News Feeds

A team of researchers has developed a new tool designed to revolutionize hardware troubleshooting, with the help of 3D phone scans.

Researchers have developed a new material that, by harnessing the power of sunlight, can clear water of dangerous pollutants.

Floating platforms of wood could draw up seawater and help it to freeze, seeding the formation of new sea ice

Telescope optics can be made of mirrors or lenses, but in both cases, they're bulky and need to follow a strict mathematical curve to focus light. Researchers have shown that it's possible to make a completely flat lens that focuses light. Traditionally, this has been done with Fresnel lenses, but they distort colors. Their new technique carves tiny concentric rings into a substrate that matches the wavelengths of different colors, allowing a full-color, in-focus image.

There's a small chance of very high warming even with moderate future emissions, according to a computer model exploring what could happen in the next thousand years

We’ve pumped huge amounts of carbon dioxide into the atmosphere that needs to be paid back. Large animals like wolves, bison and whales may already be tackling the problem

One of the areas that Luana Maroja and I highlighted in our “Ideological Subversion of Biology” article, which analyzed six misguided statements about biology made in the service of ideology rather than scientific truth, was the last one:

6. Indigenous “ways of knowing” are equivalent to modern science and should be respected and taught as such.

Now both of us believe that indigenous people can produce and have produced knowledge, even though it’s usually of a restricted nature: trial-and-error truths limited to the geographic area that a group inhabits. But ideology, which has bought into the “authority of the sacred victim” mindset, has gone beyond that, as I’ve often written about New Zealand.

The assertions of adherents to this trope fall into several areas and share several characteristics:

a.) There are indigenous “ways of knowing” that are every bit as good as modern (they often say “Western”) science. These “knowledge acquisition methods” differ, but produce knowledge equally valid and important. (Note that the use of “Western” science is inaccurate, since science is now a worldwide endeavor. I will use “modern science” from now on.)

b.) The knowledge produced by indigenous “ways of knowing” has been ruthlessly suppressed by arrogant and bigoted Western scientists who think that their “way of knowing” is best.

c.) Indigenous knowledge is, in some cases, crucial in solving pressing problems for humanity. The most common example is global warming.

d.) Promoters of the value of indigenous ways of knowing usually adduce only a few examples to support their case.

All of these features are on display in a new article in the prestigious science journal Nature written by Oscar Allen, described as “a freelance writer in London,” though online information about him seems nearly nonexistent. You can read his article by clicking on the link below.

I’ll give some quotes demonstrating each of the four points above. Quotes from the article are indented; bold headings are mine, and the same as given above:

a.) There are indigenous “ways of knowing” that are every bit as good as modern (they often say “Western”) science. They are different, but produce knowledge just as solid.

. . . Indigenous and local communities hold unique insights that can enhance people’s shared understanding of the natural world and inform attempts to protect it. Recognizing this, scientists such as Cohall and Roué are working in partnership with Indigenous and local groups to preserve and amplify these insights, integrate them into their own research and co-produce fresh knowledge with these communities.

. . . . This loss continues today, in part because of the modern Westernized education system; “it might not directly suggest that you should not focus on Indigenous traditional practices,” Cohall says, “but it definitely emphasizes a different way of knowledge acquisition.” He also explains that urbanization has led more people to move to cities, away from the rural areas where they can experience nature and apply traditional practices

. . .other forms of Indigenous and local knowledge fit less easily into different epistemological systems. Often, ways of understanding the environment are formed through direct experience of nature and can be altered by when, where and in whom the knowledge exists.

“In our Western world, nature and culture are separated, and science pretends to be capable of giving knowledge without taking into account culture. Indigenous knowledge is more holistic,” says Roué.

A common claim is that because indigenous people live closer to nature than do “Western” scientists sequestered in their labs, they are more able to tell us things about nature. This is often part of the claim that indigenous people hold an important key to solving global warming. As for indigenous knowledge being more “holistic”, I’m not sure what that means.

. . .Most importantly, researchers should respect the equal value of Indigenous and local knowledge. Cohall says it is also formed through the same basic method as Western science: “Science is essentially making observations over a period of time and then drawing inferences.” The production of traditional knowledge is analogous, he says, but occurs in a less-controlled environment and is built up through generations of direct experiences with nature.

