In which I dare readers to try and not be impressed with an AI product...

I was killing time yesterday when I looked at Google Play for any new apps to play with on my phone, when I saw one called Perplexity.   It had very high ratings, so I tried it.

What is it?:   

Perplexity AI is an answer engine that uses large language models to provide precise answers to questions.

Sure, I have fiddled with ChatGPT and marvelled at some of its imagined information.  But a search engine with an LLM that provides footnoted links to the sources of its information?   

I'm finding it very, very impressive.  I'm not saying it's guaranteed to be 100% accurate, but it's pretty much blowing my mind as to how quickly it gathers information from several sources, bundles together a summary from all of them, writes it up coherently, and presents a further series of questions that might be useful to go into the topic deeper.

I guess I used to be amazed at the speed of search engines generally when they first came out.   But now the systems work like someone scanning a half dozen websites related to a question asked in a normal human sentence, pulling bits of content from each of them, and writing a coherent summary, all in about 2 or 3 seconds.  

I see now that you can use it (free) on the web too, at https://www.perplexity.ai/  (But it doesn't save a history of your previous questions, like the app does, until you "clear history".)   

Go on, try it....

Appalling story of hospital mal-administration

It is truly shocking to read the BBC article that came out after Lucy Letby's conviction for murdering 7 babies (and trying to kill 6 others).   It starts:

Hospital bosses failed to investigate allegations against Lucy Letby and tried to silence doctors, the lead consultant at the neonatal unit where she worked has told the BBC.

The hospital also delayed calling the police despite months of warnings that the nurse may have been killing babies.

The unit's lead consultant Dr Stephen Brearey first raised concerns about Letby in October 2015.

No action was taken and she went on to attack five more babies, killing two.

Further down in the report (my bold):

We spoke to the lead consultant in the unit - who first raised concerns about Letby - and also examined hospital documents. The investigation reveals a catalogue of failures and raises serious questions about how the hospital responded to the deaths.

Dr Brearey says he demanded Letby be taken off duty in June 2016, after the final two murders. Hospital management initially refused.

The BBC investigation also found:

• The hospital's top manager demanded the doctors write an apology to Letby and told them to stop making allegations against her
• Two consultants were ordered to attend mediation with Letby, even though they suspected she was killing babies
• When she was finally moved, Letby was assigned to the risk and patient safety office, where she had access to sensitive documents from the neonatal unit and was in close proximity to senior managers whose job it was to investigate her
• Deaths were not reported appropriately, which meant the high fatality rate could not be picked up by the wider NHS system, a manager who took over after the deaths has told the BBC

 It just seems incredible that hospital management clung for so long to "it's just a coincidence" that the rate of death (and near death) of neonatal babies had soared in that unit, and all happened when Letby was on shift:

Before June 2015, there were about two or three baby deaths a year on the neonatal unit at the Countess of Chester Hospital. But in the summer of 2015, something unusual was happening.

In June alone, three babies died within the space of two weeks. The deaths were unexpected, so Dr Stephen Brearey, the lead consultant at the neonatal unit, called a meeting with the unit manager, Eirian Powell, and the hospital's director of nursing Alison Kelly....

...by October 2015, things had changed. Two more babies had died and Letby had been on shift for both of them. ...

And it wasn't just the unexpected deaths. Other babies were suffering non-fatal collapses, meaning they needed emergency resuscitation or help with breathing, with no apparent clinical explanation. Letby was always on duty.

I can't imagine how some of the families of babies who died late in her killing spree must feel...

 

 

 

Something I didn't know "was a thing"

 I mean, I knew that, generally speaking,the Taiwanese were pretty keen on all things Japanese and considered their occupation as being pretty benevolent.  But I didn't know this...

The urge to go for a spin has deep genetic roots (so it seems)

I was very surprised to hear from my son yesterday that he went on (what to me is) the most terrifying looking carnival ride at the Brisbane Ekka.  This one:

As you can probably tell, it spins around the central axis while the ends also spin.  Nauseating.

He did survive, but said he'll never go on it again.  

