This question will not go away. I am getting asked this questions at conferences now. The media also seems to think this is a compelling question, and they are fueled in part by the tech industry itself, some of whom indulge this question. I will admit the answer has become more complex and nuanced than a simple “no”, but it is still mostly no. Certainly the rise of neural networks, deep learning, transformer technology, and specifically large language models have ushered in the era of  – are we really sure current AI is not conscious? Our intuition tells us that they may be – LLMs are getting scary in how sophisticated they are and in their ability to simulate complex human conversations. Five years ago I was with the crowd who argued that LLMs are basically really good auto-complete apps, but this pithy sentiment no longer covers it. Even if they do not have a whisper of conscious awareness, they employ context-dependent logic. They are “intelligent” in their own way. And they are rapidly getting more impressive. So I get why people would be asking the question. Everyone I know who works with AI apps has their example of when they were blown away by an interaction with an LLM. For me, I was trouble-shooting a web-based app I was using, and ChatGPT was very effective at helping with the trouble-shooting and producing a differential problem list. At one point it asked for a screen capture so it could have more information, and it gave me a cogent analysis. But here’s the part I found surprising – it “noticed” another window open in the background and inferred from this that I was also using this other app in the process, and used this to generate further potential sources of the problem. As impressive as this is, I still think they are not conscious. I think the illusion of consciousness is generated by the fact that LLMs are simulating language, and we associate language with consciousness. No one thinks Midjourney, an art making program, is conscious, even though this is based on the same fundamental technology. Further, LLMs are still brittle in their own way, although the brittleness is being rapidly programmed out of them. What this means is that you can break them, or easily confuse them, in order to reveal that they are simulating language rather than truly thinking. This brittleness is revealed in several ways – sensitivity to exact prompt wording, accepting false premises in the prompt, and failure to recognize constraints. If changing the language but not the real meaning of a prompt gives different answers, then the LLM is responding statistically to the language, not genuinely reasoning about the information. Each iteration of LLMs is getting better at plugging these brittleness holes, but they are still there. Still, this doesn’t mean that there is no contextual reasoning happening. I think the best current summary is that LLMs are a combination of statistical language prediction (the auto-complete thing) and contextual reasoning. This doesn’t mean that they are conscious, however. What I believe it means is that humans find it surprising how much reasoning a machine can do without consciousness. Perhaps our own reasoning is more procedural than true deep thinking than we perceive or feel comfortable truly grappling with. We are also algorithmic machines, with different strengths and weaknesses than LLMs. But we are also conscious. So what is the difference? The difficulty in answering this question is that neuroscientists are still not certain what causes consciousness in the first place. This does not mean it’s a complete mystery or we know nothing – it means it is a complex multifactorial, and perhaps emergent, process that is not easily understood or explained. But we do have a lot of knowledge about it. There is no reason, in my opinion, to think that the different is biology vs silicon. I do believe that silicon can become conscious. Rather, it’s the process that matters, and we know some things about the process. Human consciousness requires that the brain be constantly self-activated. Turn off this activation (the brainstem activating system) and you will be in a coma – not conscious. We also know that the brain is constantly active and communicating with itself, in an endless loop of one network in the brain talking to other networks which talk to other networks and back to the original network, in a constant stream. It is this constant stream that is our consciousness. We also know there is a threshold, at least for wakeful consciousness. You need at least 45% of your cortex to be functional in order to generate wakeful consciousness. Are people who are not wakeful still conscious to some degree? Probably, but it’s difficult to understand what this means. If you are not making memories, for example, then we cannot interrogate what that consciousness was like. If people can still respond to their environment, does that imply consciousness, or could that simply be an unconscious reaction? For straightforward LLMs – prompt-response, I do not think they can be conscious, at least not human-level conscious. They are doing none of the things that are necessary for vertebrate consciousness. Things get a bit trickier for agentic AIs, that can iterate their own behavior, but still this is more like a self-iterative prompt than a true stream of consciousness. But admittedly, frontier agentic AI is starting to blur the lines a bit. But here’s the thing – even if I cannot say with absolute certainty that some of these frontier models have still not crossed over some fuzzy line into a type of consciousness, I can say that this consciousness must be fairly minimal compared to human-level consciousness. I can also say that such minimal consciousness is not what is speaking to you. It is not reasoning and generating the output of LLMs. We know what’s doing that, and it does not require consciousness or result from any process that can be reasonably inferred to be conscious (again, it may just be a prompt-response). Therefore, arguing that we cannot rule out some consciousness emerging from the processing of massive data centers does not mean you are speaking to that consciousness through your LLM. I may turn out to be wrong, but I don’t think we will back into AI consciousness through some unanticipated emergent property. I think we will have to design systems to do the kinds of things necessary for consciousness. It may be a deliberate attempt to make a conscious AI, or many of those elements may have some separate utility. Either way we will be able to look at the system and see that it is functioning in a way that can plausibly result in consciousness, and that consciousness will be proportional to the power and functionality of the system. LLMs are not it. They are not even on a path to become conscious. They are doing contextual reasoning, and it is getting steadily more impressive, but they are doing it without consciousness. The post Could AI Be Conscious? first appeared on NeuroLogica Blog.

