It didn't matter what the pompous ponce, Our Homeric Henry, was pontificating about today, there was no way that the pond could reduce him to a few sterile screen caps.
He had to live, to breathe, to be in a form where herpetology students might copy and paste, might suck in the fragrance of the written word.
The pond had a momentary pause for doubt - especially when the pond looked at the topic at hand.
Not another rant about AI.
But the pond plucked up courage, looked at the bright side, wondered if a reference to Odysseus might be the reward, knew deep in heart that some ancient philosopher held the key ...
AI no more dangerous than human stupidity
8 min
The caption for a mystifying snap. Patience, all will become clear: Barcelona in Spain is near the bottom of the list of cities with more than five million inhabitants
The clarion call to forget the paywall: Make us preferred on Google
Now some might wonder at the headline, what with human stupidity incredibly dangerous. After all, human stupidity has led to a couple of world wars, a means to end life on earth by way of nukes, and an even more splendid way to end life on earth by way of global warming (aided and abetted by denialists of the lizard Oz kind).
Ergo, AI must be incredibly dangerous.
But the pond must not pre-empt Our Henry's philosophically based conclusions.
Our Henry started strong, proving he was no top of events, by referencing rogue Anthropic refugee Evan Hubinger, Anthropic Researcher Warns There’s ‘>10% Chance’ AI Could ‘Kill All Humans’ By Next Decade
Devoted followers of Our Homeric Henry will be relieved to discover - spoiler alert - that this was just a feint, a distraction, and that the answer lies in Kant.
But first there's a journey and reassurance to be delivered to wilting hive mind brains:
Worse, it distracts from both the nature of the problem and the most sensible responses.
And by pandering to the sloppy apocalyptic thinking that permeates the climate extinction movement, it has unleashed the mother of all scare campaigns, culminating in Australia’s signing of a “feel-good” declaration urging international regulation of AI.
At the heart of that campaign sits a curiously familiar number: 10 per cent, recurring with extraordinary constancy throughout the doomsday rhetoric of the age.
That constancy is readily explained: 10 per cent is the Goldilocks of apocalyptic probability estimates – just small enough to be plausible, just large enough to be absolutely terrifying.
Goldilocks? Steady, faint of heart. Our Henry is likely referencing the Robert Southey version from the nineteenth century, a literary work full of depth and meaning ...
Our Henry cast porridge on troubled waters ...
Magnified by the far greater human costs, those figures would justify incurring vast sacrifices to avert catastrophe, as the doomsayers seem to intend. But with the dominant AI platforms having an interest in hindering new competition, and the Chinese regime in stifling AI’s development in the West, calls for a “slowdown” warrant the strictest scrutiny. And even the slightest look highlights the claims’ fragility.
There are, for example, 83 cities with five million or more inhabitants. Yet together they account for less than 20 per cent of the world’s population. An extinction event, or anything approaching it, would presumably have to devastate them all, or at least the vast majority.
Assume, however, that doing so requires 12 disastrous outages in each city, from water systems collapsing to nuclear power plants exploding.
The reptiles flung in an alarming image ... The crippled Fukushima Dai-ichi nuclear power station after the damage caused by an earhtquake (sic, so and thus, thanks AI editing) and tsunami in 2011
Our Henry was more than up to the task, and provided a statistical savaging ...
Even a hundredfold increase in the number of ways in which disaster could unfold scarcely changes that conclusion. It would take approximately 10 to the power of 998 independent pathways, rather than a hundred, before the probability ceased to be effectively zero. Indeed, with only 100 pathways, the probability of each condition being triggered would need to rise from 10 per cent to roughly 97 per cent just to nudge the dial towards the lowest single digits and raise the actuarial cost to $1.
Nor do the problems end there. Saying “10 per cent” conveys an aura of precision; but that need not rest on precise knowledge. The 10 per cent could perfectly well reflect profound uncertainty between two possibilities: the first, with a nine-in-10 weight, is that catastrophe is virtually impossible; the second, with a weight of one in 10, is that it is nearly inevitable. Taken on its own, the headline number tells us nothing about how tightly its assessment is bounded or about its practical implications.