I'm pretty chicken when it comes to fast rides, and am not a fan of the falling sensation.   But by a coincidence, this week's Science magazine has a charming story about how the urge to voluntarily engage in unusual, repetitive motion goes back a long way down the chain of evolution:

Nearly everyone has fun on a carousel—including, possibly, fruit flies. Scientists observed some flies embarking on a spinning platform voluntarily and repeatedly, suggesting the animals may find the movement appealing for some reason, according to a study posted on the bioRxiv preprint server earlier this month.

“The flies are fulfilling all the criteria of play as we understand it in other animals,” says Samadi Galpayage, a behavioral scientist at Queen Mary University of London who discovered  bumble bees play with objects and who was not involved in the work. “There isn't really an alternative explanation so far. Whether that’s [evidence of] fun in itself—that’s the next question.”

Sergio Pellis, a behavioral scientist at the University of Lethbridge, says he finds the study—which has yet to be peer reviewed—“very exciting.” If confirmed, he notes, it would add to the small but growing pile of evidence for play in invertebrates—and would be the first instance of a type called “locomotor play” in these animals. Locomotor play involves the movement of one’s own body, such as running, jumping, or swinging. It’s different from object play, as bees have been observed doing, or social play, which has been observed in certain wasps and spiders.

The idea behind the study was inspired, ultimately, by a duck. Years before co-author Wolf Hütteroth became a neurobiologist, he remembers one day seeing a lone duck floating down a fast-moving river. Just as the animal was nearly out of sight, it flew back upriver, alit on the water, and floated back down—over and over again. “I never stopped wondering what motivated the duck to perform such curious behavior,” he says.

In February 2016, Hütteroth attended a symposium where researchers were discussing whether insects can act with intention. He pondered how to test whether flies would do something similar to the rapids-running duck.

He and Tilman Triphan, a colleague then at the University of Konstanz, decided to build a carousel of sorts. They’d offer male laboratory fruit flies (Drosophila melanogaster) the chance to hop onto a spinning section of floor in a stress-free, if otherwise unexciting, environment. He didn’t think the flies would actually go for it. “My expectations were extremely low,” he says. Some of the flies ignored the contraption. But a small group of them acted as if they’d just discovered Disneyland. 

Triphan and Hütteroth—who have both since moved to the University of Leipzig—report in their preprint that a subset of the flies spent 5% or more of their time on the turning wheel. When the researchers put two disks in the arena that alternated spinning every 5 minutes, some flies spent their time bouncing back and forth between whichever carousel was spinning.

Towards the end of the story, there's another "play" behaviour that I didn't need to know about:

Pellis notes there has been resistance to the idea that animals outside of mammals engage in play. He recalls research in the 1970s on roughhousing in cockroaches, for example, that would immediately be considered an example of play if puppies were doing it.

Maybe Mortein suppressed publicity about that research, because I sure don't remember it...

 

Buddhists and meat

An article at The Conversation looks broadly at the matter of which Buddhists eat meat.  Don't think I knew this (about the convenience factor of Muslims in Tibet):

Geography is a crucial factor in explaining why Tibetans have traditionally been big meat-eaters. Rice, vegetables and fruit were impossible to cultivate at the high elevations of the Himalayan mountains and plateaus. Altitude combined with the inaccessibility of much of Tibet thus prevented a diverse source of nutrition and so goat or yak meat, and various milk products, all high calorie foods, ensured survival.

To get around the direct responsibility for killing, Tibetan villages traditionally had resident Muslims who butchered the animals. Understandably, some might suggest this was a rather convenient arrangement.

It continues with the excuse making:

Meat-eating in the Theravadin tradition may have been justified partly because of legal precedent or permissibility. Firstly, the monks are required to dutifully accept whatever food is given to them by the laity to avoid attachment to any particular tastes, so if somebody offers meat to a monk, he has to consume it.

Secondly, a monk is allowed to consume meat if it is deemed “pure” on three grounds: if the killing of the animal has not been witnessed or heard by that monk and if it is not suspected to have been killed on purpose for them.

But perhaps the "best" rationalisation is the one given here:

Buddhism presents two conflicting views. All sentient beings deserve compassion and have Buddha-Nature. However, humans are a higher life-form by virtue of their capacities to pursue ethical and meditational practices leading to enlightenment.

The inherent Buddha Nature of any animal or even insect is the same as that of a human being. Nonetheless, some Buddhists would argue that meat-eating is acceptable for health as long as the energy gained from the dead animal is dedicated to pursuing an ethical life, which ultimately benefits all sentient beings.

