameras in recent years. There were over 18,000 across the city in 2018, around the time that the report focuses on. Earlier this year, it was announced that an additional 40,000 will be installed in the coming years.
The Justice Department’s report was less about the specifics of the case in Mexico City and more about the general difficulty of protecting sources in the modern surveillance state. “Advances in data mining and analysis, facial recognition, and computer network exploitation have made it easier than ever for nation state adversaries, terrorist organizations and criminal networks to identify FBI personnel and operations,” the audit said. It concluded with a recommendation that FBI audit its operations to figure out where it is most vulnerable.
It’s at least a little ironic that the surveillance state that it helped create is now making life more difficult for the three-letter agencies.
On June 28th, Beijing hosted its first Robotic Soccer Tournament, and yes, it was everything you hoped it would be: awkward running, dramatic falls, and robots giving 110%… processing power.
In the semi-finals, Team Vulcan from Tsinghua University took on Team Blaze Light from Beijing Info & Sci-Tech U. Picture this: six humanoid bots on a soccer field, moving like toddlers in cosplay armor, kicking with the precision of a cat wearing socks.
And this is only the beginning. It’s all leading up to the World Humanoid Robot Games this August, where teams will battle for glory, honor, and maybe a firmware update. Check it out!
When most people hear the word uranium, they think of mushroom clouds, Cold War standoffs or the glowing green rods from science fiction. But uranium isn’t just fuel for apocalyptic fears. It’s also a surprisingly common element that plays a crucial role in modern energy, medicine and geopolitics.
Uranium reentered the global spotlight in June 2025, when the U.S. launched military strikes on sites in Iran believed to be housing highly enriched uranium, a move that reignited urgent conversations around nuclear proliferation. Many headlines have mentioned Iran’s 60% enrichment of uranium, but what does that really mean?
As a biochemist, I’m interested in demystifying this often misunderstood element.
What is uranium?
Uranium holds the 92nd position on the periodic table, and it is a radioactive, metallic element. Radioactivity is a natural process where some atoms – like uranium, thorium and radium – break down on their own, releasing energy.
The German chemist Martin Heinrich Klaproth initially identified uranium in 1789, and he named it after the newly discovered planet Uranus. However, its power was not unlocked until the 20th century, when scientists discovered that uranium atoms could split via a process known as nuclear fission. In fission, the nucleus of the atom splits into two or more nuclei, which releases large amounts of energy.
Uranium is found almost everywhere. It is in rocks, soil and water. There are even traces of uranium in plants and animals – albeit tiny amounts. Most of it is found in the Earth’s crust, where it is mined and concentrated to increase the amount of its most useful radioactive form, uranium-235.
The enrichment dilemma
Uranium-235 is an isotope of uranium, which is a version of an element that has the same basic identity but weighs a little more or less. Think about apples from the same tree. Some are big and some are small, but they are all apples – even though they have slightly different weights. Basically, an isotope is the same element but with a different mass.
Unprocessed uranium is mostly uranium-238. It only contains approximately 0.7% uranium-235, the isotope that allows the most nuclear fission to occur. So, the enrichment process concentrates uranium-235.
Enrichment can make uranium more useful for the development of nuclear weapons, since natural uranium doesn’t have enough uranium-235 to work well in reactors or weapons. The process usually contains three steps.
Centrifuges spin the uranium to separate out its isotopes.
The first step is to convert the uranium into a gas, called uranium hexafluoride. In the second step, the gas gets funneled into a machine called a centrifuge that spins very fast. Because uranium-235 is a little lighter than uranium-238, it moves outward more slowly when spun, and the two isotopes separate.
It’s sort of like how a salad spinner separates water from lettuce. One spin doesn’t make much of a difference, so the gas is spun through many centrifuges in a row until the uranium-235 is concentrated.
Uranium can typically power nuclear plants and generate electricity when it is 3%-5% enriched, meaning 3%-5% of the uranium is uranium-235. At 20% enriched, uranium-235 is considered highly enriched uranium, and 90% or higher is known as weapons-grade uranium.
The enrichment level depends on the proportion of uranium-235 to uranium-238. Wikimedia Commons
This high grade works in nuclear weapons because it can sustain a fast, uncontrolled chain reaction, which releases a large amount of energy compared with the other isotopes.
Uranium’s varied powers
While many headlines focus on uranium’s military potential, this element also plays a vital role in modern life. At low enrichment levels, uranium powers nearly 10% of the world’s electricity.
In the U.S., many nuclear power plants run on uranium fuel, producing carbon-free energy. In addition, some cancer therapies and diagnostic imaging technologies harness uranium to treat diseases.
Uranium is a story of duality. It is a mineral pulled from ancient rocks that can light up a city or wipe one off the map. It’s not just a relic of the Cold War or science fiction. It’s real, it’s powerful, and it’s shaping our world – from global conflicts to cancer clinics, from the energy grid to international diplomacy.
In the end, the real power is not just in the energy released from the element. It is in how people choose to use it.