Solar power as a source of electricity has been on a meteoric rise, and has crossed some encouraging milestone recently. Every way you look at it, solar is booming. This is driven primarily by the decrease in the cost of adding solar power, so let’s start there. The Berkley Lab has been tracking the cost of solar power in the US for years, so they are a reliable source of information. They get direct data from the actual price paid by consumers, and break down the cost of panels, the cost of installation, and all the “soft” costs that  are part of the industry. In 2009 the total cost of installing solar between $7 – 8.70 per watt. An average US residential home installs a 6-8 kW system, with the average increasing over the years. In 2025 the average size was 7.7 kW. Let’s use a 7 kW system, at $8 per watt, that’s $56,000 total installation cost. Today the average price of installed residential solar in the US is $3.6 per watt (but it is $3.0 if you pay upfront, $4.50 if you finance). So that same 7 kW system now costs $25,200. (If you pay for it outright, the cost drops to $21,000.) So over the last two decades the total const of installed residential solar has dropped by about 60%. This is without considering any tax breaks or incentives. The reason for the decrease is partly that solar panels themselves are cheaper, and they are more efficient, so a 7kW system requires fewer panels. Installation costs have decreased by about 50% over this time, largely due to economies of scale. However, the US still has higher soft costs for solar than many other industrialized nations, and this is mostly a matter of the regulatory system. So there is some regulatory efficiency to be gained. Solar companies themselves also have some possible efficiency gains. Fortunately, the net price for solar continues to go down, with another 50-60% decrease in total price possible even without further technology gains. Commercial and grid scale solar is even cheaper – with large non-residential installations down to $2.4 per watt. It is primarily for this reason that solar is now the most common new source of added grid capacity. Worldwide, wind and solar are responsible for 85% of new grid capacity, with solar making up 73%. Worldwide there is now about 3 TW of installed solar capacity. One TW was added in just the last two years, showing the exponential increase in solar installations. Solar now generates 9-10% of the world’s electricity. This is led mostly by China, which as 50% of the world’s installed solar. For a long time, while wind and solar were increasing rapidly, they were more than offset by the total increase in electricity demand. This is still much better than not installing renewable or low-carbon energy, but it meant that fossil fuel use was still increasing. For the first time, however, total electricity from fossil fuels decreased, by 0.2%, without being caused by an economic downturn (like COVID). This is a tiny decrease, but it is a potential milestone if we have truly turned a corner. Even better, this decrease is driven primarily by coal, which is the dirtiest form of energy. Natural gas plants are still increasing, because they are still necessary for peaker plants, and because of the increased demand of data centers. The rise of solar is helping push down demand for fossil fuel energy, but there are two other factors that will be critical to push them down further. The first is grid battery storage. Battery technology is also improving rapidly. The advent of cheaper sodium ion batteries (which also bypass the lithium bottleneck) is seeing battery grid storage rise considerably. More and more of those home solar installations include battery back up. There are also new innovations, such as plug and play home battery backup, where you can add storage capacity in a modular fashion without any installation costs – just plug them in. Regulations need to catch up to this technology, allowing them to be used for whole home backup, but for now they can be used for targeted backup, like for your refrigerator, or your work station. Grid storage has the capability of reducing reliance on natural gas plants. So does nuclear. We seem to be at the beginning of a revival of the nuclear power industry. This is necessary even if we wish to simply maintain the current percentage of nuclear on the grid, which will help keep fossil fuel use down. The bottom line is that the cost of solar has dropped precipitously, and so if you have not looked into it recently, you should take another look. The prices I gave above were for a fairly large home. The total price of the system scales with the size of the house, which is a marker for the financial resources of the owner. Let’s say you have a more modest house that only needs a 5 kW system. If you pay for it outright, the total cost would be $15,000. The federal tax incentive just ended on Jan. 1, 2026, which is unfortunate but is defensible due to the fact that the cost of solar has dropped so much. However, 30 states still have some incentive, which can save thousands of dollars. But that aside – $15,000 is very affordable. How much this would save depends on a lot of variables, but for an average home in the US it would be between $800 and $1,500 per year. That means the payback time is between 10 and 18 years. After that, you essentially have free electricity. The useful life expectancy of a new system is 25-30 years, roughly double the payback time. Most solar companies now bundle solar installation with batter backup systems. This allows for peak shaving, and backup during outages. There is potential money savings here as well (including every time you don’t loose a freezer full of food). Further, many states offer incentives for batteries as well, if you allow them to use your batteries for grid storage. A battery system can also keep your solar panels functioning during a power outage. The trend lines for solar and battery grid storage are pretty dramatic and there is every reason to predict that they will continue, as further incremental technology advances take place and economies of scale kick in. The post Solar Is Booming first appeared on NeuroLogica Blog.

