Examining the history and fate of many technologies, and reading the analyses of others, I have come to understand some general rules, one of which is that some technologies look great on paper, but are terrible in reality. The reason they look so promising at first is because they have some key advantage that just seems to make sense. Using hydrogen as an energy storage medium, I think, is one of those technologies (at least for some applications). The idea of hydrogen is very appealing – it can be stored as a gas, transported in pipelines, and burned for energy, combining with oxygen to make water as the only combustion product. No CO2 or other greenhouse gas is produced. That sounds really good. But the more you crunch the numbers, the less good hydrogen looks (at least with current technology). Researchers recently did just that, looking at the utility of hydrogen for heating homes. Their assessment was soundly negative.

The idea is that hydrogen would be produced by electrolyzing water, using excess energy from intermittent sources like solar and wind. That hydrogen can be stored, and then supplied to homes when needed for heating. This would replace fossil fuels for heating, which would have multiple advantages (especially if you happen to live in Europe during a conflict that reduces your fossil fuel supply). But when you take a closer look, the shine fades. Jan Rosenow, the report’s author and Europe Director at the energy think-tank the Regulatory Assistance Project, had this to say:

“However, all of the independent research on this topic comes to the same conclusion: heating with hydrogen is a lot less efficient and more expensive than alternatives such as heat pumps, district heating and solar thermal,”

The simple reason for this is that hydrogen is just expensive. Using hydrogen for heating would cost six times using a heat pump, which is an existing proven technology. Any money spent doing further research and development into a hydrogen heating infrastructure would be better spent simply subsidizing heat pumps. These systems can be a bit expensive to install, but they are cheap to run, very energy efficient, and do not directly burn fossil fuels.

Continue Reading »

We appear to be in the middle of an explosion of AI (artificial intelligence) applications and ability. I had the opportunity to chat with an AI expert, Mark Ho, about what the driving forces are behind this rapid increase in AI power. Mark is a cognitive scientist who studies how AI’s work out and solve problems, and compares that to how humans solve problems. I was interviewing him for an SGU episode that will drop in December. The conversation was far-ranging, but I did want to discuss this one question – why are AIs getting so much more powerful in recent years?

First let me define what we mean by AI – this is not self-aware conscious computer code. I am referring to what may be called “narrow” AI, such as deep learning neural networks that can do specific things really well, like mimic a human conversation, reconstruct art images based on natural-language prompts, drive a car, or beat the world-champion in Go.  The HAL-9000 version of sentient computer can be referred to as Artificial General Intelligence, or AGI. But narrow AI does not really think, it does not not understand in a human sense. For the rest of this article when I refer to “AI” I am referring to the narrow type.

In order to understand why AI is getting more powerful we have to understand how current AI works. A full description would take a book, but let me just describe one basic way that AI algorithms can work. Neural nets, for example, are a network of nodes which also act as gates in a feed forward design (they pass information in one direction). The gates receive information and assign a weight to that information, and if it exceeds a set threshold it then passes that along to the next layers of nodes in the network. decide whether or not to pass information onto the next node based on preset parameters, and can give different weight to this information. The parameters (weights and thresholds) can be tuned to affect how the network processes information. These networks can be used for deep machine learning, which “trains” the network on specific data. To do this there needs to be an output that is either right or wrong, and that result is fed back into the network, which then tweaks the parameters. The goal is for the network to “learn” how it needs to process information by essentially doing millions of trials, tweaking the parameters each time and evolving the network in the direction of more and more accurate output.

So what is it about this system that is getting better? What others have told me, and what Mark confirmed, is that the underlying math and basic principles are essentially the same as 50 years ago. The math is also not that complicated. The basic tools are the same, so what is it that is getting better? One critical component of AI that is improving is the underlying hardware, which is getting much faster and more powerful. There is just a lot more raw power to run lots of training trials. One interesting side point is that computer scientists figured out that graphics cards (graphics processing unit, or GPU), the hardware used to process the images that go to your computer screen, happen to work really well for AI algorithms. GPUs have become incredibly powerful, mainly because of the gaming industry. This, by the way, is why graphics cards have become so expensive recently. All those bitcoin miners are using the GPUs to run their algorithms. (Although I recently read they are moving in the direction of application specific integrated circuits.)

