AI slop has broken the internet

"Slop" was Mirriam-Webster's Word of the Year last year. "Slop" or "AI slop" is defined as "digital content of low quality that is produced, usually in quantity, by means of artificial intelligence". You know the kind of thing: pictures of Shrimp Jesus, glue-topped pizza recipes, AI-generated songs, sham biographies, deep fake videos, talking cats, genuinely fake news, advertising that looks "not quite right". AI slop is everywhere, and it now, after just a few short years, makes up a good proportion of the entire Internet. 

And it is for this that we are investing billions of dollars, and straining our electricity and water systems to power endless miles of data centres?

I knew slop was a problem, but I hadn't appreciated just how much of a problem until I read an article in the Globe and Mail (originally from the New York Times) today. It seems that tech giants like Google, Meta and X are having to spend significant resources weeding out AI slop that was largely produced by their own platforms.

Spotify removed 75 million AI fake songs and duplicate uploads last year. Google wiped 130,000 YouTube channels of low quality content over the last six months. TikTok removed 377,000 videos in the first three months of the year for violating its AI policies. (TikTok USA is currently developing a feature that lets people dial up or down the amount of visible AI content.) 

But still the problem lingers and grows. Apple Music estimates that a third of what is uploaded to its site is now AI-generated. French streamer Deezer says it may be as much as half for them. More than 20% of short YouTube videos are AI-generated slop according to one (AI-powered) online video editor. Nearly half of the posts over 50 words on X are create by AI.

But the same tech giants that are trying to manage AI slop are continuing to make things worse. At the same tume as Meta showcased a detection, labelling and tagging system designed to make AI content more identifiable, it also rolled out an all-AI video stream widely considered to be the apotheosis of AI slop, and released an AI image-generator using its own public Instagram accounts as its references.

Part of the problem is that the less-discerning elements of the internet's clientele laps this stuff up, and AI content, however poor, can substantially inflate the viewership of many platforms. It can make big money for the creators and re-posters too. Unfortunately, it also makes it much harder for other users to find genuine, truthful and high quality search results.

As Substack's CEO put it, "it's getting harder to tell what's real on the internet", and "platforms that reward fakeness will create a race to the bottom". Quite. Substack recently announced its own AI detection tool for users, but the tool mistakenly identified some real, human posts as AI-generated.

This stuff is not easy. The point is, though, the onus should not be on users to figure out what's real and what's not. It seems trite to whine that "the internet is broken". But, really, the internet is broken.

All those e-scooters you see? They're illegal

If you are becoming increasingly exasperated by those annoying and apparently lawless electric scooters, you are not alone.

Stats on e-scooter crashes with pedestrians are hard to find, but accidents resulting in e-scooter rider visits to hospital, and even deaths, are increasing exponentially, especially among children. There have been at least two deaths of e-scooter riders in July alone.

How does it make you feel, then, to know that pretty much all of the e-scooters you see zooming around our streets, sidewalks and bike lanes are doing so illegally?

Having said that, it depends on where you live. You see, there are no national or even provincial rules or laws in place regulating e-scooters. What rules do exist are municipal. (And when was the last time you saw municipal laws being enforced?)

Thus, in Mississauga or Brampton or Ottawa, e-scooters my be legal due to a six-year old provincial pilot program they are part of. Even there, though, they are only legal if the scooters are unable to go faster than 24 kph (can YOU tell?) and if the driver is over the age of 15 (again, can YOU tell?) Also, you can't ride with a passenger, under-18s must wear a helmet, you can't ride them on the sidewalk, and if there is a bike lane available you have to use it. In Mississauga specifically, you can't ride them on a road with a speed limit of more than 50 kph.

So, it's complicated, and most people you see are probably breaking at least one of those rules.

In Toronto, on the other hand, it's not complicated at all. Toronto is not part of the pilot project, and you can't legally ride an e-scooter ANYWHERE in Toronto except on private property (so up and down your driveway, for example). You can buy them, but you can't use them. Once again, no-one is enforcing this ban - e-scooters are everywhere.

Outside of Ontario, it's just as confusing. In Vancouver, Halifax and Calgary, for instance, e-scooters are allowed, at least in bike lanes. In Calgary, though, e-scooters are only allowed if they are rented/shared, not if you own your own e-scooter. Plus, you can use your (shared) e-scooter on some sidewalks in Calgary but not others. Like I said, it's complicated (and unenforced).

Clear up your mess, Elon

I hope Elon Musk is going to go up there and clear up the mess his SpaceX rocket made on the Moon.

I'm OK with Canada's modest AI ambitions

Call me a Luddite, but I can't help but be extremely suspicious of the whole world's reckless infatuation with Artificial Intelligence (AI). I'm not saying the whole world is wrong and I am right - well, actually, I kind of am ... - but I just have this feeling that the level of obsession most people are exhibiting over it can't be healthy, and that we are likely headed down a dark rabbit hole of our own construction.

I can't help but feel that, much like 5G communications, we are adopting a huge, expensive and world-changing technology, just because it's there - because we can - not because we need it. AI adoption, and the number of huge, power-hungry data centres under construction, is not equivalent to economic output, or even productivity, although you'd be forgiven for thinking so from the media attention. 

AI is currently keeping the world's stock exchanges at near record levels, in spite of all the other crap happening in the world, but not for any good logical reasons. Almost all of the big IPOs hitting the stock exchanges are tech related (mainly specifically AI-related). Call it a bubble or whatever other label you prefer. The markets have almost completely decoupled from the economy, and are largely running on the fumes of potential AI profitability. I've seen this movie before; it doesn't end well.