We are therefore entirely in the dark as to the estimate’s significance and robustness. Yet it is not merely a truism that extraordinary claims demand extraordinary evidence: it is a foundational proposition of modern statistical inference.
None of that proves Hubinger’s estimate is incorrect.
At this point came another very modest set of billy goat butts ...
A simple analogy helps explain their concern. Tell a travel agency you must fly from Perth to Melbourne for the AFL grand final, and you would be horrified if, finding all flights booked, it arranged for a ticketed passenger to be assassinated, thus freeing a seat – and astonished too, since your request presupposed the agency would act within unspoken limits you assumed were understood.
In this dangerous world of analogies, who should arrive to give excellent advice? Eminent German philosopher Immanuel Kant, as always a voice of reason
Inspired by the presence of Kant, Our Henry reached new levels of rhetorical insight ...
Instead, whatever AI systems display is induced statistically from data and feedback, rather than experience, and because every decision rule that process induces needs a further rule to fix its application, the training mechanism can never fully close off the risk of error.
Just as that has been known for decades, so too has the broad architecture of the solutions.
The danger, Wiener wrote, lies in using “to achieve our purposes, a mechanical agency with whose operation we cannot efficiently interfere once we have started it, because the action is so fast and irrevocable that we have not the data to intervene before the action is complete”.
The answer, first conceptualised in the 1960s by Herbert Simon – another of AI’s greatest pioneers, and alone among them in holding both the Turing Award and the Nobel Memorial Prize in Economics – has three components.
The reptiles interrupted with an AV distraction, because mad King Donald is still ruler of them all ...
Trump administration scolds Albanese over ‘censorious’ tech crackdown
News24 host James Macpherson says the Trump administration has “scolded” the Australian Labor Party for their censorious Digital Duty of Care laws.
And so to Our Henry's Kantian konclusion:
These measures are not costless, and easier said than done. But they are no more than good engineering practice, checkable and even certifiable. Better still, they do not require a “slowdown” in technological progress, raising entry barriers or allowing China to steal the race. And they are far less socially harmful than that ultimate folly, the precautionary principle, which has become technophobia’s figleaf of choice.
The real risk is not that AI will kill us; it is that relying on increasingly intelligent machines will make us increasingly stupid, further eroding our collective capacity for clear thought and entrenching the hysteria that now so frequently drowns out the voice of reason.
As usual, Kant said it best: “Deficiency in judgment is what is properly called stupidity; and for such a failing there is, regrettably, no remedy.”
Exactly, precisely so.
When dealing with AI, always revert to Kant, born in Königsberg in the Kingdom of Prussia in April 1724 for the final word ...
Is there an infallible Pope in the house?
With the reptiles continuing to spoof the intermittent archive, the pond's range of reptile references continues to be limited.
The Australian Daily Zionist News couldn't resist this big lie ...
Luckily the reptiles used AP as their source, so correspondents could head off to other versions of the big lie, as at France24, Israel must win 'because we’ve got no other choice’, PM Netanyahu tells UN
There was much lying and disingenuous downplaying ...
“I can’t stand that. I won’t stand for vigilantism,” the prime minister said before accusing foreign leaders and the international press of distorting the scale of violence in the West Bank.
Netanyahu’s government is dominated by settlers and their allies, including the Cabinet minister who oversees the police. There have been few arrests of settlers involved in attacks on Palestinians.
Israel captured the West Bank and East Jerusalem in the 1967 war, and the Palestinians want both as part of their future state. The Palestinians and much of the international community view the settlements as illegal and as an obstacle to creating a viable, contiguous Palestinian state.
Netanyahu has said no Palestinian state will be established while he is Israel’s leader and has argued that the Palestinian Authority is not a partner for peace.
So keen was the pond to pander to the pretentious referential pundit that it didn't even bother checking out other worthy contenders, eager for attention. But there was one on the extreme far right of the reptile rag that caught the eye ...