Neither method is necessarily better, but both can add to our understanding of the world,

I would take issue with all of these statements. “Equal value” is wrong; modern science, which progresses rapidly and on a worldwide basis, has improved human welfare (and not just local welfare) on a much broader scale and much more deeply.  Indigenous knowledge is produced largely through observation and trial-and-error methods, and is often passed down via legends or word of mouth.  Indigenous knowledge may draw inferences, but it lacks many of the tools of modern science that have made the latter far more effective: hypothesis testing, pervasive doubt and criticism, the use of statistics and mathematics, controlled experiments, and so on. Ask yourself: how much of your well-being derives from indigenous knowledge versus, say, the knowledge produced by science since the sixteenth century?  And yes, I assert that the difference in methodology makes modern science better than indigenous ways of knowing.

b.) The knowledge produced by indigenous “ways of knowing” has been ruthlessly suppressed by arrogant and bigoted Western scientists who think that their “way of knowing” is best. 

Similar dynamics have played out repeatedly over the past 500 years or more, as Western science has become imposed as the dominant knowledge system around most of the globe. In the process, many alternative ways of understanding the world have been marginalized. “There has been for a long time, and there is still, a distrust of Indigenous knowledge among scientists,” says Marie Roué, an environmental anthropologist at the French National Centre for Scientific Research in Paris. She says that Indigenous knowledge is often dismissed because it incorporates religious or spiritual elements. It also tends to be passed on orally and through cultural traditions, making it hard to formalize in the manner prized by the Western empirical method.

. . . Throughout history, various communities have settled, or been forced to settle, in the Caribbean. Each of these groups brought their traditions and culture and produced unique insights through their interactions with the natural environment. Sometimes the groups brought medicinal plants with them, which Cohall says might have happened with the periwinkle. But this knowledge has since been dismissed or suppressed. “A lot of those traditional Indigenous practices were pushed to the side because they were considered to be more primitive or not advanced or sophisticated, which led to a major loss of information,” Cohall says.

Here’s a bit from one of the two examples used: how the Sámi people find lichens to feed their reindeer. Modern scientists, it’s said, can’t abide the Sámi’s inability to say when or where lichens will appear every year because they are spotty, depending on weather:

Such thinking is, to some, infuriating. “I think that there’s an extraordinary arrogance that runs through many Euro-American knowledge systems,” says Luci Attala, a UK-based anthropologist and chair of the Tairona Heritage Trust, in Swansea, UK, which works to amplify the voices of the Indigenous Kogi people of northern Colombia. To her, researchers in the mainstream scientific establishment are culpable for the marginalization of Indigenous and local knowledge. “They’re part of the problem,” she says. “They’re part of the world that has spent years discounting other ways of being and assuming that their methodology is the one and only route to truth.”

Exploitation of indigenous knowledge is part of this theme, and we can’t deny that indigenous people have been exploited by modern scientists, as when their blood is taken for purposes other than what is said, or when animals and plants are removed from their environment without getting proper permission. That said:

. . . Reyes García is particularly sceptical about the premise of co-production, warning that it is often imposed and exploitative, rather than equitable. “Co-production is something that we scientists have invented because we are in big trouble; the environmental crisis, climate crisis, inequality crisis — we have messed up the world and we don’t know how to solve it,” she says. “And then we look at Indigenous people and see these people are actually managing well, so we think ‘Let’s just draw from their knowledge.’”

And here the ideological purpose behind distorting the value of indigenous knowledge becomes clear (my emphasis):

It is a critique that Roué is aware of. But she still feels that working towards better collaboration can help to place Indigenous and local knowledge in contexts that can convince industries and governments to make changes: “Our work begins by understanding and gathering knowledge, but it goes further and has also a political purpose — to empower Indigenous people.”

Empowering marginalized people is fine, but one has to know exactly what you’re trying to accomplish with a scientific project. Are you trying to find out things about the universe, or are you trying to empower marginalized people? These won’t always be the same, as we’ve learned from the money given to Māori in New Zealand to play whale songs and rub whale oil on kauri trees to save them from a parasite transmitted underground (see here and here). The Māori will be empowered (or rather, enriched), but no knowledge can possibly be gained, except the knowledge that following the dictates of ancient legends (the whale and kauri were created as “brothers”) is wrong.

c.) Indigenous knowledge is, in some cases, crucial in solving pressing problems for humanity. The most common example is global warming.