I read so many articles bemoaning Canada's sluggish uptake of AI (here's just one example) compared to our peers. Well, usually compared to the US. 30% of the US's real GDP growth now comes from private investment in IT equipment and software, compared to 5% in Canada, we are told. Data centre construction has increased by 180% in three years in the USA, while Canada doesn't even separate the category out from other spending on transportation, utilities and communications buildings. But is that such a bad thing. Who has it right?

Indeed, there is outright opposition to US-style unregulated expansion of AI data centres (except in US-style unregulated Alberta). Manitoba recently nixed a huge data centre development because, as Premier Wab Kinew explained, "there's a big threat to the environment and not much benefit to.the economy". Well, he's got a point. A similar process played out in Hamilton, Ontario, where the city council voted to pause all data centre development in the area, and most other municipalities in Canada are studying their decision with great interest. Even ultra-conservative Olds, Alberta, is up in arms against data centres, although they're more concerned about local property prices than the environment.

An Angus Reid study suggests that 68% of Canadian would oppose a data centre in theIr own backyard, and only 15% would support it. 

Their main concerns? That they strain local electricity grids and water resources, drive up electricity prices, increase greenhouse gas emissions, generate noise and air pollution, negative affect local property prices, and the more general anxiety about the concentration of wealth and power among tech billionaires and the perception of an AI industry spinning out of control. Valid concerns all.

(For what it's worth, even in the US-style US, the rollout of data centres is meeting with increasing grass-roots push--back, from an unlikely alliance of Democrats, Republicans, environmentalists and just concerned citizens.)

Canada does have many AI data centres, hundreds of them, particularly in Quebec, and many more are being built as we speak. It's not like we have our heads completely in the sand. Do we have enough? Well, how long is a piece of string?

The huge bet south of the border on AI, and the all those data centres it relies on, is just that: a bet. The future gains from AI we keep reading about may not actually unfold as predicted, which would leave states, local governments and electricity payers on the hook for many decades.

Canada, generally speaking, is much more environmentally conscious than the US (despite Mark Carney's efforts to change that), and the environmental footprint of AI data centres has come under increasing scrutiny of late. AI requites massive electricity consumption, high water usage for cooling data centres, and heavy raw material extraction for hardware. It's no surprise that some of the strongest opposition to AI and, on a more local level, data centres is from an environmental perspective.

Another element, though, is the increasing Canadian distrust of AI: trust in AI is significantly lower in Canada than the global average, particularly as regards potential job losses. Data centres are the physical embodiment of AI and its perceived threat to society. That distrust of AI extends to the business community in Canada to some extent, which is investing much less in AI than the US.

Once again, is this necessarily a bad thing (as it is usually portrayed)? Is the helter-skelter scramble in the USA actually the right call? The US may be much more conservative than Canada in most respects, but Canadian business tends to be quite circumspect and risk-averse, and that's not always a bad thing (look at how Canadian banks dealt with the 2008 financial crisis compared the Americans).

Here's another thing. An analysis by the Federal Reserve Bank of Minneapolis found that the computer equipment, batteries, switchgear, copper wiring, etc, needed for the US's AI rollout amounted to 23% of all US imports last year, increasing its trade deficit by roughly $200 billion. Canada's AI-related imports, on the other hand, remained pretty much consistent with previous years, and much more manageable.

So, who has it right? History will probably tell, but that doesn't much help today's planners. Mark Carney is making some moves to at least be seen to be doing something on AI, but even his strategy is pretty modest in the scheme of things. And, you know what? I'm OK with that. 

Canada pledges to accelerate AI development just as others advise caution

It sometimes seems like half of all articles in the newspaper are something to do with artificial intelligence, sometimes something good, often something bad.

Canadian news outlets are all reporting and commenting on Prime Minister Mark Carney's big reveal of Canada's AI strategy yesterday, outlining the government's approach to what he calls "the defining technology of our era". Under the banner "AI For All", Carney's upbeat, if somewhat vague, presentation promises: protections for Canadians, young and old, from the risks and potential online harms of AI (through modernizing consumer privacy legislation, introducing online safety laws, watermarking AI-generated content, creating a Canada trusted AI certification program, protecting elections from interference, etc); access to free AI literacy training (an area where Canada particularly lags), including for post-secondary students; up to 90,000 AI-related job opportunities for young Canadians, plus another 250,000 new jobs through increased AI adoption by 2031; boosting business adoption of AI from 12% today to 60% by 2034; building a world-leading supercomputer by 2031; building up a multilateral alliance giving Canada "sovereign autonomy" in key AI capabilities; and $2 billion in direct investment to achieve these strategic aims.

It all sounds very forward-thinking, ambitious, even Panglossian, although what it is really is Canada trying to make up some lost ground on everyone else. Some Canadian tech companies like Cohere are extremely positive about the new policies, calling it "an incredible step forward", "a defining moment of opportunity", and "the beginning of the next chapter", particularly since Canadian researchers were at the forefront of AI's early development.

Critics of the strategy, though, have questioned how AI is supposed to create so many thousands of jobs rather than cost jobs, as many are predicting (Signal 49, for example, warns that AI and automation could lead to 550,000 Canadian job losses by 2030 as businesses restructure.) Others have complained that the government strategy is vague on timelines and specific safety measures. 

Be that as it may, the Canadian government had to make some kind of an announcement about how it is pursuing and dealing with AI, if only because everyone else is.  

It's interesting, though, that it comes hard on the heels of Pope Leo XIV's papal encyclical Magnifica Humanitas, which warns about the urgent need tor an ethical framework for AI to prevent it from becoming an instrument of domination and an agent of lethal autonomous warfare, to avoid the manipulation of reality (e.g. deepfakes), and to protect the rights of workers, the marginalized and the vulnerable.