Continue Reading »

Carbon is an extremely useful element. Carbon-containing compounds can be food, fuel, fertilizer, or building material. We also have an overabundance of carbon in the form of CO2 in the atmosphere, with industry producing over 34 billion tons per year. This is why one of the current technological “holy grails” is to develop a cost and energy efficient method of recapturing that carbon and feeding it into a useful production stream at industrial scale. This way a pollutant can be turned into product.

The problem is that CO2 is a stable molecule, and so it costs a lot of energy to break it apart – reversing reactions that produce the energy in the first place. Specifically, we need to split one oxygen off the CO2 to make CO (carbon monoxide). CO can be used in a variety of useful chemical reactions, making hydrocarbons, for example. The way to make reactions happen on useful industrial scales is with catalysts – a molecule that makes a reaction go faster (often by orders of magnitude). Of course the reaction also requires energy, because we want to go from a low energy molecule (CO2) to higher energy molecules (CO and O2). The challenge has been bringing all these elements together.

A new study introduces a new element into the equation – DNA.  This may seem counterintuitive at first, but it makes sense when you put the whole picture together. Researchers at MIT were trying to crack this very specific problem – how do they bring together CO2 dissolved in liquid with a catalyst on the surface of an electrode that will be providing the energy? All these elements need to come together in the most efficient way. Further, catalysts can tend to break down with use, and we also need to get the old catalysts off the electrode and replace them with fresh catalyst. You can do this just by diffusing CO2 and catalyst in the liquid with the electrode and let randomness get it done, but this is highly inefficient, and efficiency is the game.

Continue Reading »

One of the things I enjoy about writing this blog is that it is a conversation. My essay is often just the opening salvo in what turns into an interesting exchange on the topic, and I often learn new facts, gain deeper insight, and if nothing else get better at communicating my ideas. This is why I have a high tolerance for commenters with very different views. I do get rid of the worst trolls that I find are destructive to the conversation, but as my regular commenters know, I set a pretty high bar. I do recommend everyone try to engage meaningfully with other commenters and not just try to “win” with snark and insults. If we all agreed here, the comments would be pretty boring.

Sometimes, however, I feel like I have enough to say in response to the comments that a follow up post is warranted. The conversation about AI art is one of those times, partly because the conversation focused on elements of my post that I feel were ancillary. My post was not really about art. It was about how we respond to disruptive technology, and one way in which some technologies are disruptive. Specifically some technologies automate the technical aspects of creation, rendering obsolete (or at least to a much diminished role) entire sets of skills. My three examples were woodworking, photography, and the recent AI algorithms that can generate art.

In response some commenters noted that crafting a chair from wood is not art. Unfortunately this lead to a discussion about “what is art”, which is interesting, but entirely misses the point. That was not the analogy, and crafting furniture does not have to be art for my analogy to hold. The point was, that a profession of skilled artisans was essentially rendered obsolete by modern technology. Sure, there are people who keep this craft alive, and there is a high-end market for hand-crafted items. But the industry has fundamentally changed. A 19th century woodworker would have a hard time finding employment outside a historical village.

Continue Reading »

Recently an artist names Jason Allen won the Colorado State Art Fair’s competition in the category of digital art with a picture (shown) that was created by an AI, the Midjourney software. This has triggered another round of angst over computers taking our jobs. Some have declared it the end of art, or that it will destroy the jobs of working artists. This development can certainly be a job-killer, but we have to get over it. This, in my opinion, is just an extension of the advance of technology, which ruthlessly destroys jobs while creating new jobs and opportunities. We should not waste a moment shedding tears about those lost jobs, but rather put our energy into adapting to the new reality.

