Plenty of science news this month but I’ve avoided talking about AI. Things are moving too fast in that space and with everything still up in the air, it seems unwise to tout early findings and announcements before things have settled down.
Almost all of the stuff I have this month are about biology so we’ll start with the ones that aren’t. This article talks about reliable methods to detect when someone is lying. Obviously there are all kinds of devices that purport to do so with dubious efficacy and law enforcement officers are trained to look for telltale signals. This group of researchers posit that it is necessary to look for only one type of signal: the level of detail in the story the person is telling. They even tested this by separating groups of students who really did an activity and who were told to simply lie about it. Those who actually did were able to provide much more detail in their account and that in turn is a more reliable method of telling who is lying and who isn’t.
Next we have a paper about a particular Mayan calendar which had stumped researchers for a long time. The calendar is known for having a 819-day cycle and researchers had difficulty synchronizing it to the movements of the visible planets. This paper shows that by increasing the length of the calendar to 20 periods of 819-days, it is possible to create a pattern that matches with the synodic periods of all of the planets visible to the Mayan civilization. What blows my mind is that this is effectively a 45-year calendar proving just how long the Mayans must have watched the skies to learn and understand the larger cycle.
Moving on to biology, we have a paper that discusses how plants emit ultrasonic sounds when under stress. The experiments they conduct confirm that plants do emit airborne sound that can recorded using microphones but that machine learning models that differentiate between the sounds to tell when the plants are stressed by drought, cut or normal. To be fair, this doesn’t mean that plants are sentient as they lack a nervous system and the sounds that are being emitted are most likely a physiological consequence of the different conditions they are subjected. But as the authors note, this is still information that can be used, either by humans or other plants and animals in the vicinity.
Next we have two articles that are vaguely related. First is further news about the discovery of fungi that can consume plastics, specifically polypropylene that is used in takeaway containers and cling film. The fungi are already naturally present in plants and soil so there is no fear of them being dangerous. It still takes a great deal of time to fully degrade the plastics though the process is accelerated by ultraviolet light and heat. But if it could be scaled up, this could be very useful way of breaking down plastic waste so that it won’t stay forever in the environment.
In what makes for a great example of the familiar ‘life finds a way’ tagline, researchers have discovered that coastal creatures have made the so-called Great Pacific Garbage Patch their home. This is the well known concentration of plastic debris in the Pacific Ocean that most people will have already heard of. The researchers found that many species of coastal organisms, meaning crabs and anemones that usually live in coastal areas and not in the middle of the ocean, are colonizing the plastic waste and forming communities there. Of course the garbage patch remains a problem but it serves as a powerful reminder to us that life adapts and goes on in spite of what humans do.
Finally on a lighter note, here’s an article about how pet parrots were taught to use touchscreens to make video calls. They found that the parrots were indeed able to make such calls to other birds and seemed to enjoy socializing with each other through the screen. The project is too small in scale to draw firm conclusions but as parrots are usually kept as solitary pets while they live in flocks in the wild, this may actually be a useful way to help alleviate the psychological issues that many pet birds develop.
Obviously generative AI is the in-thing now. I’ve been playing around with ChatGPT and every one of the publicly available models for months now and it really makes me feel like we’re living in a sci-fi scenario. There are new announcements in the field practically every day and it’s exciting to try to keep up.
We’ll still start with some non-AI news first. Most people will have heard of how sperm counts in men have been falling. This study talks about how men who do physically demanding jobs have much higher sperm counts compared to the baseline. This may be linked to how they also have higher testosterone levels. Most other studies don’t show a direct link between testosterone levels and sperm count but it’s possible that they could be linked in more indirect ways.
Next we have a story about an American man who developed an Irish accent after being afflicted with prostate cancer. This is thought to be an instance of paraneoplastic neurological disorder in which a cancer patient’s nervous system is attacked by their immune system. This isn’t an altogether new finding but it is the first instance of the cause being prostate cancer metastasizing and affecting the brain. Plus I’m always fascinated by how personalities and demeanors of people can change drastically as a result of injury or disease.
There are far too many announcements in the AI space to cover so I’ll limit myself to news that is more applicable to the mainstream. By now most people will have heard of GPT4 and how it accepts images as an input, in addition to text. This article goes more deeply into such multimodal inputs for AI, integrating text, images, video and sound. It talks specifically about an in-house model made by Microsoft called Kosmos-1 that is unavailable to the general public and is supposedly far better than anything else. The important thing is that such a model would be much closer to a general purpose AI that we think of in science-fiction, able to receive all manner of sensory input from the outside world and respond accordingly.