I am back from an almost three week trip to LA, Sydney and Christchurch. Sorry I did not have time to keep up with my blog over that period – in which we recorded four live SGU shows, produced three days of conference content, put on two stage shows, and hosted several other events. It was a tremendous amount of fun, but also a great deal of work. I am now jet-lagged and fighting off the remains of a viral infection I picked up along the way. Such trips always reinvigorate my dedication to science communication and scientific skepticism. I appreciate every conference-goer who came up to tell me about their personal journey and the role the content I have helped produce had played in it. I thought I would share some further observations from the trip. First, while the skeptical movement, such as it is, has been through a lot, it is still alive and well. The crowds we are able to attract to these events remain robust, and in fact are larger than many past events. There is a lot of enthusiasm for science and critical thinking. There is also a deep hunger for developing the skills to navigate our increasingly complex world. How do we deal with what social media, growing misinformation, deep fakes, and now artificial intelligence has wrought? People also have a deep desire for community – the kind of community that comes from getting physically together for a shared purpose, not just online.  The number of people we can reach in these physical conferences and meetings is much smaller than online, but I think it remains a critical complement to online content. It also reinforces what I think many of use have learned from the “social media” phenomenon. Online communities are not the same as in-person communities, just online. People interact differently in person and online. The dynamics of social media were simply not what many of us expected. I don’t think we should abandon social media (although I know people who make a reasonable argument that we should), but rather we need to have a more realistic view of its strengths and weaknesses, its vulnerabilities and psychological effects.  The conference goers were pretty similar, demographically, to what they have been over the past 30 years, which means there seems to be a steady state of younger skeptics entering the movement. There are more women than in the past, but still very few people of color – a social nut we have yet to crack. The issues about which people are interested are dramatically different than in the past. At a similar conference in the 90s or even oughts, there would be many questions about Bigfoot, UFOs, astrology, snake oil, and similar pseudoscience. Today people are interested in science denial, misinformation, AI, the trans debate, and similar societal issues (although snake oil is still prominent, perhaps more so). This is also the first big international trip I have taken since COVID and Trump 2.0. People in Australia and New Zealand had similar reactions, it seems. They still consider the US an ally, but they are deeply saddened, frightened, and perplexed by US politics. The first Trump administration could be dismissed as a fluke. The second, they reasoned, is evidence of something deeply wrong with American politics, something that might endure past Trump. This means that the US is no longer a reliable partner, and they must hedge their bets. They need more independence and need to carefully calibrate their relationship with China. There was also a strong consensus that this change is permanent – we will never fully recover our place in the world post-Trump. Interacting with people outside the conference was also interesting. Many people were still friendly, but some were cold or even passively hostile when learning we were Americans. It was nothing we couldn’t overcome with a bit of humor, but it was still disconcerting. It was also interesting being away from American media for three weeks. New Zealanders are chill, what a local New Zealand political scientist described to me as “low voltage”. Everything seems to be working fine and their politics are fairly low stakes. Australians are a more rowdy bunch, but still pretty low drama. It was just refreshing being away from the constant culture wars and American media stoking outrage and controversy at every turn. It really brought home how toxic American politics has become. But now I’m back. I guess I have to start looking at American news again. Perhaps I’ll wait until my cold is over. The post Back From Down Under first appeared on NeuroLogica Blog.