I do think it is reasonable to consider AI artists as just another form of automation and using tools to enhance our ability to create stuff. We can go back to just before the industrial revolution, when, for example, a highly skilled wood worker would make a chair entirely by hand. Even by then automation had had an effect – a productive shop would likely have an assembly line where specialists focused on different aspects of making the chair. Lathes and other tools were used to speed the process and improve precision, but still a great deal of technical skill, developed over years, was required. But soon the job of the highly skilled woodworker would be destroyed (outside of historical theme parks) by machines. A high-quality wooden chair can be made without the need of a single skilled woodworker, assembled by people who only need the skill to operate the machinery. At the time such products were denigrated as cheap knockoffs for the masses.

There are countless such examples. Getting closer to the artistic realm – do you take photographs with either your phone or a dedicated camera? Do you manually set the ISO, f-stop, aperture setting, or measure ambient light levels? Unless you are a professional photographer, the answer is likely no. Computer chips in the camera can do all of that for you. Even professional photographers will use these automated features – they rarely will measure light levels, for example, but let the camera do it. The point is that technology has reduced the technical skill necessary to take a good picture. Now, all you have to do is focus on the composition – the more artistic and creative aspects of photography.

Continue Reading »

As you are likely aware, NASA’s latest big project is the space launch system (SLS) which is the rocket system that will be used by the Artemis program to return astronauts to the Moon. The SLS also contains the Orion capsule, which is a deep space craft capable of holding four crew for missions up to 21 days. It is currently the only deep space capsule, capable of the high speed reentry required for return from the Moon.

Artemis I, and uncrewed test mission, was scheduled to launch on Monday August 29th. This launch had to be scrubbed because of the main engines were not at the right operating temperature. The problem turned out to be a faulty sensor. However, there are limited launch windows (only a couple of hours) and the problem could not be identified and fixed within the launch window, so the launch was scrubbed. It was then rescheduled for Saturday September 3rd. This time the problem was a real leak in the hydrogen fuel tanks, likely a problem with one of the seals. They failed to fix the problem on the launch pad so again had to scrub the launch. Leaking hydrogen is a serious problem; beyond a certain point there is a risk of the leaked hydrogen exploding on launch, and they were well beyond that safety point.

This shows how delicate this whole process is. It may be possible to fix the seal and the leak with the rocket still on the launch pad. However, the batteries used for the abort system are getting to the end of their optimal readiness window, and those batteries have to be swapped out in the engineering building. So the rocket has to be taken off the pad and brought there to reset everything to be ready for launch. This puts the next earliest launch date about six weeks off, in mid October.

Are these launch delays routine and expected or are they evidence that the SLS is a boondoggle, as its harshest critics maintain? I think it’s a little of both.

Continue Reading »

One of the major challenges of space travel is that there are no ready-made resources there. Mars, for example, has no food, shelter, oxygen, fuel, or power. It likely has water, but it’s not certain how much and how accessible. So for now any human mission to Mars will have to bring all recourses from Earth. Getting stuff to Mars is massively expensive, and resupply can take 6-9 months, during optimal launch windows. Keeping humans alive on Mars for any length of time is therefore a very tenuous and expensive endeavor.

One obvious solution is what NASA calls “in-situ resource utilization,” – using stuff that is already there. The Perseverance rover on Mars contains an experiment called MOXIE (Mars Oxygen In-Situ Resource Utilization Experiment) that has taken the first step toward the goal of in-situ resource use. MOXIE was developed by engineers at MIT, it is a lunch-box sized experiment onboard the Perseverance that makes oxygen from the Martian atmosphere. It is essentially a proof-of-concept, and if it works may be the forerunner of far larger machines in the future.

MOXIE draws in Martian air, filters out dust and other contaminants, then pressurizes it. It then applies a process known as the Solid OXide Electrolyzer (SOXE) to electrochemically convert CO2 in the Martian atmosphere into oxygen and carbon monoxide (CO). The elemental oxygen is then isolated in a chamber where it combines into breathable molecular oxygen, O2. MOXIE then releases the O2 and CO back into the atmosphere, but obviously working versions would collect these gasses for use.