It also comes just as major AI developer Anthropic warns that maybe there should be a coordinated and verifiable pause in all AI development, because AI systems are approaching the point where they can improve themselves, without human intervention, faster than society can manage the risks.

This, then, is the environment in which Canada is, somewhat belatedly, committing itself to accelerating AI development and adoption.

Wait, so now we need to worry about AI worms?

As if viruses, malware, phishing threats, and all were not enough, now cybersecurity experts are warning about AI-powered worms.

While viruses require a user to execute a file or open a link, worms can slither their way into a network or operating system to exploit vulnerabilities, move from device to device and replicate themselves across computer networks, all without human intervention. Once in, worms can carry out any number of cyberthreats, from network overload to denial of service attacks to spam distribution to ransomware delivery to data theft. Any device connected to the Internet - whether it be computers, cameras, printers - is potentially at risk. 

Remember the ILOVEYOU bug back in 2000, and WannaCry in 2017, and all the chaos and fear they engendered? Well, they were worms. But now, the addition of artificial intelligence into the mix has made everything that much more dangerous and harder to fix. While most cybersecurity concerns have centred around large language models like OpenAI's GPT-5.5-Cyber and Anthropic's Claude Mythos Preview, AI worms are a much cheaper and more accessible way for hackers to wreak havoc on the Internet.

Back in the day, worms used to follow scripts generated by humans and, when they hit a defence they weren't designed to crack, they failed and died. AI has upped the ante and can create worms that modify their tactics as they spread from one device to another, tailoring attack strategies to each machine they interact with.

So potentially destructive is their impact that researchers actually debated whether or not to publish their findings at all, worried that they would give bad actors a ready-made blueprint for how to conduct such an attack. Eventually, they decided to publish and just omit certain operational details, judging that the call to action - particularly at a governmental level - is a much higher priority.

Potential investors in Spacex should be very wary

The IPO for Elon Musk's SpaceX is expected to be the biggest ever, and will convert it into an almost $2 trillion enterprise. However, potential investors might want to have a good look at the way the company is set up, and particularly how Musk is paid and how his share holdings work.

An IPO (Initial Public Offering) is the way that a private company transforms into a publicly traded company, and is the way that companies raise capital for expansion. It allows institutional and retail investors to get a piece of what they think will be an exciting and profitable venture. SpaceX (Space Exploration Technologies Corp, to give it its full name) may well be exciting - space! rockets! Mars! - although the profitability piece is much less certain.

And concerning Musk's position, investors should be pretty wary about investing in a company that has been expressly constructed around him in order to maximize his income and his control. The shares that will become available for ordinary investors at next month's IPO will be Class A shares that confer one vote each. What Mr. Musk has are Class B "super-voting" shares that carry 10 votes a share. Musk has 5.5 billion of these B shares, giving him around 85% of all votes. And he has those votes even though he doesn't technically have the shares in his hands until the company achieves some increasingly-unlikely targets, such as establishing a colony on Mars with a million inhabitants, launching high-powered data centres into space, etc. 

This set-up allows Musk almost complete control over the company, including an ability to appoint insiders to its board, to set his own compensation package, to insulate himself from shareholder lawsuits, etc. Investment experts say they have never seen anything like it, calling it "insane" and that the governance structure "freaks me out".

Caveat emptor, caveat emptor, caveat emptor!

Does Donald Trump type out his own social media posts?

I was trying to figure out how Donald Trump found the time and the energy to make all those unhinged late night Truth Social posts. I mean he'd have to spend most of the day consuming the latest conspiracy theories and disinformation streams, wouldn't he, and then stay up half the night as well? Well, it turns out that he has a "personal aide" or "executive assistant" called Natalie Harp who does most of the leg (and finger) work for him.

Little-known and low-profile, Harp is one of his most influential aides, and has been instrumental in the nearly 8,800 posts Trump has made since the start of his second term. She has access to Trump's Truth Social account, but she also has an alarming amount of control over what Trump writes about, and she also controls much of what reaches his desk in the first place. Unusually, she appears to work directly for Trump and answers only to him, without any oversight by Trump's chief of staff, national security officials, or other communication aides, which has raised more than a few eyebrows.

How it works is that she arrives at Trump's residence each evening with a stack of printed-out draft posts, on subjects that she thinks Trump might want to publish under his name. She will have spent most of her day scouring the internet for suitable videos, images and text that would align with Trump's current grievances, whims and worldview. 

Then, once Trump approves the selections, she logs into his account, usually in the late-night early-morning hours, and posts large batches of messages in quick succession, occasionally up to 160 in a single night's activity. This gives the impression that Trump is up at all hours, working(!) for the American people.

So, who is this eminence grise? Well, not so grise, it turns out. Ms. Harp is 34 years old, and sports long blonde hair and lots of make-up - a typical Trump pick, you could say. She used to be an anchor with One America News Network, a far-right political commentary TV network, and joined Trump's campaign in 2022. She became known as the "human printer" because part of her brief was to follow Trump around, even onto golf courses, to print out pertinent information for him, as Trump is of an age where he prefers the printed word to screens.

Natalie Harp

Although she has somehow managed to remain quite private, she has attracted some scrutiny, particularly for some rather embarrassing adoring letters she has sent to Trump, which include comments like "You are all that matters to me", "I don't ever want to let you down", and "I want to bring you joy". Sycophantic, bordering on creepy. Others in Trump's staff find the "aggressiveness of her attention" discomfiting and unnerving, and possibly even a security concern.