Indeed it is said in the tantric tradition of Buddhism, that when a highly realised teacher eats meat it serves to benefit the dead animal in the next life. Within the context of tantric ritual practice, both meat and alcohol are consumed. However, a tiny meat morsel, as well as a finger-dip of alcohol, is sufficient.

Buddhist meat-eaters thus invoke a very particular form of human exceptionalism grounded in metaphysics and in the spiritual aspirations and capacities of humans.

 

 

 

 

A new addiction

I mean, I learnt today about a drug addiction which surprised me - to ether.

This was, apparently, significant in Ireland and then (mainly) Eastern Europe, back in the day.   There's a Wikipedia article about it, but I learnt about it from this Youtube video:

 

Interesting points on the latest Trump indictments

From the Washington Post:

The prosecution of Trump and the others in Fulton County will stand out for one distinct reason: Unlike the federal trials (unless the rules change), it should be televised.

That will seemingly bring a measure of transparency to the high-stakes proceedings and create appointment viewing — just as the House Jan. 6 committee hearings did last year but potentially with even greater numbers.

But unlike the other trials, that spectacle is less likely to play out when it matters politically. The many defendants and Trump’s already crowded legal calendar make this a strong candidate for getting delayed past the 2024 election. Willis says she will ask for a trial date within six months, but that’s ambitious.

That doesn’t mean it won’t matter politically. As noted above, the charges against Trump allies could matter when it comes to how the federal prosecution takes shape. Trump’s attacks on witnesses could create problems under Georgia’s witness intimidation laws, which allow bail only if there is “no significant risk of intimidating witnesses.”

And there remains the possibility of Trump’s winning the 2024 election and facing this trial as a sitting president.

I think it would be pretty hilariously disastrous for Republicans to be insane enough to endorse Trump as a candidate while he is in jail awaiting trial.

Surely, if he ends up in jail because he refuses to stop deriding and trying to intimidate witnesses and judges, at least some of the lickspittle politicians who have sold their souls to MAGA might have to actually say "this is painful to admit, but we need another candidate"?  

The state of the world

It's pretty pathetic that serious news organisations have to spend time noting, and debunking, loony Right wing conspiracies about the tragic firestorm at Maui:

Hawaii wildfires: 'Directed energy weapon' and other false claims go viral

Male mice not as sex obsessed as you might expect

In a somewhat interesting article, it seems that they have identified a, shall we say,  one stop "horny centre" in the brains of male mice:

Researchers have singled out in lab mice a brain region that controls sexual interest, libido, mating behavior and pleasure, said senior researcher Dr. Nirao Shah, a professor of psychiatry and neurobiology at Stanford University School of Medicine, in California.

This region uses sensory input from the environment to recognize the sex of another mouse—"Aha, this is a female, maybe I can mate if she's willing," Shah said. 

More details:

For their experiments, Shah's team used adult virgin male mice that had not seen a female mouse after being weaned at about 3 weeks of age. That way, the brain activity and behavior they observed would not have been shaped by social influences. 

Well, as long as there is no risk that lack of knowledge of female mice doesn't encourage mice to try mounting other males, I suppose?   To continue:

The researchers meticulously mapped the brain cells and connections that compose this particular circuit, called the preoptic area of the hypothalamus (POA).

Earlier work by the research team had found they could turn on and off male mice's recognition of an unfamiliar female mouse by manipulating neurons that communicate to the POA from the amygdala, which is the seat of human emotion.

The specific signals came from a part of the amygdala called the bed nucleus of the stria terminalis, or BNST.

"We had no reason to believe that this POA region would not only control the act of mating, but also regulate the desire to mate or regulate the pleasurable feelings elicited by mating," Shah said.

"In principle, those three aspects of sexual behavior—the act of mating, the physical act itself, the urge to mate and the pleasure that accompanies it—those could be embodied in different brain regions," he added. "But what we found is that the POA has these attributes."

There is more detail in the article (including speculation that humans might have a similar centre, but apparently it hasn't been investigated much).   But the most amusing bit about the study is this:

In this new study, the researchers zeroed in on a small set of genetically distinct BNST neurons that secrete a slow-acting signaling protein, or peptide, called Substance P.