I often get questions like the one below: “I live part time in Japan. Everyone sees hydrogen powered generators, trains, trucks, etc regularly. High density population means it’s easy to get enough synergy to justify the infrastructure. So sad we aren’t doing more to utilize this great tech. Hydrogen combustion engines emerge as cheap alternative to fuel cells.” https://asia.nikkei.com/business/energy/hydrogen-combustion-engines-emerge-as-cheap-alternative-to-fuel-cells This is often how news about advances in hydrogen tech are framed. The problem is – the limitations with hydrogen technology have nothing to do with the tech itself, so tech advances are mostly irrelevant. Also, hydrogen combustion is not a better solution for most use cases then hydrogen fuel cells. Fuel cells use an electrochemical process to combine hydrogen and oxygen, producing electricity and water. They are about 60% efficient, produce no pollution, and have no moving parts. Hydrogen combustion operates more like a regular engine, but with hydrogen as the fuel. They are about 40% efficient, produce nitrogen oxides as pollution, and have moving parts that operate at high pressure and temperature. They can be, however, more powerful for heavy applications and are cheaper to build. For cars hydrogen combustion is a terrible idea. Maybe there is a use case for large generators, trains, and heavy trucks. Even there, however, there are current limitations that the technology of hydrogen engines or fuel cells do not address. 1 – Storage is a problem. They never figured out the storage problem, so just reverted to compressed hydrogen gas. There are efficiency, range, and safety issues with this. Plus, hydrogen is very leaky and destructive to infrastructure like pipes. 2 – Only about 1% of the world’s hydrogen production is green. Most of the rest is gray – essentially stripped from hydrocarbons. This is actually worse then just burning the hydrocarbons for fuel. If we do manage to ramp up green hydrogen production, it should be used first in industry, like steel production. Massive green hydrogen for transportation is a long way off. 3 – Even if we solve 1 and 2, hydrogen cars are less efficient than battery EV, and always will be (60% vs 80% efficient). This is just physics. Further, battery tech has simply advanced more quickly than hydrogen, and it continues to advance. Hydrogen lost this technology race. We may be able to fix the first issue with new materials, but until we do this is a major limitation. This is the main reason that the “coming hydrogen economy” promised back in the early 2000s never happened. There are three promising ways we may solve the second issue. The first is scalable green hydrogen production. If we had solar arrays or wind farms generating electricity to electrolyze water into hydrogen and oxygen, that could produce green hydrogen. The problem here is – we would be better off using that green energy for electricity. Round trip energy efficiency (electricity to hydrogen back to electricity) is only 30-40%. Better to use the electricity directly. Until we have decarbonized the energy infrastructure, don’t use green energy to make hydrogen for light transportation. Where there may be reasonable use is for hydrogen for heat-intensive industries, like steel, and for heavy vehicles like trains and ships. The second possible solution is if there turns out to be vast reservoirs of hydrogen under the ground we can tap into (so-called gold or white hydrogen). This remains to be seen, however. A third potential source is as a byproduct of nuclear reactors. Any reactor can make some hydrogen by radiolysis – splitting water by radiation. High temperature reactors can also make hydrogen through thermal methods. And any reactor can use their electricity for electrolysis, but this has the same issue as using renewable power. The third issue I think is just inherent to these processes. Battery EVs are likely to be always more efficient than hydrogen fuel cells. There may have been a window 20-30 years ago where hydrogen fuel cells could have leap-frogged BEV’s, but that window is now closed. Battery technology won, and also continues to steadily improve. Even if we make progress in hydrogen fuel cells or hydrogen combustion, we still have a hydrogen storage and transportation bottleneck. There have been advances here as well, but they come at a cost. Liquid Organic Hydrogen Carriers (LOHCs) can bind hydrogen to a fluid for easy storage and transportation, then heat the fluid to release the hydrogen. However, this has massive infrastructure and energy requirements, and would further reduce the energy efficiency of hydrogen. We have also engineered better storage tanks – Type IV tanks, which feature a seamless polymer liner fully wrapped in high-strength carbon fiber. This doubles the pressure under which hydrogen can be stored, doubling the range of hydrogen fuel cells. But it takes 12-15% of the energy stored in the hydrogen to compress it to these higher pressures. A hydrogen economy for transportation would have massive infrastructure needs, from production to pipelines, storage, and distribution, likely to take decades. This is all just to get us to a system that is less efficient than BEVs with similar range. Meanwhile there is already existing battery technology with twice the range of common BEVs today (silicone anode Li ion), or similar range at half the cost (Na ion). Solid state and lithium air batteries could potentially five times or more today’s energy density (1,500 vs 300 kWh). Meanwhile we are already near the theoretical limit of compressing hydrogen (700 bar). At 1000 bar hydrogen atoms repel each other and you get exponential energy requirements for further compression. If we are going to invest in infrastructure, those investments should go to fleshing out a fast-charging network for EVs and securing raw materials for making batteries. The post Hydrogen Tech first appeared on NeuroLogica Blog.