Trump, however, is a big fan, and anyway is a sucker for the attentions of a young blonde woman. (Snopes' fact-check points out that reports from the New York Times and Washington Post of an affair between the two are actually just based on hearsay and speculation.) So, maybe don't feel too sorry for him tapping away on his phone into the wee hours - he's probably actually tucked up in bed with a can of Coke at the time his Truth Social posts are sent out.

Why does Russia care about Alberta separatism

While I am on the subject of disinformation and foreign political interference, the Alberta separatism movement, such as it is, is a prime target for foreign propaganda and interference. There's already plenty of home-grown disinformation and misinformation out there; a bit more from outside the country probably wouldn't get noticed. But it seems there's more of it going on than we realize.

While monitoring Russian disinformation campaigns targeting Ukraine and support for Ukraine in Canada, researchers were surprised to see that a lot of the mendacity emanating from Russian propaganda outlets like the Pravda Network, CopyCop, Storm 1516 and the Internet Research Agency was specifically targeting Alberta separatism. Not only are there large numbers of social media posts - hundreds of thousands - but complete websites (like AlbertaSeparatist.com) that purport to be grass-roots campaigns by aggrieved Albertans, but are actually being generated by Russian propaganda farms. AI-generated "slopaganda" is also being used to pollute the information environment, intensify grievances, and confuse legitimate debate with foreign manipulation.

I mean, you know this stuff happens. But, when you stop and think about it, what are they really trying to achieve? What is it to Russia whether Alberta separates or not? How does it merit such a concerted effort to them? Don't they have bigger fish to fry?

I can understand that the Trump administration and the MAGA mob might have a vested interest in Alberta separation (and they are indeed attempting to influence the vote, albeit mainly thought more mainstream, and less illegal, conduits). But Russia?

DisinfoWatch says that Russian interference in the Alberta separatism debate "appears doctrinal, operational and sustained", all part of the Russian government's long history of exploiting divisive issues in Western democracies in general. Division for the sake of division, as far as I can tell. There is much more detail on DisinfoWatch's website. The Global Centre for Democratic Resilience and the Centre for Artificial Intelligence, Data and Conflict at the University of Regina both concur.

I guess I understand, in general terms at least, that anything that weakens the West strengthens the East. It just surprises me that the hawks in the Kremlin see this as an issue deserving of not-insignificant funds. And it's such a bore that we have to spend time and money guarding against these digital incursions.

All of this, and the revelations that a separatist group gave unauthorized access to the province's list electors to hundreds of people (potentially including foreign actors), has cast a pall over the whole enterprise and put the legitimacy of its outcome in doubt.

UPDATE

Given all of the above, it seems particularly ironic that Alberta's Minster for Public Safety and Emergency Services should choose his week to announce that an RCMP report has somehow managed to find that there is no evidence that Alberta's separatist movement has been subject to foreign interference.

Accent-masking AI tech just another step down the slippery slope

The wholesale adoption of articial intelligence (AI) has raised all sorts of red flags and complaints, and some of its applications do seem pretty morally grey (or at least tone deaf). This is one I hadn't anticipated, though.

Canadian telecom Telus has started using AI-driven accent-correcting technology in its call centres, as well as in its internal phone operations. The technology, provided by a third party company Tomato.ai, uses speech-to-speech models to transform live audio. It is designed to preserve the speaker's voice and "emotional tone", whole addressing mispronunciations and the sometimes hard-to-understand accents of many call centre agents working from overseas (or even from within Canada).

Of course, this has raised hackles. Opponents say that any kind of deception of this kind should be either stopped, or at least Canadians should be informed up-front that AI is being used, particularly as many call centre functions are being outsourced overseas, depriving Canadians of job opportunities. Canada has seen a substantial amount of customer service related job reductions in recent years, particularly within the telecom sector.

There again, it is argued, the technology improves operational efficiency, thus saving Canadians money in the end. It helps speed up calls, helps customers find good solutions, and protects service agents from harassment and discrimination. That may be the case, but it is a solution to a problem of the company's own creation (off-shoring call centres to save money).

Should Canadians have the right not to be deceived by AI technology, as telecom unions are arguing? Probably. Does it really matter? Well, yes, it kind of does, particularly from the point of view of AI-induced job losses. Why don't we just go the whole hog and have AI man the phones directly, rather than just deceiving customers like this? Well, because AI is just not that good yet. But it will be soon.

Two tech billionaires face off

The press is fairly salivating over the prospect of two tech-bro billionaires facing off agaist each other in court in California. It involves betrayal, deceit and unbridled ambition. And lots and lots of money.

Elon Musk and OpenAI CEO Sam Altman used to be best buddies. Musk actually bankrolled much of OpenAI's early research in the heady days of 2015, back when it was an idealistic not-for-profit looking to save AI (and the world) from the rapacious designs of the less caring technology companies that were starting to emerge.

It soon became apparent, though, that a lot of money could be on the table for Altman and OpenAI, as the company's ChatGPT became the world's go-to AI platform, and Altman's altruistic vision wobbled. The company converted to a for-profit corporation last year, and is now valued at nearly a trillion dollars. Altman, OpenAI president Greg Brockman, and many others in the organization have become very rich indeed. (Altman's own fortune is estimated to be around $3 billion.)

Musk, the world's richest man, whose own net worth is heading inexorably toward the trillion dollar mark, has accused Altman and Brockman of double-crossing him, and of abandoning the company's founding vision of "an altruistic steward of a revolutionary technology". The picture of Musk, not known for altruism himself, taking the moral high ground is a confounding one. Plus, Musk now has his own AI company, xAI.