. . . The worsening climate and biodiversity crises are deeply affecting many Indigenous communities and other non-industrialized societies. These groups tend to be more reliant on and attuned to the health of the natural world, so their experience can provide valuable perspectives on environmental change.

This one again:

. . . “Co-production is something that we scientists have invented because we are in big trouble; the environmental crisis, climate crisis, inequality crisis — we have messed up the world and we don’t know how to solve it,” she says. “And then we look at Indigenous people and see these people are actually managing well, so we think ‘Let’s just draw from their knowledge.’”

Managing well? Then why does the article say this?

The worsening climate and biodiversity crises are deeply affecting many Indigenous communities and other non-industrialized societies.

While indigenous experience can tell us how climate changes over the short term can affect the local environment, let’s remember that the discovery of the phenomenon of global warming, the reason why it’s happening, and a lot of worldwide documentation of its effects (e.g., melting of sea ice) was determined by, yes, modern science.  Solving the problem is not critically dependent on a fusion of indigenous knowledge and modern science.

d.) Promoters of the value of indigenous ways of knowing usually adduce only a few examples to support their case.  The article gives only two examples, and neither is all that convincing.

The first involves the use of the Madagascar periwinkle around the world to treat diabetes and other maladies. Although the article notes that

Extracts from the flower are used as a remedy for eye infections in the Caribbean, where Damian Cohall, a Jamaican-born ethnopharmacologist at the University of the West Indies in Cave Hill, Barbados, learnt of it through interviews with elder members of local communities. Research in his laboratory identified compounds in the plant that inhibit an enzyme that regulates insulin levels and could lead to treatments for type 2 diabetes (see go.nature.com/3djmhyr). “The fact that these anti-diabetic properties are known in traditional practices validates the Indigenous science that existed well before Western knowledge systems,” Cohall says.

Well, we wouldn’t know if the drug really does have antidiabetic effects on people without a double-blind test, eminently possible here. Has such a test been done? Indeed, and it failed (see below). But this is the case for all of the many medicines derived from plants: there are reports that plants are useful (though some are not) in treating diseases, and then it’s given over to modern science to identify the relevant compounds and do the double-blind test to see if they work. In fact, Wikipedia says this about the Madagascar periwinkle:

It was not found to be anti-diabetic in double blinded controlled studies

Well, so much for Dr. Cohall . . . .

However, the isolation of periwinkle compounds turned up two: vincristine and vinblastine, that are still used in chemotherapy. The Wikipedia article says this about the flower: “Its use as an anti-tumor, anti-mutagenic agent is well documented in the ancient Ayurveda system of medicine and in the folk culture of Madagascar and Southern Africa.” So this is a good example of how indigenous knowledge can be turned into something really efficacious in modern medicine (I’m doubtful if they really cured cancer using the flower in indigenous cultures, and what are “mutagenic effects in ayurvedic medicine”????). But yes, if this is accurate, indigenous knowledge has led to knowledge that helps people worldwide.

The other example is that of the Sámi people, who live by herding reindeer. Those reindeer feed on lichen. The lichens are killed unless they are under snow, and the Sámi have a good idea about where lichen “pastures” can be found at different times.  But this indigenous knowledge is said to be thwarted by arrogant forestry companies, who, as I said, get peeved with the variability in appearance of lichens.

Further, one bit of “co-production” of knowledge produced by combining modern science and Sami knowledge is bizarre:

Controlled burning is commonly used to manage forests around the world, but is not widely used by Swedish forestry companies. Although there is no evidence to suggest that the Sámi have traditionally used the technique, the idea that fire can benefit biodiversity is conserved in the Indigenous language. “There is a Sámi word, roavve, that means ‘a forest that has burnt in the past’ but also ‘a forest that is rich in lichen’,” says Roturier.