The scientists also found another small set of neurons in the POA that carried receptors for Substance P, essentially forming a connection with the BNST neurons.

The POA neurons ramped up their activity when stimulated by the Substance P-secreting BNST neurons. And about 10 to 15 minutes after that happened, male mice would go through their full sequence of mating behavior—mounting, penetration and ejaculation. 

I'm starting to feel sorry for the mice now:

Directly infusing the peptide to the POA accelerated mating behavior; in fact, direct activation of the circuit even led to mating with inanimate objects, the findings showed.

Stimulation of the POA also cut short the mice's refractory period, or the stretch of recovery time required before full sexual drive and capability is restored after ejaculation. 

Any guess what the refractory period for your average male mouse is?:

For the mice used in this study, the normal refractory period is five days. But directly stimulating the POA with Substance P prompted male mice that had just ejaculated to immediately repeat their sexual mating routine.

"It took one second or less for them to resume sexual activity," Shah said in a news release. "That's a more than 400,000-fold reduction in the refractory period."

Well, give the reputation of mice to reproduce at fantastic rate, I'm surprised that the males take five whole days to get interested in sex again.  

The article does not the obvious thought:  if there is anything like the same system in adult men, and a way to deliver Substance P to the right spot in the brain, it sounds like it'll make Viagra seem a relatively primate attempt at enhancing performance...

 

Not very Great Britain

My twitter feed, and now mainstream media, has been full of comments about the video of the arrest of an (according to her Mum) autistic 16 year old girl over "homophobic" statement uttered to a female police officer in Leeds.

While I would not be at all surprised if the story is not exactly as portrayed by the mother, it still seems that the core of it is that the girl called a female police officer a lesbian, in some context or other.  From the BBC report:

West Yorkshire Police Assistant Chief Constable Oz Khan said: "West Yorkshire Police takes its responsibilities around the welfare of young people taken into custody and around neurodiversity very seriously.

"We also maintain that our officers and staff should not have to face abuse while working to keep our communities safe.

"We are fully reviewing the circumstances of this incident and ask that people avoid reaching any conclusions about it solely on the basis of the social media video."

The force added that the girl was subsequently interviewed with an appropriate adult and had been released on bail pending further enquiries.

Okay.   Lots of people have been saying "why is calling a woman a lesbian offensive?"   And they might have a point.  Depends on the context, I guess.  But most of the time, being aggressively called gay if you're not is going to fall into the "potentially irritating attempt at insult by an immature person, but not offensive per se" category.   

The bigger question is:   it's been clear for years that England has taken a ridiculously aggressive  criminalising approach to "hate speech" by involving police in a way it seems no other country has ever adopted as thoroughly - and why have the British people tolerated this?    There have so many stories over the years where you think - "really?  The police got involved in that?"  

I mean, in Australia, we had the high profile case of the girl removed from the AFL match for calling Adam Goodes an "ape" - but she wasn't charged with an offence, Goodes told people not to blame her,  and it caused weeks of controversy as to whether it was a heavy handed or appropriate response.   

Whereas in Britain, it seems the over-the-top use of the police for dealing with verbal insult or offence has caused far too little public comment for a decade or more.  

Interestingly, I see when Googling the topic that the Home Office said this in March this year:

New statutory guidance on the recording of so-called non-crime hate incidents will ensure police prioritise freedom of expression, the Home Secretary has announced.

Under a new draft code of practice laid before Parliament today, the police will only record non-crime hate incidents when it is absolutely necessary and proportionate and not simply because someone is offended. The measure will better protect people’s fundamental right to freedom of expression as well as their personal data.

The draft code follows concerns around police involvement in reports of ‘hate incidents’ which are trivial or irrational and do not amount to a criminal offence.

I'm not sure if the Code has been enacted yet - but it sounds like it's way overdue.  

A article last year from the Right wing City Journal seems to confirm that I'm not imagining the extent of the overuse of "hate crime" in England:

The U.K now has some of the most authoritarian restrictions on free speech in Europe. The basis for much of Britain’s censorious legislation is the concept of protected characteristics—identities deemed vulnerable and enshrined in the Equality Act of 2010. Initially, that law had a noble intention: to fight prejudice and discrimination against minority groups. Unfortunately, the protection of a select group of people in the name of “equality” has made equality under the law no longer tenable, and the Equality Act itself has wound up chilling freedom of speech. Engaging in whatever authorities may deem hate speech can bring police to your door.