Next is a paper about the image generation model Stable Diffusion. The researchers talk about how they are able to take images of human brain activity from fMRI and run them through the model to reconstruct an image of what the human subject was seeing at the time of the FMRI scanning. The results are extremely impressive and the surprising thing is that the model is able to do this without any additional training or fine-tuning.
There’s been a scarcity of interesting science news this month though there is one announcement that is pretty big news. Also, a couple of bits about technological applications. I ordinarily don’t like to cover these but there’s little else to talk about.
The big news is a couple of new papers that links black holes with dark energy. They claim that old black holes seem to be growing far faster than the models predict and propose that black holes contain a well of vacuum energy. They further argue that the amount of energy stored in black holes in this way matches the amount of dark energy that has long been posited to explain the accelerating expansion of the universe but has never been observed. Needless to say that this would be a major discovery if proven correct but at this point should be treated as a speculative idea rather than anything substantial.
Next, a couple of articles on technology. The first one talks about toroidal propellers. They’re so strangely shaped compared to the usual ones that we see on common objects like fans that you really need to look at a picture of one to understand how different these are. The original motivation for developing these was to reduce noise for aerial vehicles such as drones and these are indeed quieter. But when they researched if there would be a thrust-efficiency penalty they discovered that they are actually more efficient and are structurally stronger too. The downside is that these shapes are much more complex and so cost more to manufacture. But the gains are so significant that I imagine they will be the new standard soon especially since it applies underwater too and in military applications. I like this discovery particularly because people have been searching for more efficient propellers for a very long time and it’s kind of insane that there is still low-hanging fruit to be taken advantage of in this way.
Then there’s this highly speculative proposal for the USA to harness the Yellowstone Supervolcano to generate electricity. The technology, which involves drawing out the heat of the volcano to drive turbines, seems conventional but the scale of the effort is breathtaking and being able to do it safely is a big question. The authors point out that energy produced would be green by definition and it would even relieve some of the pressure of the volcano, forestalling the long-predicted and disastrous eruption. This will never actually be attempted of course. when even much better understood applications like nuclear power plants face so much opposition and are so difficult to fund. But I like the science-fiction aspect of such grand yet sadly impossible proposals.
Finally this feels like something that shouldn’t need to be said given the preponderance of evidence already available but some people just won’t give up. A large, randomized, placebo-controlled trial has just found that use of the drug ivermectin did not improve outcomes for patients with COVID-19. This trial and its conclusions should be the final nail in the coffin that the drug holds any effectiveness for this purpose but of course, given how people are, it won’t put this particular conspiracy theory to rest.
This month’s pick of developments worth paying attention is more substantial than usual and I think at least a couple of them will have serious long-term implications.
Starting with lighter news first, the announcement that lab-grown retinal cells are capable of reaching out and connecting to other cells is exciting. Lately many researchers have been experimenting with all kinds of organoids, organized clusters of cells that can be regarded as very primitive versions of organs. This is a similar example that involves coaxing stem cells to develop into light-sensitive retinal cells. The latest development is to show that these cells are capable of forming synaptic connections to other cells. We’re still a long way from growing artificial eyes in this way but it may be possible to use this technique to treat certain eye disorders and injuries.
Next is a paper that claims that chimpanzees possess risk-taking behaviors that are similar to humans. This means their risk assements are broadly correlated across domains, they shy away from ambiguous situations, males are more willing to take risks than females and that appetite for risk peaks in young adults, patterns that are broadly similar to humans. From a philosophical perspective, these results are interesting as chimpanzees are of course not influenced by human culture, suggesting that there may be an evolutionary basis to risk-taking preferences.
Then we have this longer article about how the old model of animal learning, famously known from Pavlov’s experiments with dogs, is slowly being debunked. The old model looked forward and stipulated that learning works by trial and error. A prediction is made in response to cues and adjustments are made to future predictions as a result of how correct it was. The new model looks backwards, such that when a reward or punishment is received by the animal, it looks back in its memory tp search for what might have caused the event and learns accordingly. New experiments suggest that the new model seems closer to the truth. Though the difference between the two models may be subtle to most of us, this amounts to a revolution in neuroscience with major implications.