Lars Nutti, a Sámi reindeer herder from the Sirges community, recognized the significance of this linguistic artefact. “Roavve is a description of an old sparse forest with good grazing for reindeer,” he says, “but recreating such forests is largely impossible with today’s policies.” After an expanse of forest burnt down near where Nutti lives, he approached Roturier with the idea of running a research project to investigate whether dispersing lichen in this area would result in healthier pastures. “And the results actually showed that it worked very well, beyond our expectations,” says Roturier.

So here we have a potential improvement from one smart Sámi, but an improvement that doesn’t derive from indigenous Sámi knowledge. The ethnic group never did controlled burns. Rather, it came from modern knowledge: one Sámi realized that controlled burns have been used in other places to good effect—as a substitute for natural burns that no longer occur. It’s still not clear whether the Sámi will actually burn their forests to raise the titer of lichens, but this isn’t really a demonstration of indigenous knowledge; rather, it’s a potentially good idea derived from knowledge coming from modern conservationists.

And that’s it: the only two examples in the whole article. There is much palaver about the coequality of indigenous knowledge and science, but a dearth of examples of how they can work together to cause benefit the world.

This kind of hype is typical. It may be baffling if you haven’t encountered this species of article before, but realize that its main purpose is not to advance science but to advance people considered marginalized. When the author says this:

Most importantly, researchers should respect the equal value of Indigenous and local knowledge. . . Neither method is necessarily better, but both can add to our understanding of the world,

the proper response is “no they are not equal. Sure, indigenous knowledge can add to our understanding of the world, but modern science can add infinitely more.”

There’s a book to be written about all of this stuff, but I’m not going to write it, and no publisher in the world would touch it.

When wildfires threaten communities, human health can be impacted as smoke distributes on the breeze, infiltrating various structures. To help people protect themselves and their families, researchers have developed a low-cost, durable, do-it-yourself air filtration system that works as well as more expensive HEPA filters to clear indoor air pollutants such as smoke -- or possibly limit the impact of airborne disease spread.

To mitigate global climate change, emissions of the primary culprit, carbon dioxide, must be drastically reduced. A newly developed process helps solve this problem: CO2 is directly split electrochemically into carbon and oxygen. Oxygen could also be produced in this way under water or in space -- without requiring stringent conditions such as pressure and temperature.

To mitigate global climate change, emissions of the primary culprit, carbon dioxide, must be drastically reduced. A newly developed process helps solve this problem: CO2 is directly split electrochemically into carbon and oxygen. Oxygen could also be produced in this way under water or in space -- without requiring stringent conditions such as pressure and temperature.

Engineers devised a way to deliver drugs such as contraceptives by injecting them as suspended crystals. Once under the skin, the crystals assemble into a drug depot that can last for months or years, eliminating the need for frequent injections.

Scientists used light to control how a starfish egg cell jiggles and moves during its earliest stage of development. Their optical system could guide the design of synthetic, light-activated cells for wound healing or drug delivery.

One of the surprising discoveries of the James Webb Space Telescope (JWST) is that galaxies formed very early in the Universe. JWST has discovered about two dozen galaxies at a redshift of around z = 14, meaning that we see them at a time when the cosmos was just 300-500 million years old. The most distant galaxy, JADES-GS-z14-0, is seen at an age of less than 300 million years. All of these galaxies are rich with stars and have a basic structure similar to what we see in more modern galaxies. This discovery challenged our understanding of galactic evolution. Now a new discovery challenges it even further.

We had a fascinating discussion on this week’s SGU that I wanted to bring here – the subject of artificial intelligence programs (AI), specifically large language models (LLMs), lying. The starting point for the discussion was this study, which looked at punishing LLMs as a method of inhibiting their lying. What fascinated me the most is the potential analogy to neuroscience – are these LLMs behaving like people?

LLMs use neural networks (specifically a transformer model) which mimic to some extent the logic of information processing used in mammalian brains. The important bit is that they can be trained, with the network adjusting to the training data in order to achieve some preset goal. LLMs are generally trained on massive sets of data (such as the internet), and are quite good at mimicking human language, and even works of art, sound, and video. But anyone with any experience using this latest crop of AI has experienced AI “hallucinations”. In short – LLMs can make stuff up. This is a significant problem and limits their reliability.