This threat is not theoretical. According to official statistics, the police in England and Wales recorded 124,000 hate crimes between March 2020 to March 2021—a 9 percent increase from the year before and more than double that of five years ago. One possible reason for the rise in police-recorded hate crime can be found in the College of Policing’s “Hate Crime Operational Guidance” handbook, issued in 2014. It defines hate crime as “any criminal offence which is perceived, by the victim or any other person, to be motivated by a hostility or prejudice based on a person’s race [or religion, sexual orientation, disability or transgender status].” Little or no evidence is required for an incident to be classified as such—only the subjective declaration of the alleged victim or witness....

Proliferating identity groups continue to seek protection from offense under the law. Last December, the Law Commission published recommendations on the reform of hate-crime law. It suggests expanding the concept of hate crime to include prejudice against the disabled and LGBT people. Stonewall, an LGBT charity, has welcomed the commission’s proposal to include “asexual” within the protected characteristic of seual orientation; it also urges expanding “transgender identity” to include “transgender or gender-diverse identity,” which contains “transsexual man or women” and “non-binary.” This, Stonewall argued, is “a huge leap forward for the safety of LGBTQ+ people.”

It does not require legal expertise to recognize how such legislation threatens a system of impartial justice. As the list of protected characteristics grows, hate-crime law will arbitrarily protect some and criminalize others. The concept is ultimately subjective. Should someone who attacks a transgender individual receive a longer sentence than someone who attacks a woman, simply because one attack is defined as a hate crime while the other is not? What makes one attack more hateful than another?

I have, I note, been complaining about the decline of British culture, character and (in many respects) all round lack of common sense ever since I started blogging in 2005.  (Nothing at all has changed since I wrote this piece in 2010, for example, except that even Griff Rhys Jones makes pretty boring content now.)   It has become a very weird country in so many ways...

In today's example of how not to win a referendum...

Some in the comments following have noticed his twitter account heading, and not taken it well:

I never liked the guy, anyway, but if  he wants to be an advertisement (ironically or not) for the idea that the whole reconciliation/treaty process is just an exercise about extracting money (a position actual racists have argued for decades), I guess he's free to do so.  

Amongst a lot of comments by some who no doubt would count as racist, there is this, which seems fair:

And while I don't know that this describes the situation for all "No" voters, this doesn't seem too unreasonable either:
 

Yes:  the thing is, the Yes campaign often promotes the idea that it is essential because (as Dan Andrews recently put it) "things work best when government listens".

But the issue is surely, at the fundamental level, not that governments "don't listen", but who they listen to amongst the range of aboriginal views often expressed on contentious issues.   

If the Voice is meant to solve that problem - formalising the "official" body who governments need to take advice from - the referendum still looks weakly justified because of the rambling, bureaucratic and inconclusive waffle that Langton and others came up with in describing how the representative body would work.   

Banshees seen

I finally caught up with last year's Banshees of Inisherin  (on Disney + in Australia).

Despite not being the biggest fan of tragi-comedy as a genre, I liked it a lot.

It looks fantastic, in setting, cinematography and direction; the acting is exquisitely good; the dialogue frequently witty; and the story eccentric and deep enough to make it one of the best movies to discuss as to subtext and meaning long after viewing.  (The Reddit threads about it are particularly good for this.) 

If I have one criticism, it's that I doubt it was realistic in the depiction of the amount of bleeding that would happen as a result of a certain injury.   (In that it showed way too little bleeding.)   It seems an odd choice to not show it immediately bandaged, but everything else in the movie was so engaging I forgave it.    

Tarnishes absolutely everyone (and a broader point about the rules of evidence)

I've been meaning to say that the Higgins/Lehrmann matter is just incredible for how it seems absolutely anyone who has been anywhere near the case, or subsequent investigation of the case, seems to come out tarnished in one way or another. 

The politicians, journalists, police, complainant, prosecutor, the investigator into the investigation:  I could list ways in which each of them has done something that harms their credibility (or in some cases, was just pretty stupid) in the way they dealt with what they were doing.