Finally some of the early results from the James Webb Space Telescope are in and they are suggesting that cosmologists’ ideas of how the universe formed may be all wrong. The new telescope effectively lets astronomers see further back in time than previously possible and the surprising finding is that there are too many mature galaxies out there and that they are too bright. The researchers expected to see more chaotic and primitive structures from the distant past, yet it seems that the past looks too similar to the present than what is predicted by current models. No prominent astronomer has quite come out to explicitly say it yet and they’re all still reviewing the results but I believe this shows how little we really know about the formation of the universe and the models we have are really just educated guesses.
Fewer news articles this month but that is also because I have decided not to highlight some of the more overrated and salacious articles that have gone viral recently including the claim of obtaining net energy from a fusion reaction and the paper that female snakes are discovered to have clitorises.
The easier to understand bit of news I have this month is this paper about progress being made towards the development of a universal flu vaccine. As most people should know, existing flu vaccines are seasonal as doctors have to observe which particular strain of flu is currently most widespread and choose the appropriate vaccine. So it makes sense to try to use the new mRNA technology to create a vaccine that encodes antigens from all 20 known influenza A subtypes and influenza B lineages. Testing on mice and ferrets showed that it worked to generate the desired subtype-specific antibodies when challenged by different viral strains.
The next article is too difficult for me to really understand, let alone summarize in any reasonably accurate manner. Doing my best, I will simply say that one of the most important problems in pure mathematics is the Riemann Hypothesis which claims that the distribution of prime numbers is not completely arbitrary. The news is that a Chinese-American mathematician Yitang Zhang has made progress in this topic. To be fair, he only claims to have disproven a weak version of a conjecture that serves as a counterexample to the hypothesis. Nevertheless if proven correct, this does represent progress and given how important prime numbers are, in modern cryptography for example, should be considered a huge deal.
I usually skip over particular inventions but I like this one. It’s a new way to stop condensation forming on glass surfaces like spectacle lenses and windows that doesn’t require an outside electricity supply. It’s actually a thin layer of gold in between layers of titanium oxide. This absorbs solar radiation and heats up the glass enough to get rid of condensation. Unfortunately it doesn’t work at night and while you might be concerned about the cost of gold, so little is used that the cost is actually quite low. This is one invention that I’m confident will make it to the market almost immediately.
ChatGPT has been making waves lately and I’ve been playing with it myself. However there’s is no one definitive article I can link to for now as everyone is still experimenting and learning. In general this year has been the big breakout year for AI and one of my favorite examples is the development of a bot that can play the boardgame Diplomacy. This article summarizes the rules well enough. The upshot is that it’s a wargame set in Europe but the rules are simplified and the game emphasizes communication between the players. They must make deals with one another to survive and there are inevitable betrayals. The bot is able to convincingly act like a human including talking to the human player in plain text messages, making deals, relating what is being said to what is happening on the board and making plans. I just love this YouTube video of what it feels like to play against the bot.
We’re back to having more science news than I can reasonably cover. Hopefully this represents a decent selection of the most significant announcements.
Starting with the easier to grasp and happier pieces of news, we have the welcome success of lab-grown blood transfused into a human. Stem cells are extracted from a normal donation of blood, encouraged to grow and then guided to become complete red blood cells. This effectively creates a larger usable supply from small donated amounts. So far the trial uses only very small amounts to determine safety and the process is too expensive for widespread use. It may be a viable means to help with extremely rare blood types however.
Another piece of good news talks about some unexpected benefits of protected marine sanctuaries where fishing is banned. The normal benefit is that population numbers of valuable species inside the protected area boom and overspill into outside areas where fishing is allowed. A study in Norway however also found that lobsters inside the protected grow larger and exhibit bolder behaviors. In effect, as laws ban the harvesting of lobsters below a certain size, lobsters have grown smaller and become more timid to avoid being caught in traps. The existence of the protected areas counteract that, which improves the quality of the fishing just outside of them as well as the quantity.
In less happy news, a set of studies examined whether or not cannabis use can somehow promote creativity. Using outside evaluators to assess the quality of creative work, the studies found cannabis use didn’t significantly improve the creativity of users. But it did make them happier and made them believe themselves that they were more creative. But as the article notes, it’s still possible that cannabis use boosts cognitive abilities that weren’t measured or that already creative people are simply more likely to use cannabis in the first place.