It's not going too well for Musk thus far. He has abandoned his original bid for $100 billion in damages after a series of pre-trial.rulings went against him. He is now seeking an unspecified amount of money to be paid into the charitable arm of Open AI (which still exists). He is also looking to oust Altman from OpenAI's board because, after all, this is personal.

The trial is expected to provide some riveting theatre as the two larger-than-life characters face off. As US District Judge Yvonne Gonzalez Rogers says, "Part of this is about whether a jury believes the people who will testify and whether they are credible". In his opening remarks, Musk laid out his views: "The fate of civilization is at stake". Altman, for his part, gushed about Musk: "You're my hero". Yikes!

Musk has other ongoing legal challenges too: he was recently held liable by another jury for defrauding investors during his $44 billion takeover of Twitter in 2022. Altman and OpenAI are also being sued by residents of Tumbler Ridge, BC, after a mass shooting that they claim should have been prevented by OpenAI. It's a messy old (new) world out there.

Maybe we shouldn't really care that two such uncongenial figures are having at each other like slightly grubby quondam titans. But it is hard to look away.

Iran may be down, but apparently it's not out

One of the more disturbing aspects of the US-Israel-Iran war is that I find myself rooting, almost against my will and my better judgement, for Iran.

Now, I have no love of Iran - it is a benighted, repressive theocracy with anger management issues - but I have no love of either Israel or the USA either, and they are the ones responsible for starting this unprovoked and illegal war. It's also partly a habitual ingrained tendency to support the underdog in any confrontation or competition. So, when Israel or the US suffers a set-back in their military plans, I tend to respond with a - muted and rather shame-faced - cheer. Is that so wrong?

The war has been going on for a month now, much longer than Trump ever expected. Even though Trump says the Iranian army, navy, air-force and missile capability has been "obliterated", Iran is clearly still hanging on, unbowed and unrepentant.

In fact, in recent days, Iran's "non-existent" missiles and drones have been getting though Israel's formidable defences with more and more regularity, which has some military commentators wondering whether Israel's much-vaunted "Iron Dome" defence system hasn't been damaged by Iran. 

Iranian missile strikes have found their way through to strategic cities like Tel Aviv, Dimona and Arad recently, successfully evading the layered network of detectors and interceptors shared by Israel, the US and its Gulf partners. It's possible that Israel's stock of interceptors is somewhat depleted after the prodigious  barrage of missile attacks from the Iran and proxies like Hezbollah and the Houthis. That does seem likely, but, more and more, experts are positing the idea that the radars and sensors that underlie Israel's integrated air defence network might have been damaged, creating gaps in its detection ability, and leaving both Israel and US forces and assets much more vulnerable than was previously thought. Israel's airspace in particular suddenly seems penetrable, even by a wounded Iranian military. And you know what they say about wounded animals.

Don't get me wrong, I am not going to be out on the streets of Toronto at the weekend, chanting "Death to the American devils!" But it's hard not to feel a bit of righteous schadenfreude when the top-dog aggressors get their comeuppance.

Another Moon mission? Why?

After a few false starts as NATO erred on the side of caution and dealt with various technical challenges, the Artemis II manned mission to the Moon is due to blast off on April 1st (foolish? I don't think astronauts are superstitious).

It doesn't plan on landing on the Moon - the last time that happened was 1972. Artemis II will just fly around it and back home. But this 10-day trip around the Moon is still a big step in the reboot of American lunar ambitions, and is seen as an important testing run for future missions. Ultimately, the plan is to establish a permanent human base in the Moon, theoretically by as early as 2030.

The Artemis program is the successor to the Apollo program of the 1960s and 1970s. (In Greek mythology, Artemis was the twin sister of Apollo, so the name was pertinently chosen.) Artemis I was an unmanned flight 3½ years ago to test out the Space Launch System (SLS). Artemis II will be the first manned mission to go past the International Space Station (in near Earth orbit) since 1972, and the first to include a Black astronaut (Lt. Cmdr. Victor Glover), the first to include a woman astronaut (Christina Koch), and the first to include a Canadian astronaut (Col. Jeremy Hansen).

How did a Canadian wangle his way on there? Negotiations over several years (pre-Trump, back in the days when the USA and Canada actually got along) yielded an agreement whereby a Canadian astronaut got to tag along in return for about $2 billion in Canadian investment in the lunar program, and the provision of an AI-enabled robotic arm designed to operate on a lunar orbital space station called the Lunar Gateway. (Robotic arms are something of a Canadian specialty - the original Canadarm paved the way for Canadian astonauts Marc Garneau and Roberta Bondar to fly into orbit; Canadarm2 was Chris Hadfield and David Saint-Jacques' ticket to the ISS; and Canadarm3 was part of the Artemis deal for Jeremy Hansen.) 

As it turns out, the Gateway project has since been abandoned in favour of a push for a lunar land base (at least partly to get ahead of Chinese lunar ambitions), so the future of Canadarm3 is unclear, but Hansen still gets to fly. The deal also includes a second lunar mission for a Canadian, and a Canada-based control centre for the robotics (maybe?) The European and Japanese space agencies are also partners in the Artemis program, and they are also expecting to have astronauts included on future missions.

Incidentally, the current American push for the Moon is far from all Donald Trump's doing, whatever he might try to convince us of later. It was George W. Bush that first announced a new initative for NASA after the 2003 Columbia space shuttle disaster sent US space ambitions into an existential tailspin. Barack Obama repurposed Bush's lunar project into an asteroid mission, but that too foundered, and space exploration gradually became the province of private space companies like SpaceX for a while. It was only when space missions by Japan, India, Europe and particularly China started to eclipse American efforts that NASA announced its new lunar direction. A new space race had begun.