No place in shrine for questionable frontier history
5 min
The caption for the snap designed to produce sympathy: A Last Post Ceremony at the Australian War Memorial Picture: David Whittaker
A cry from the wilderness behind the paywall: Make us preferred on Google
At last the knave behind the fuss stepped forward:
I am the co-convener of the Defend Sacred Ground campaign, which aims to put a stop to the inclusion of frontier conflict in the Australian War Memorial. Former PM Tony Abbott is our patron.
The onion muncher is your patron? Let the pond be clear, contempt is the first response, together with the thought that this must be black bashing of a familiar ancient English colonial kind ...
I have no doubt there are some, among the cohort pushing for inclusion of frontier conflict, who are genuine in their belief that this is an overdue and benign acknowledgment of our warts-and-all history. But the history being discussed is by no means settled. One narrative that has been widely and credulously accepted is that up to 100,000 Aborigines died in these so-called “Frontier Wars”. That figure is not supported by any empirical evidence. It has no place in the War Memorial.
If you hand your enemy a weapon, don’t be surprised if he uses it. Those well-meaning people will not be consulted when an activist minority in the Aboriginal community seizes upon this and trumpets it as an official acknowledgment that Australia was invaded, that the peaceful Aboriginal population was subjected to dispossession and genocide resulting in trauma that persists to this day. In short, that the nation-state of Australia is illegitimate.
Note the canny way that the war is reduced to a "frontier conflict" and images designed to terrify the hive mind are included ... People from the Aboriginal Tent Embassy and their supporters protesting at the Australian War Memorial in Canberra in 2014 Picture: Kym Smith
The pond was startled by the sense of a Colonel Blimp outraged that anyone would think to honour the fuzzy wuzzies ...
That this AWM proposal is contentious, and that the AWM Council is aware of this, cannot be in doubt. Prominent people have been speaking out about it since at least 2022.
In a recent piece in this newspaper, AWM director Matt Anderson states: “Before the development began, we consulted nationally, and the feedback was overwhelming: Australians expect difficult content in our galleries.” If you don’t remember being consulted, it’s probably because you are not a member of the Aboriginal and Torres Strait Islander community or a suitably vetted historian. As far as we can tell, that is the extent of the consultation undertaken by the AWM.
It is likely that most Australians would oppose this proposal, if given the chance. But until now, they have not been given that chance. And they should have been, because what is proposed is, on the face of it, in contravention of the Australian War Memorial Act 1980.
Just to alienate the pond even more, an image of the onion muncher, patron of all that demeans .. Former PM Tony Abbott at an Anzac Day service in 2026. Picture: Martin Ollman/NewsWire
More knaves came forward ...
We understand that the Council’s assertion that the proposal is in line with the Act is not based on independent legal advice but, as AWM chairman Kim Beazley has stated, on a “broad interpretation” of the Act. That would be a tenable approach if a) the Act were ambiguous, and b) the desired outcome were either in line with the clear intent of parliament or, alternatively, it now reflected an overwhelming and verifiable consensus among the Australian people. None of those conditions applies here.
The reprehensible "Bomber" was part of the act ... Chairman of the AWM, Kim Beazley, in Canberra in 2024 Picture: Martin Ollman
There came the usual mealy mouthed platitudes ...
The Council members may be quite genuine in their desire to seek “justice” for Aboriginal people and present a true account of our history, but it cannot be said that they have acted in good faith.
Rather than seek independent legal advice and put the matter beyond doubt, or at least give themselves a figleaf of credibility, they have relied on their own “broad interpretation” to further their own agenda. And they did this knowing that the matter is highly contested. Surely it is incumbent for chairman Beazley, as the statutory custodian of our national shrine of remembrance, to ensure everything is above board. The Australian War Memorial belongs to all Australians.
We call upon minister Keogh to direct the Council to remove all current depictions of frontier conflict, and to abandon its plans for the new Pre-1914 Gallery. As a minister of the crown, it is his duty to enforce the legislation.