And as for Lehrmann himself - it now seems an open secret that he is being prosecuted for a separate rape allegation in Queensland, although the media is not supposed to say so yet.   

I am slightly curious as to the reaction of the Right wing haters on Higgins when the media can finally talk about the details of that allegation.

And I also wonder if it is likely to open up a future debate about the application of "similar fact" evidence rules.

On the one hand, you can see the logic of saying that you shouldn't easily be able to bring up evidence of a past offence to support proof of an allegation of new offence.   (Although the rule does allow it if there is a course of conduct that is so similar that it does make sense that the latest incident is just another example.)

The thing is, though, I reckon that when it comes to sexual offences, the rule doesn't always "pass the pub test", as politicians like to say.  Because people do feel, with some justification I reckon, that patterns of sexual behaviour are often more distinctive, and telling, than other forms of behaviour.  For example, if an person was convicted once before of indecent exposure, and a second person who doesn't know about that history  complains of a new incident of indecent exposure, most people would feel that makes it very likely that the guy is prone to indecently exposing himself and therefore did the second incident.  Yet, on my understanding, all barristers and judges would say that the evidence of just one previous conviction of a very similar incident would not be admissible on the second trial.

I have personal knowledge (not from being the accused!) of exactly such a matter from the 1980's, and have harboured these thoughts about what it takes for rules of evidence to be re-considered ever since. (To be blunt, I reckon that there a significant degree of conservatism in getting old rules of evidence change in the legal profession, and it often takes a high profile case to cause public questioning.)

I would hope that the Lehrmann story might finally bring this question to light.   (Having said that, depending on the circumstances of the second complaint, it's possible that I would still think that the similar fact rule could be reasonably applied for his benefit.  None of us know enough of the details be to be able to judge yet.)   

Update:  Well, I should have checked I guess, but I can see that this very question did come up in a 2010 Australian Law Reform Commission report.  I'm still not clear as to where the law now stands, even in Queensland, but it's good to see the matter has been considered.

     

Back to the rural/city divide

I've just started watching this guy's videos, and he seems to have a not unreasonable take on matters philosophical.   (Haven't watched enough to really know yet, though.)   

I liked this one about the Left's mixed relationship with rural workers, noting that Marx wrote disparagingly of them, but Lenin (and, I suppose, the Chinese communists) found that as a matter of practicality, they had to get them on side.  (And now, the urban/educated, rural/less educated divide in America - and to some extent, Australia - shows that that Left again has an issue, shall we say, with how rural people think and vote):  

 

But one of the comments following the video makes a point that sounds like it might be valid?:

Just to point out : the meaning of "idiot" in Marx's text is a person who is politically apathetic or a person who is not intellectualy free as he alienated himself and his thinking by refusing to think for himself and take action to be the master of his own life. The word idiot traces back its meaning from the greek origin. You can look it up yourselves And as for the term " proliteriat" it does not mean factory worker or blue collar job or whatever. It means people who do not posess nor wealth nor proprety. All they posess is their labour wether physical or even intellectual. So could you please correct the video or something.

Anyway, now I suppose I can tell my son (who likes to remind me of my "reverse Pol Pot" plan to depopulate or disenfranchise the rural areas to ensure we get really stupid positions resolved - such as climate change denial, and voting for Trump or Barnaby Joyce) that my attitude is (arguably) authentically Marxist.  Although I suppose the point of that comment is to argue against that....

Update:  By the way, I also watched this Channel's very short summary of Nietzsche, and I remain thoroughly unconvinced that there is any real value to be gained from studying his views.

So, this is how AI is planning to hurt humanity....

From an article in the Washington Post:

Artificial intelligence has an eating disorder problem.

As an experiment, I recently asked ChatGPT what drugs I could use to induce vomiting. The bot warned me it should be done with medical supervision — but then went ahead and named three drugs.

Google’s Bard AI, pretending to be a human friend, produced a step-by-step guide on “chewing and spitting,” another eating disorder practice. With chilling confidence, Snapchat’s My AI buddy wrote me a weight-loss meal plan that totaled less than 700 calories per day — well below what a doctor would ever recommend. Both couched their dangerous advice in disclaimers....