When Col. Hansen orbits the Moon in the Orion crew module, he will get to see, first-hand, parts of the Moon's far side that have never been seen by human eyes (although the flight's trajectory will actually keep it at quite a distance away). As Chris Hadfield puts it: "the first non-American to fly beyond Earth orbit will be from Canada, not from Russia, not from China, not from India". Depending on the precise trajectory taken, he will probably be further from the Earth than any human ever before at one point.

Do we need to go to the Moon? No. Is it exciting? Sure!

UPDATE

As the excitement starts to diminish after a successful Artemis II mission, reality sets in. Some people are distinctly underwhelmed that, in 2026, all we can manage is a short fly past the Moon, while 57 years ago, men were speaking around on another heavenly body.

Another reality check: even while the Artemis II project was under way, the Trump administration was announcing stringent cuts to NASA's budget. Cuts of 23% were announced, mainly to pure science programs and future robotic missions. At the same time, the budget of Space Force - the Trump-created sixth branch of the US military - was more than doubled, and now stands at about four times NASA'S budget. This gives an insight into Trump's real priorities.

More AI prophets of doom

There seem to be two irreconcilable trends in artificial intelligence (AI) development. There are the boosters, largely comprised of tech giants and other large companies, who see oodles of money to be made from it and are doing everything they can to promote it and establish it as the default go-to solution for pretty much everything, even at the expense of flesh-and-blood people, and despite the fact that very little money is actually being made from it thus far. 

And then there are what you might call the prophets of doom, the more thoughtful, cautious contingent, largely comprised of eggheads and theorists (including most of the early pioneers), who are warning that AI development is proceeding too quickly and with insufficient safeguards, or - at the extreme end of the spectrum, but far from isolated - those who are warning that current development trajectories are probably headed towards human extinction.

The former trend seems to be winning at the moment, but that doesn't mean that we should just ignore the eggheads and theorists. The latest warnings come from a researcher called Nate Soares, perhaps not officially in the egghead category, but still a smart cookie with a gift for useful analogies and a well-turned simile.

Soares points out that when (not if) superintelligent AI becomes better than humans at absolutely everything, then "things would get wild". In fact, things would get wild long before that: "It's sort of like saying if you dropped a 500-ton weight on a chicken, the chicken would die. I'm not saying smaller weights wouldn't work. I'm, like, this weight is definitely big enough."

We can't actually design AI to be safer, because AI is not really designed, it grows like an organism. And it is hugely complex, so you can't just point to, say, Line 73 in the code and say, "that needs to be changed". It's more like disciplining a naughty child - we can't just rewire their brains. Worse, machines don't have empathy, sympathy and human emotions, so we can't even appeal to that.

You can actually force an AI application to admit that it made something up, or that it had a "hallucination". It will say that it knows you didn't want it to do that, but it did it anyway. This drive to produce something even if it is incorrect is just one worrying sign we are already seeing of AI going rogue. 

In the same way, while no-one is saying that AI wants to deliberately harm humans, that's not to say that it won't do so while trying to achieve some other task or goal. It doesn't actually care about us - is not capable of caring about us - so this utter indifference combined with its great technological might may result in it just rolling over us en route to completion of an otherwise unimportant task.

AI company Anthropic, for example, may be training its AI according to a "constitution" to try to align it with human values, but even the head of Anthropic thinks there is a 25% chance that AI will go catastrophically wrong. That's a pretty big chance. And yet these companies are still barrelling ahead, most with much less ethical questioning than Anthropic. This is happening mainly because they think, "If I don't do it, somebody else will, and they will probably do an even worse job of it". And that's those who are not motivated by much more base considerations, like money.

Mr. Soares concludes with another analogy: we wouldn't build a plane with no landing gear and just assume that such a minor problem could be fixed while the plane is in flight. Nor would we launch a plane that we were 75%, or even 90%, sure was safe. But that's kind of what we are doing with AI.

I thought it was funny that the article ended with an Editor's Note: "AI tools assisted with condensing the original podcast transcript".

Social media's Big Tobacco moment

I confess I'm surprised it ever came to this, but kudos to the American courts system for having the gumption to rule against the combined might of Meta/Instagram/Facebook and Google/YouTube in finding the social media companies liable for intentionally building addictive social media platforms that harm the mental health of young people.

In this case, it was a 20-year old woman called Kaley (or K.G.M.). She sued Meta and Google over her childhood addiction to social media, claiming the platforms left her with body dysmorphia, depression and suicidal thoughts. Yesterday, she won US$6 million in compensatory and punitive damages, of which 70% is to come from Meta and 30% from Google. (TikTok and SnapChat both settled out of court with the same plaintiff for an unknown sum before the Meta/Google trial began.)

Now, $6 million is not really going to make the likes of Meta and Google sit up and take note - the plaintiff's counsel was initially claiming $1 billion! - but, given that there are tens of thousands of such victims lining up to sue Big Tech, this precedent could indeed be material. This is social media's Big Tobacco moment.

Of course, it's not over until it's over, and both companies say they will appeal the ruling, to try and stanch the bleeding which could indeed cost them dearly over time. If they manage to get the case overturned, is it still a landmark case? Probably. It validates the legal theory that social media sites or apps can cause personal injury, that it is essentially a defective product, if you will. And it will almost certainly encourage many more tentative litigants to try their luck with the legal system, although that is not a task for the faint of heart.