Continue Reading »

In my upcoming book, which I will now shamelessly plug – The Skeptics Guide to the Future (release date Sept 27th, but you can pre-order now) my co-authors and I spend a lot of time extrapolating cutting edge technology into the near and medium-term future. What are the technologies that are on the cusp of disrupting current technology and changing our lives? One of them is the technology to build ever-smaller and more capable machines – the technology of the very small. We can dream of having mature nanotechnology, robots at the nanoscale that can manipulate matter at the molecular level, but this is likely still centuries in the future. Between now and then there is a lot of territory, however.

In the meantime we can imagine what the most likely early applications will be. What is the low-hanging fruit? For medical purposes there are some likely early applications, even for robots that are not quite at the nanoscale, perhaps at the micro or even centimeter scale. Tiny robots can be useful as surgical aids. They can be injected through the skin (no incision necessary) where they can make precise interventions, such as removing tumors or suturing blood vessels. This could take microsurgery (which is already a thing) to the next level.

When such robots can be more autonomous or easy to control remotely, another possible early application is to have them crawl along the inside of the large and even medium-sized blood vessels, clearing up plaque and removing clots.  Or they can move along the inside of the intestines, removing polyps and scanning for cancers. In recent decades we have been transitioning from having to open up major body cavities in order to do surgery, to being able to do the same procedures through small holes using cameras and specially designed instruments. This has made many surgeries significantly less invasive, with dramatically reduced trauma and recovery time. Micro surgical robots have the potential to take this to the next level over the coming decades.

Continue Reading »

The big science and technology news today is the Artemis I launch, an uncrewed test flight that will orbit the Moon on a six week flight. I thought I would be writing about that today, but prior to the launch I actually don’t have much to add to the extensive reporting. I’ll probably have something to say after the launch. But there is other space news, this one from the European Space Agency (ESA). The ESA is considering a proposal for space-based solar power, which also makes a nice follow up to my previous post updating solar technology.

This is one of those ideas that, when I first heard it, I thought it was a great idea. There are some obvious benefits to placing solar panels in orbit, then beaming the energy down to Earth. Above the atmosphere in geostationary orbit, the solar panels would receive sunlight 24 hours a day and without any concern for clouds or weather. Panels can also be optimally oriented to the sun without having to worry about the Earth’s rotation. How much of a factor is this? Depends on location of ground-based solar, but for example Arizona can expect to have 7.5 peak solar hours per day, while New Jersey has 4 hours. There is still some off-peak energy production, but less. Overall light exposure efficiency can vary from 20-40%. In space we can get 100% light exposure efficiency. Being based in orbit also solves the intermittency problem. Solar can become a reliable baseload source of energy.

The obvious downside is that it’s expensive to get stuff into orbit, but until you run  the numbers it may not be obvious if space-based solar can be cost effective, compared to other options like nuclear or geothermal. This is the primary reason for the ESA feasibility study. Unfortunately, when you start to run the numbers, and consider all aspects of logistics, space-based solar looks worse and worse.

Continue Reading »

When I first started reporting about solar cell technology around 2005 the best commercially available silicon solar cells (photovoltaics) had an efficiency rating of about 11% (the amount of the sunlight hitting the panel that ultimately gets converted into electricity). They were expensive, heavy, and didn’t product that much electricity, but still it was enough for early adopters and we were at the beginning of the commercial solar industry. That was just before the inflection point when solar started to take off.

I remember reading many solar power news item, detailing some incremental advance, but still with some limitations and uncertainty. But slowly, inexorably, these potential advances added up. Every year solar panels because a little better and a little cheaper. Now silicon crystal solar panels commercially available have efficiencies of over 20%, they have a minimum lifespan of 20 years but many are rated for 35-40 years (and some report 40-50 years), and their price has plummeted, down about 90% compared to 2010. The ultimate potential efficiency of silicon solar cells is often cited as 29%, but using various techniques higher efficiencies have been reported, such as this one in 2019 reporting a 31% efficiency.

The question is – how long will this trend continue? It’s hard to say, but it’s clear that advances in silicon solar panels are not done yet. Already the cheapest form of new power, it will likely continue to get cheaper and more efficient for the next 10-20 years, producing phenomenally cheap energy by historical standards. Even without any major new technological breakthroughs, just incremental tweaks on existing solar technology, this is likely to happen.

Continue Reading »