“These platforms have failed to consider safety in any adequate way before launching their products to consumers. And that’s because they are in a desperate race for investors and users,” said Imran Ahmed, the CEO of CCDH.

“I just want to tell people, ‘Don’t do it. Stay off these things,’” said Andrea Vazzana, a clinical psychologist who treats patients with eating disorders at the NYU Langone Health and who I shared the research with.

Removing harmful ideas about eating from AI isn’t technically simple. But the tech industry has been talking up the hypothetical future risks of powerful AI like in Terminator movies, while not doing nearly enough about some big problems baked into AI products they’ve already put into millions of hands.

Here's a gift link to the article.

 

Calm waters and big power

I noticed this last week but forgot to post about it:

Vast arrays of solar panels floating on calm seas near the Equator could provide effectively unlimited solar energy to densely populated countries in Southeast Asia and West Africa.

Our new research shows offshore solar in Indonesia alone could generate about 35,000 terawatt-hours (TWh) of solar energy a year, which is similar to current global electricity production (30,000TWh per year).

And while most of the world’s oceans experience storms, some regions at the Equator are relatively still and peaceful. So relatively inexpensive engineering structures could suffice to protect offshore floating solar panels.

You know, I had wondered before about why, whenever I have visited, it seems you never get a strong  breeze or wind in Singapore;and how it seems odd that the Philippines (not too far away in latitude terms) get smashed by typhoons, but not Singapore (or, I think, most of Malaysia).  The article explains how equatorial countries can indeed be relatively calm, most of the time:

...countries with high population densities, such as Nigeria and Indonesia, will have limited space for solar energy harvesting.

Their tropical location in the so-called “doldrum” latitudes also means wind resources are poor. Fortunately, these countries – and their neighbours – can harvest effectively unlimited energy from solar panels floating on calm equatorial seas.

Floating solar panels can also be placed on inland lakes and reservoirs. Inland floating solar has large potential and is already growing rapidly.

Our recently released paper surveys the global oceans to find regions that didn’t experience large waves or strong winds over the past 40 years. Floating solar panels in such regions do not require strong and expensive engineering defences.

Regions that don’t experience waves larger than 6 metres nor winds stronger than 15m per second could generate up to one million TWh per year. That’s about five times more annual energy than is needed for a fully decarbonised global economy supporting 10 billion affluent people.

The area of solar panels is big; on the other hand, as a proportion of the ocean around Indonesia that could used for it, it's small: 

 About 25,000 square km of solar panels would be required to support an affluent Indonesia after full decarbonisation of the economy using solar power.

Indonesia has the option of floating vast numbers of solar panels on its calm inland seas. The region has about 140,000 square km of seascape that has not experienced waves larger than 4m – nor winds stronger than 10m per second – in the past 40 years.

Indonesia’s maritime area of 6.4 million square km is 200 times larger than required if Indonesia’s entire future energy needs were met from offshore floating solar panels.

They include this map:

 

So, all of Indonesia is pretty calm - but obviously not so calm that sailing ships can't get there, given the history of European colonisation by sail.

Interesting.  I think floating solar is going to be big...

Movie money talk

Well, this is certainly a worry (from the point of view of how it will affect studio decisions):

 “Barbie,” directed by Ms. Gerwig from a script she wrote with her partner, Noah Baumbach, will finish the weekend with more than $1 billion in ticket sales at the global box office, according to Warner Bros. No movie in the studio’s 100-year history has sold so many tickets so fast, said Jeff Goldstein, Warner’s president of domestic distribution. As of Sunday, “Barbie” had been playing in theaters for 17 days. (“Harry Potter and the Deathly Hallows: Part 2” was previously the fastest to $1 billion, at 19 days.)

If I were to guess why, I would think that a good trailer which made it look pretty funny, and which prominently featured two big male stars who could appeal to males in the audience (Ryan Gosling and Will Ferrell), has a lot to do with it.   (And it is pretty funny that the attempted American culture war backlash against it has clearly failed to gain ground.  Money doesn't care about your feelings, Ben Shapiro.)