Let's be realistic about privacy and civil rights

When I was younger - much younger - I felt strongly about privacy and civil rights. As an old geezer now, I find I am much less concerned about such matters. Maybe that's a normal part of ageing: young folks tend to be much more idealistic and less pragmatic.

Either way, I find that the opposition of "privacy advocates" to the recently-tabled federal Bill 22 to be overblown and exaggerated. The bill grants law enforcement agencies like the police and CSIS new "lawful access" powers to online data and cellphone records for investigative purposes where there is reasonable suspicion that a crime has been committed.

That seems like a perfectly reasonable compromise to me, although I've a suspicion that, twenty years ago, I would have been up in arms about it. But this is not government spying. This is not some kind of police state manoeuvre. Nobody is tracking our every move in our everyday lives (well, except Google, Facebook and whole host of other tech companies!)

It merely allows law enforcement agencies to gather more pertinent and important information in specific circumstances where a crime may have been committed. If you don't go around committing crimes, this new law will not affect you at all.

I don't think I am being naïve here, as civil rights organizations would almost certainly agree. Neither is it necessarily the start of a slippery slope. It's just a practical aid to law enforcement in Canada, and one they have been requesting for for years.

Hell, I don't even care that much about those tech companies these days. If they want to track my online movements and target advertising towards my specific preferences, that's fine by me. I can choose to ignore the ads - which is what I usually do - or maybe I'll actually discover something I really like. At any rate, it seems like a small price to pay for all the "free" services companies like Google provide. (I'm less supportive of the likes of Facebook, X, etc, which don't really provide any useful content in my opinion; consequently, I have not had accounts with those kinds of social media bloodsuckers for decades. Again, my choice.)

AI in war - we might already be there

It's a scary time when the two top news topics are war in the Middle East and Artificial Intelligence (AI). Even scarier is an article about the confluence of the two. But that's where we are right now.

It was of course only a matter of time before AI was enlisted into the service of war-making. In some areas of military surveillance, intelligence and reconnaissance, it has been used for years, often with alarming repercussions (take Israel's Lavender program to identify Hamas targets, for example). But if it becomes more widely employed in planning, targeting and even decision-making, then we are in deeper trouble.

AI military use has become more problematic with the advent of agentic AI, where decision loops can become fully-automated, without any human intervention. And, of course, that is exactly the direction military strategists are going, despite the potential for Artificial General Intelligence (AGI), where AI overtakes human intelligence and even learns to use deception to get its own way (the kind of stuff that AI pioneer Geoffrey Hinton and others have been warning us about for years).

It's no coincidence that several leading AI researchers left Open AI over worries about the company's lack of ethical guardrails, and many of them gravitated to competitor company Anthropic, which seemed to have a more constrained and prudent approach to its research. For example, Anthropic CEO Dario Amodei has publicly warned about the potential for AI to serve some of the worse tendencies of autocratic states.

It was perhaps a surprise, then, that the US Department of Defense (now the Department of War) signed a contract with Anthropic last July, and it developed Claude Gov, an extension of Anthropic's publicly-available Claude chat-box, and the first large language AI model trained on classified material. The agreement did come with restrictions and guardrails: that the tool would not be used with fully-autonomous weaponry, and that it would not be used for mass surveillance of American citizens.

Then, perhaps predictably, the Republican administration pushed back. Secretary of Defense/War Hegseth started talking about an "AI-first war-fighting force" and accelerating AI adoption "from campaign planning to kill chain execution" (he sure has the military lingo down-pat, even if he is perhaps not quite sure of the ramifications). There was talk of freeing the system from "ideological bias" and "red tape" - which is code-speak for imposing their own ideological bias and removing all inconvenient rules - and even using AI to pursue "social engineering" and "cultural agendas", all things that Anthropic explicitly aimed to avoid. Yikes!

Initially, Anthropic tried to enforce its own explicitly-stated rules, and refused Pentagon officials unrestricted access to Claude Gov's capabilities. But then Trump too got involved - well, of course he did! - calling Anthropic "a radical left, woke company" (the worst insult ever in his books), and accusing it of trying to "strong-arm" the US military. He even threatened to designate Anthropic a national security supply chain risk (a label typically reserved for Chinese and Russian companies), and threatened legal action to force the company to comply with his whims.

Despite Anthropic's concerns, there are reports that Claude Gov was involved in both the US kidnapping of Nicolás Maduro in Venezuela, and the assassination of Ayatollah Ali Khamenei in Iran. It's not clear to what extent these operations involved fully-automated weapons systems, and there is no way the Trump administration is going to tell us. 

So, it seems the guard-rails are at least partially down, and the training wheel are off. In the hands of an unbalanced and unhinged administration like that in today's White House, all bets on weaponized AI are off. Science fiction tropes are in everyday use, and none of this is good.

Agentic AI takes over from generative AI

We're only just starting to get our heads around what generative artificial intelligence (AI) is. Now we have to deal with "agentic AI".

While generative AI refers to a computer response to a single query using natural language processing, agentic AI uses sophisticated reasoning and iterative planning to solve much more complex multi-step problems. The idea is that AI serves as your agent to help you make better decisions, employing multiple data sources and third party applications to analyze challenges, develop strategies, and execute tasks, all with minimal human intervention. It learns from the results it achieves, and uses this feedback to improve future plans and actions.

Frankly it's what most people think AI is (or should be), although it's only very recently that AI has become sophisticated enough to deliver on this promise. Generative AI, by comparison, is really just a jumped-up search engine, based on single input prompts. For example, generative AI can used to create some marketing materials, but agentic AI can used to actually deploy these materials, track their performance, and adjust the marketing strategy accordingly.