The next paper is from the field of macroeconomics and discusses dynamic stochastic general equilibrium (DSGE) models. These models are the predominant framework of macroeconomic analyses. The authors of this paper subject a version of this model to some statistical tests, effectively testing if the model can predict its own simulation. They also test how well the model fits nonsense data. Even with centuries of data, the forecasting error is very high and swapping in nonsense data might actually yield better results. I don’t know enough about how these models work and about statistics to judge the validity of this paper for myself but it seems like a serious challenge that the field of macroeconomics must answer.
Finally my favorite of the announcements this month is about a novel way to measure time. It involves pumping atoms with lasers such that they enter a high-energy state called a Rydberg state. The movements of electrons under such conditions are subject to quantum effects and are described as a Rydberg wave packet. More than one such wave packet interfere with one another and this interference can be used as a measurement of time. The interesting part is that there is no need to predefine a starting time so scientists can measure any event they want to observe and compare it to the signature of interfering Rydberg states to note how long it lasts. Plus this measurement of time is entirely self-contained and does not rely on some other measured quantity.
In addition to the Nobel Prize announcements this month and the reactions and commentary that always follows, there’s been plenty of cool news science, enough that I’ve had to pick and curate.
We might as well start with the image that has captured everyone’s imaginations this month. It’s an update to the iconic Pillars of Creation image originally taken by the Hubble Space Telescope in 1995. The new image was taken by the James Webb Space Telescope using its Near-Infrared Camera to view that region of space about 6,500 light-years away. The breathtaking visual captures proto-stars being formed amidst clouds of dust and gas, the powerful gravitic forces involved propelling the clouds of materials around to form these distinctive shapes.
As amazing as that image was, the one article that captured my imagination this month is this one about implanting corticoid organelles into rats’ brains. I’ve covered the subject of these organelles before, but briefly these are small agglomerations of human nerve cells, cultivated from pluripotent stem cells. The idea is to study simplified versions of complex organs, this one being a simplified model of the brain. This particular experiment involved implanting the organelles into the part of the rat’s brains responsible for the sense of touch. After the human and rat nerve cells had connected up properly, they tested if the organelles could properly respond to sensory input, blowing air on the rats’ whiskers, and if it could direct the rat’s behavior. Both proved true and though the ethical issues with such work are worrying, it makes the important point that such artificial, simplified brains can in principle be made to integrate with live animals.
Another great article is this one about how pandemics that happened far in the past continue to affect us today. Analyzing the DNA extracted from victims of the Black Death in the 14th century plus those who died many decades after the plague, the team pinpointed a variant of one particular gene that seems to confer some protection and showed how it became more pervasive in those who survived the plague. Experiments with cultured cells further showed that the variant version have macrophages better able to kill the bacterium that causes the plague. Yet there is a downside as this variant is also linked to a greater susceptibility to autoimmune disorders which essentially means that the immune system has been tuned to be overactive against all kinds of threats.
I don’t like to put too much weight on socioeconomics studies so consider this as just one data point among many. This paper studying how participation in markets affect moral behavior uses data from experiments done in some villages in Greenland. After controlling for other factors, it finds that increased market participation leads to more universalism in moral decision-making, meaning that the villagers saw themselves as part of a wider community instead of valorizing their own co-villagers above outsiders. It’s the kind of finding that is intuitive and perhaps a little too good to be true but I certainly would like it to be.
Next is another paper that is sure to be politicized. It summarizes the current state of knowledge regarding vegan and vegetarian diets to argue that strict adherence to a purely vegan diet results in too many nutrient deficiencies. In any case, our ancestors consumed plenty of meat, eggs and seafood so our bodies are adapted to it. The paper recommends a diet of mostly unprocessed plant-based foods balanced with modest amounts of wholesome animal foods.
Finally here is a long article from Google about using AI techniques to discover novel algorithms. It uses the example of matrix multiplication that most people who have some mathematics education should know how to do. In 1969 the German mathematician Volker Strassen showed a way to do the calculation more efficiently at least on 2 x 2 matrices yet until now no one knows how to extend this to larger matrices or if even better algorithms are possible. The article talks about Google using a system they call AlphaTensor to gamify the process of searching for better algorithms and actually succeeds in finding novel solutions though it take a far mathematician than myself to understand how to use the new algorithm. Since matrix multiplication is used in many, many fields of computing even the slightest optimization makes a huge difference. But this also raises the old fear that AI-led discoveries will soon lead us into territory that human minds will struggle to understand.