Let me make one final point. We do not oppose this proposal because it is contrary to the Act. We oppose it, in principle, because it is a bad, divisive idea.
Lieutenant Colonel (rtd) Peter O’Brien is a Vietnam veteran and author of five books.
Discord on public messaging? The same might be pretty much said about this hot-cold-lukewarm analysis ...
There came a last gobbet ...
"The AWM is not a war museum. It is not a history museum" says this Abbott-adjacent loon. Maybe he should read up on Charles Bean, the Memorial's instigator. Bean was quite clear it was both a museum and a memorial. And any objective view would would establish that both functions are treated separately at the AWM.
ReplyDeleteYou’ve said it before I could, fellow Anony. If the AWM isn’t all of those things ,then why a e the finishing touch currently being put on a half a billion $ upgrade, primarily on the museum function? By that logic, shouldn’t all of the AWM’s areas other than the central shrine be either knocked down or hived off to, say the National Museum? The kindest interpretation is that “Defend Sacred Ground” is incapable of understanding that a site or organisation may have more than one function ( which isn’t implausible for a group under the patronage of the Onion Muncher…..). A perhaps more likely explanation is that their underlying motivation is good old fashioned racism and imperialism. There’s some irony in the group’s title - how many of its members would accept the concept of indigenous sacred sites?
DeleteAnd also to the point, Anony, that whatever purpose was originally envisaged for something, its purpose and use will inevitably change over time.
DeleteYes, who better to supply the guiding principles for the development of modern technology than a man who would have been gobsmacked by the sight of a steam engine or a Spinning Jenny?
ReplyDeleteThere is, dare I say it, a certain logic to some of the Hole in the bucket man’s comments, what with his talk of including engineering safeguards and the like. Unfortunately the HITBM appears totally unaware of the “move fast and break things l approach of modern tech bros and the blithe acceptance by so many of their “”Hey, we’re really smart, so trust us!” assurances. He rarely allows real events to intrude on his philosophical musings.
There’s also something almost admirable in the way that Our Henry manages to heroically equate the concept of a 10% chance of extinction with a loss of 10% of population and resources, and to then find that to be an acceptable, recoverable risk. It brings to mind that apocryphal news headline l small earthquake - not many dead”, as well as being a reminder that the man really missed his true calling as a planner of military operations, calculating acceptable casualty numbers.
I wonder if the Henry tries to keep up with discussion among human mathematicians about the claims of AI to have solved actual problems or interesting conjectures of the kinds that people like the obsessive Paul Erdõs scattered around the profession.
ReplyDeleteSome of that discussion is along the lines of ‘what exactly did the AI machine do in its millions of lines of code?’ In simplest terms - did the machine run every possible combination of whichever equations it claimed to have solved, and found that either they all worked, or at least one didn’t, so consider them ‘proved’ or ‘disproved’, - or did the machine develop an entirely new way of analysing the proposition - which we don’t actually understand,
Recent comment in ‘New Scientist’ from Terence Tao ‘often described as the world’s leading mathematician’, (and raised in Adelaide, Australia) emphasises that what the machines claim to be delivering, is not being explained and accepted in a way that can be ‘digested into the corpus of work that is taught to the next generation’ of human mathematicians.
Bit of trivia on the way - Erdõs met Terence Tao, then a 10 year-old-prodigy, on a visit to Adelaide in 1985.
Back to the Henry and his calculations that, he says, test a stated probability (10%) that AI could end human life. Implicit in what the Henry writes is that he thinks only of known ‘outages’ in cities that might end large numbers of human lives. He mentions water systems ‘collapsing’ to nuclear plants ‘exploding’. His attempt at analysis cannot make any assumption about other kinds of ‘outages’, because we have not experienced them. But some of the message from the discussion amongst mathematicians now is that AI may be capable of finding potential ‘outages’ of kinds that we have not imagined; very likely cannot imagine.
Someone who has read as much of the principles of philosophic thought as the Henry claims, should have considered what might be possible, if not necessarily calculable, outside their capacity to imagine completely different circumstances.