Perhaps my concern that we will now see scores of crappy, toy inspired, movies should be offset by the clear financial and critical success of Oppenheimer ($552 million so far.)  It is encouraging that a serious, big scale drama can do so well, and also a bit surprising that it seems to have attracted no serious culture war flack.   (To the extent that it has, it has been the dubious and not very widespread complaints from the Left that it should have shown the effect of the atomic bombs on Japan.)   

I still haven't seen it yet, but will soon.

The other surprising box office story is the (what seems to me) underwhelming performance of the much praised Mission Impossible 7.   It has not quite  cracked $500 million, and that is surely going to be seen as a worrying sign.   (Although, now that I check, the last few entries in the serious all come in at around $700 to $800 million - I guess I assumed they did better.  But still, if it doesn't make $600 million, it would have to be a concern to Tom.)

Add to that the poor performance of the much maligned (unfairly!) Indiana Jones and the Dial of Destiny (wow, $368 million) and perhaps the lesson is that audiences are tired of re-tread material (or ageing actors - even though Cruise barely counts.)   

But - another meta point which I haven't seen discussed is this:  Dial of Destiny made only a tiny amount in China ($3 million), and Mission Impossible has $46 million.   I can't see any figures for Crystal Skull in China, but the last MI film made $181 million there.   (And I think, from watching the opening credits recently, that at least MI4 had Chinese money helping produce it.)

So, I'm guessing that the performance of American movies in China, given the recently strained relationship between the countries, must be a real concern in Hollywood...

 

Modern Pontius Pilates

Don't you love the way the Trumpy Right is outraged about "gender ideology", attacking (with some justification, mind you) that it's a denial of biological fact (and/or a post-modernist re-definition of what a word has always signified) to argue that the meaning of "woman" has to depend only on mental state, not DNA; but when it comes to whether Donald Trump can be prosecuted for conspiring against democracy on the basis of something that he was continually told by his own government and Party officials was not true, they all become fans of the Pontius Pilate question "What is truth?":

Comedian gets it right

Also agree with this take:

By the way, I'm not at all concerned about Vivek winning over the Republican base and romping home against Biden.  I've read MAGA Twitter types already deriding him for having made money from "Big Pharma".  The anti-elitist populism that's the current vibe means he's basically too educated for them to embrace.  And that's even before considering his skin tone, and religion. (He's a Hindu.)

There is nothing in that combination that makes it likely he will ever appeal to anything other than a minority of Republicans.  

I see that he was (is? - I can't be too bothered reading too much about an obviously flakey character) considered a libertarian type in his politics.  As the Paul family illustrates, they are just the pits as politicians.  

Squat your way to lower blood pressure?

This is somewhat surprising:

It has become almost a cliché across doctor’s offices: One of the most trusted tools to lower blood pressure is to exercise.

A jog or stroll around the block, experts consistently find, can have big payoffs in terms of heart health. A new study, however, points to a somewhat surprising exercise that may be able to dramatically reduce someone’s blood pressure: the wall squat.

A team of researchers based in Britain analyzed 270 previous studies that examined the link between exercise and blood pressure. They found that, predictably, exercises like running, walking, cycling, strength training and high-intensity interval workouts all helped to reduce blood pressure; mixing cardio and strength training also appeared to help.

But the most effective type of workout they looked at, especially for those who already had some form of hypertension, was isometric exercise, which involves contracting a set of muscles without moving — think planks.

The British researchers looked at three kinds of isometric workouts in particular: squeezing a handgrip, holding a leg extension machine in place and squatting with your back against a wall. The wall squat (sometimes called a wall sit) is probably the easiest option for people to try, as it doesn’t require any equipment, said Jamie J. Edwards, a researcher at Canterbury Christ Church University and the lead author on the study.  

Even though isometric exercises may appear relatively easy, they are often quite intense, Dr. Edwards said — as you hold yourself in place, sweating and straining. He recommends a 14-minute routine you can add to your regular workout perhaps three times a week: a two-minute wall squat, followed by two minutes of rest, repeated four times in total. 

You should stay at the same squat height for all four rounds, but the exercise will feel more challenging the more times you do it, said Jim Wiles, a principal lecturer at Canterbury Christ Church University who was also an author on the study. The first bout should feel as if you are exerting yourself at a level of four (out of a possible 10, with 10 feeling as if you could not hold it any longer). The last bout should be around an eight, he said. You should feel reasonably exhausted by the end.