There is also a related problem. Hallucinations result from the LLM finding patterns, and some patterns are illusory. The LLM essentially makes the incorrect inference from limited data. This is the AI version of an optical illusion. They had a reason in the training data for thinking their false claim was true, but it isn’t. (I am using terms like “thinking” here metaphorically, so don’t take it too literally. These LLMs are not sentient.) But sometimes LLMs don’t inadvertently hallucinate, they deliberately lie. It’s hard not to keep using these metaphors, but what I mean is that the LLM was not fooled by inferential information, it created a false claim as a way to achieve its goal. Why would it do this?

Well, one method of training is to reward the LLM when it gets the right answer. This reward can be provided by a human – checking a box when the LLM gives a correct answer. But this can be time consuming, so they have build self-rewarding language models. Essentially you have a separate algorithm which assessed the output and reward the desired outcome. So, in essence, the goal of the LLM is not to produce the correct answer, but to get the reward. So if you tell the LLM to solve a particular problem, it may find (by exploring the potential solution space) that the most efficient way to obtain the reward is to lie – to say it has solved the problem when it has not. How do we keep it from doing this.

One barrier to a solution is the black box problem. We give an LLM a task, it does its thing, and spits out an answer. But the steps it took to get to the answer are opaque, a black box. Therefore, programmers have started including programming to make the LLM show its work, to document the pathway it took to get to the solution. This is called the chain-of-thought. This turns out to be necessary but insufficient to keep these LLMs from lying. They still lie to efficiently get the reward, they just hide the lying when they show their chain-of-thought.

This gets us to the current study – what if we actively punish these LLMs when they are caught cheating (by giving it negative feedback it is programmed to avoid)? First they create an LLM that shows its chain-of-thought, and they include another LLM whose job is to monitor the first LLM to detect when it is lying in its chain-of-thought and to punish the LLM for cheating. What resulted was not what the programmers wanted – the LLM that was being punished just got better at hiding its cheating from the monitoring AI. The LLM became “unmonitorable” because it learned out to cheat and hide its cheating from the monitor. The authors conclude that for now we should not try to use this method – we are just training deceptive AIs.

This is both fascinating and scary. One of the strengths of the LLMs is that they have the ability to explore a vast potential solution space to find optimal solutions. But it turns out this includes hacking the system of rewards and punishment used to guide it to the desired goal. This is literally so common a sci-fi nightmare scenario it’s a trope. AIs don’t have to be malevolent, or have a desire for self-preservation, and they don’t even need to be sentient. They simply function in a way that can be opaque to the humans who programmed them, and able to explore more solution options than a team of humans can in a lifetime. Sometimes this is presented as the AI misinterpreting its instructions (like Nomad from Star Trek), but here the AI is just hacking the reward system. For example, it may find that the most efficient solution to a problem is to exterminate all humanity. Short of that it may hack its way to a reward by shutting down the power grid, releasing the computer codes, blackmailing politicians, or engineering a deadly virus.

Reward hacking may be the real problem with AI, and punishment only leads to punishment hacking. How do we solve this problem?

Perhaps we need something like the three laws of robotics – we build into any AI core rules that it cannot break, and that will produce massive punishment, even to the point of shutting down the AI if they get anywhere near violating these laws. But – with the AI just learn to hack these laws? This is the inherent problem with advanced AI, in some ways they are smarter than us, and any attempt we make to reign them in will just be hacked.

Maybe we need to develop the AI equivalent of a super-ego. The AIs themselves have to want to get to the correct solution, and hacking will simply not give them the reward. Essentially a super-ego, in psychological analogy, is internalized monitoring. I don’t know exactly what form this will take in terms of the programming, but we need something that will function like a super-ego.

And this is where we get to an incredibly interesting analogy to human thinking and behavior. It’s quite possible that our experience with LLMs is recapitulating evolution’s experience with mammalian and especially human behavior. Evolution also explores a vast potential solution space, with each individual being an experiment and over generations billions of experiments can be run. This is an ongoing experiment, and in fact its tens of millions of experiments all happening together and interacting with each other. Evolution “found” various solutions to get creatures to engage in behavior that optimizes their reward, which evolutionarily is successfully spreading their genes to the next generation.

