A very warm welcome to our new subscribers. As a quick catch up, tomorrow's world seems to be backsliding into today, and yesterday's sci-fi looks increasingly more sci than fi.

Artificial intelligence is all very exciting, but we still don’t really know how human human intelligence works, and it would be really nice if we did considering how frequently our brains, and the artificial ones we create, go wrong. 

That’s part of the reason Johns Hopkins researchers have now created mini-brains that exhibit electrical activity and respond as a network. The researchers literally stuck different brain regions together with biological superglue and watched them wire themselves up. With 6-7 million neurons (compared to tens of billions in adult brains) they express 80% of the cell types normally seen in early human brain development. 

🧐 What's in it for me? Besides providing a live action Frankenstein sequel, this could also revolutionise how we study things like autism, schizophrenia, and Alzheimer's, i.e. diseases that affect the whole brain, not just isolated regions. Perhaps more importantly, drug development could finally move beyond animal models that fail 96% of the time in clinical trials. 

💵 Out of the Lab:

Imagine someone asks you to check their CV, you’re happy to do so, and so you do. Now imagine they ask you to proofread their 9 volume history of Anglican Door Knobs. Best thing to do in this situation is say it looks lovely, and move on. Oddly, it seems that AI may be doing something similar, and if it’s not, there could be something much more concerning going on. 

As tech giants pour billions into "thinking" AI models, Apple researchers quietly uncovered something unsettling: Large Reasoning Models (supposedly the more intelligent successors of Large Language Models) may have hit a fundamental ceiling that no amount of computing power can break through. Their study reveals that these models exhibit "counterintuitive scaling limits", where they actually reduce reasoning effort as problems get harder, despite having adequate computational resources to continue.

The implications are potentially staggering. Apple found that the reasoning models outperformed in moderate complexity tasks, but completely collapsed for anything high complexity. Even when explicitly given the solution criteria, the models couldn't improve, suggesting we may have reached the limits of current AI architectures' ability to genuinely reason.

Or they’re just trying to make excuses for why Siri still sucks…

🧐 What's in it for me? This research potentially explains Apple's conspicuous restraint in the AI arms race. If Apple is right, the entire industry may be chasing diminishing returns on a fundamentally flawed approach. For businesses betting big on AI reasoning capabilities, this suggests current systems may never achieve the general intelligence everyone's banking on, no matter how much compute you throw at them.

💵 Out of the Lab:

Lithium, the metal that powers your phone might be the key to defeating Alzheimer's according to Harvard researchers. After seven years of investigation, they found that lithium naturally occurs in our brains but can get hijacked by amyloid plaques, creating a vicious cycle of cognitive decline.

The breakthrough came when researchers identified lithium orotate, a form of lithium that can dodge amyloid's grasp. In mouse studies, this compound not only prevented Alzheimer's but actually reversed memory loss, even in mice with advanced disease. Unlike lithium carbonate (the drug used for bipolar disorder), lithium orotate worked at doses so low they showed no toxicity. Before you starting licking your iPhone though, note the limitations of animal models mentioned in Scientists Grow Miniature Brains That Think.

🧐 What's in it for me? This could transform Alzheimer's from an inevitable decline into a preventable condition, potentially through something as simple as monitoring lithium levels in routine blood tests. Given that Alzheimer's affects 1 in 9 Americans over 65, this discovery could spare millions of families from watching their loved ones fade away. Plus, lithium is dirt cheap and can't be patented, though that might actually slow development since there's less profit motive…

💵 Out of the Lab:

The golden apple snail, once rivalling mosquitos for the most unwanted organism award, has a superpower that could make the darling of doctors worldwide: it can completely regrow its eyes. No one had studied this before, mostly because snails are terrible lab subjects, but a UC Davis researcher realised that apple snails breed like rabbits and are infuriatingly adaptable. 

The snails' eyes are structurally identical to human eyes, complete with corneas, lenses, and retinas packed with photoreceptor cells. After amputation, a new eye grows back in about a month through a carefully orchestrated biological process. The team used CRISPR to knock out the pax6 gene, the same one crucial for human eye development, and found that snails without it simply don't develop eyes at all.

🧐 What's in it for me? If researchers can crack the regeneration code, we might eventually treat everything from macular degeneration to traumatic eye injuries by simply switching on the right genes. Currently, humans can't regenerate damaged eye parts and even successful eye transplants haven't managed to reconnect to the brain. But these snails suggest the genetic machinery for eye regeneration might already exist in humans, just waiting for the right molecular switch.

💵 Out of the Lab:

If you haven’t seen a deepfake video yet, you almost certainly have and just didn’t realise. They’re everywhere, and now Google have built UNITE to tackle this. The AI system can detect video manipulation even when there's not a face in sight. Previous deepfake detectors were basically useless if someone wasn't explicitly face-swapping celebrities.

UNITE uses transformer-based deep learning to analyse entire video frames, spotting subtle spatial and temporal inconsistencies that humans miss entirely. The secret is something called "attention-diversity loss," which forces the AI to examine multiple regions of each frame rather than fixating on faces.

🧐 What's in it for me? As AI-generated videos become indistinguishable from reality, this could be the difference between a functioning society and complete informational chaos. Social media platforms, news organisations, and fact-checkers desperately need tools like this to prevent synthetic videos from going viral and destroying lives. Ironically, the same technology making deepfakes more convincing is now being used to detect them.

💵 Out of the Lab:

Penn State researchers discovered that gold nanoclusters behave like individual atoms with quantum superpowers, including the ability to tune their spin properties by changing their molecular decoration. These "super atoms" can mimic the best properties of trapped gaseous ions (currently the gold standard for quantum computing) while being much easier to scale up.

The breakthrough came when researchers realised they could adjust the clusters' spin polarisation (a crucial quantum property) from 7% to 40% simply by swapping out the molecules surrounding the gold core. Unlike traditional quantum materials with fixed properties, these gold clusters can be chemically tuned like adjusting the settings on a quantum radio.

🧐 What's in it for me? Current quantum computers require dilute gaseous ions that are notoriously difficult to scale up. Gold clusters could provide the same quantum advantages in a solid-state system that's actually practical to manufacture. Think quantum computing that doesn't require a laboratory the size of a small building to keep a few atoms cold enough to work properly.

💵 Out of the Lab:

By 2050, more than three-quarters of countries will have fertility rates below replacement level, meaning populations will shrink for the first time since the Black Death. Mexico went from seven children per woman in 1970 to just 1.6 today. South Korea hit 0.75, so low that its population peaked in 2020 and is now falling at an accelerating pace.

The ripple effects will reshape entire industries. Housing prices could collapse in regions losing population while skyrocketing in the few places still growing. Agriculture will face labour shortages just as older populations need more healthcare. Universities will close while retirement homes multiply. Sub-Saharan Africa, meanwhile, may account for half the world's babies by 2100.

However, there may be a light at the end of the tunnel: many cite financial constraints as the primary reason for not having children. As populations shrink and labour becomes scarce, wages should rise and housing costs fall, potentially creating conditions for fertility rates to stabilise. The question is whether societies can adapt quickly enough to manage the transition without economic collapse.

Or we just double down on this:

Until next time, stay curious.

Like what you're reading? Share toast with a friend.