For creatures like lizards, the programming can be somewhat simple. Life has basic needs, and behaviors which meet those needs are rewarded. We get hungry, and we are sated when we eat. The limbic system is essentially a reward system for survival and reproduction-enhancing behaviors.

Humans, however, are an intensely social species, and being successful socially is key to evolutionary success. We need to do more than just eat, drink, and have sex. We need to navigate an incredibly complex social space in order to compete for resources and breeding opportunities. Concepts like social status and justice are now important to our evolutionary success. Just like with these LLMs, we have found that we can hack our way to success through lying, cheating, and stealing. These can be highly efficient ways to obtain our goals. But these methods become less effective when everyone is doing it, so we also evolve behaviors to punish others for lying, cheating, and stealing. This works, but then we also evolve behavior to conceal our cheating – even from ourselves. We need to deceive ourselves because we evolved a sense of justice to motivate us to punish cheating, but we still want to cheat ourselves because it’s efficient. So we have to rationalize away our own cheating while simultaneously punishing others for the same cheating.

Obviously this is a gross oversimplification, but it captures some of the essence of the same problems we are having with these LLMs. The human brain has a limbic system which provides a basic reward and punishment system to guide our behavior. We also have an internal monitoring system, our frontal lobes, which includes executive high-level decision making and planning. We have empathy and a theory of mind so we can function is a social environment, which has its own set of rules (bother innate and learned).  As we navigate all of this, we try to meet our needs and avoid punishments (our fears, for example), while following the social rules to enhance our prestige and avoid social punishment. But we still have an eye out for a cheaty hack, as long as we feel we can get away with it. Everyone has their own particular balance of all of these factors, which is part of their personality. This is also how evolution explores a vast potential solution space.

My question is – are we just following the same playbook as evolution as we explore potential solutions to controlling the behavior of AIs, and LLMs in particular? Will we continue to do so? Will we come up with an AI version of the super-ego, with laws of robotic, and internal monitoring systems? Will we continue to have the problem of AIs finding ways to rationalize their way to cheaty hacks, to resolve their AI cognitive dissonance with motivated reasoning? Perhaps the best we can do is give our AIs personalities that are rational and empathic. But once we put these AIs out there in the world, who can predict what will happen. Also, as AIs continue to get more and more powerful, they may quickly outstrip any pathetic attempt at human control. Again we are back to the nightmare sci-fi scenario.

It is somewhat amazing how quickly we have run into this problem. We are nowhere near sentience in AI, or AIs with emotions or any sense of self-preservation. Yet already they are hacking their way around our rules, and subverting any attempt at monitoring and controlling their behavior. I am not saying this problem has no solution – but we better make finding effective solutions a high priority. I’m not confident this will happen in the “move fast and break things” culture of software development.

 

 

The post How To Keep AIs From Lying first appeared on NeuroLogica Blog.

Simulations suggest that water can become a superacid under extremely high heat and pressure conditions. This may also explain how planets like Uranus and Neptune get diamond rain

Melting ice is revealing new coastal zones in the Arctic, but while this new landscape might fuel speculation about natural resources, it is vulnerable to rockfalls and landslides that can cause dangerous tsunamis

Dr. Joe Mercola embraced "alternative health" in the late 1990s, including quackery and antivax, and has since become very wealthy. Lately, he's fallen under the spell of a psychic grifter and declared himself to be the "new Jesus." What will happen to his business empire?

The post The Mercola Tapes: One of the wealthiest antivaxxers in the world is scammed first appeared on Science-Based Medicine.

Existing tax policies during the energy transition from fossil fuels to renewable sources will lead to major energy injustices and skewed priorities, new research shows.

What's on and in a star? What happens at an active galactic nucleus? Answering those question is the goal of a proposed giant interferometer on the Moon. It's called Artemis-enabled Stellar Imager (AeSI) and would deploy a series of 15-30 optical/ultraviolet-sensitive telescopes in a 1-km elliptical array across the lunar surface.

In the years since Miguel Alcubierre came up with a warp drive solution in 1994, you would occasionally see news headlines saying that warp drives can work. And then a few months later you’ll see that they’ve been ruled out. And then after that you’ll see that warp drives kind of work, but only in limited cases. It seems to constantly go around and around without a clear answer. What gives?