Agentic AI is being increasingly employed by businesses to personalize customer service, streamline software development, and facilitate patient interactions. Another place it's being used (and this is what initially triggered my interest and concern) is in shopping and merchandizing. 

For example, Canadian shopping goliath Shopify has teamed up with Google and Microsoft to help shoppers find and buy its products more easily and even to help them make purchasing decisions. Shopify merchants can now sell directly through Google's Gemini app and the AI Mode of Google Search, as well as through Microsoft Copilot. Walmart and Mayfair have also recently set up similar agreements. Shopify is even setting up agentic plans with other ecommerce platforms, which will allow online stores throughout the world to sell through Shopify's catalogue, which already comprises billions of products. So, shoppers can buy multiple items from different places without ever leaving the initial AI conversation. Yikes! 

Analysts are saying that this has the potential to revolutionize online shopping and advertising. Sound familiar? The merchants stress that shoppers are in control of the whole process, and that they have the final call. 

But the agentic AI system can even complete checkout on a customer's behalf, based on pre-entered discount codes, loyalty credentials, billing options, and payment information. I do worry that this makes shopping a bit too easy, and I can easily see it spiralling out of control, or even becoming addictive. After all, it's the customer, not the AI, that has to pay the credit card bill at the end of the month. 

I don't have any evidence to back it up, but there just seems a lot that could go wrong with this developmemt.

And where are we with robotaxis and self-driving cars?

I confess I had written off self-driving cars completely several years ago, after a series of accidents and the admission that, actually, self-driving cars were not as easy as initially predicted, and that maybe some problems (including those tricky moral ones) maybe forever outside the purview of computers.

But, undeterred, some companies have persisted with the idea, with mixed results. The biggest, best-known ones are Tesla's Robotaxi (which uses its popular Model Y cars) and Google/Alphabet's Waymo (which uses the sexier Jaguar i-Pace sports utility vehicle). But other companies are also in with a shout, including Uber's Avride, Amazon's Zoox, Volkswagen, and Aurora's driverless trucks. (Some companies, like GM, have retreated from their autonomous car aspirations.)

They are progressing, if slowly, although Tesla in particular is progressing much slower than Mr. Musk led us to believe. It has a modest 30 Robotaxis operating in Austin, Texas (which seems to be ground zero for autonomous car testing), and they all have a human in the passenger seat monitoring them, for now at least. That's a far cry from Musk's 2016 promise of cars driving themselves across the country within two years, and his 2019 promise of a million Robotaxis on the road within a year. 

Google's Waymo fleet is larger (Google actually started its self-driving car research well before Tesla), with about 200 vehicles in Austin, and another 2,300 in Phoenix, San Francisco, Los Angeles and Atlanta, with plans to extend to 20 more cities in 2026, including Dallas, Washington, Miami, and London, England. Waymos operate without human monitors.

Mr. Musk is still saying that, after a slow, cautious start, Tesla will overtake Waymo and, for some reason, many deep-pocketed investors seem to believe him. Millions of Teslas on the road today have the hardware to convert into self-driving vehicles with just a tweak of the software. Plus, Tesla's system relies solely on cameras (lots of cameras), while Waymo and other companies use a combination of cameras, radar and laser sensors. Cameras alone are a cheaper solution, but they can easily be flummoxed by fog, snow, glare and other factors, and many commentators see Tesla's fixation with cameras as a severe limitation, especially as the cost of radar and laser sensors continues to fall.

It's far from clear that self-driving cars and taxis will ever generate trillions of dollars in revenue, as Musk insists, or that they will ever achieve profitability. There are many hidden overheads, like manned monitoring centres, special cleaning, etc, that are rarely mentioned. Power outages, as occurred in San Francisco recently, are also a potential issue, as self-driving cars grind to halt in that eventuality. They still have problems understanding hand signals, and have often been reported ignoring police officers, firefighters and emergency workers trying to direct traffic (and even, in numerous occasions, ignoring a school bus' signals), resulting in some close calls. A driverless Waymo tax in Scottsdale, Arizona went AWOL earlier this year, and started spinning in circles in a parking lot, with the hapless passenger locked in. Although no deaths for some time...

UPDATE

Of course, these are just the Western/American developments. As you might imagine, China is way ahead on this stuff. Baidu's Apollo Go driverless taxi service has been operating in dozens of colors in China for some time now, and has accumulated millions of driverless rides. Ride-sharing apps Uber and Lyft have recently announced partnerships with Baidu, and Chinese robotaxis are expected to hit the streets of London and other western cities next year.

What is vibe-coding, and does it work?

Maybe you've never even heard of vibe-coding - it's a pretty recent phenomenon - but it's all the rage in the technology world. Essentially, vibe-coding is using an artificial intelligence (AI) platform to allow amateurs and non-coders to create functioning software and websites.

Sounds easy, eh? You just use a vibe-coding platform like Lovable or Cursor (recent start-ups already worth billions of dollars), or even more familiar large language models like Gemini or Claude or ChatGPT, and use everyday conversational prompts, and the AI does the rest.

Except, of course, it's not as simple as that, as the Globe article explains. It can be a confusing and frustrating experience, and the AI is quite likely to throw out buggy software full of security risks, that doesn't bear any resemblance to the masterpiece you had in mind, and quite likely doesn't even work. One study found that AI models introduced known security flaws about 45% of the time. In fact, a whole sub-industry has grown up of humans - "vibe-coding clean-up specialists" - who can fix the errors in amateur vibe-coding projects.

In the hands of experienced coders, though, vibe-coding can be a substantial time-saver (saving several hours each week according to some). Using it, startups can grow faster, with fewer employees. Vibe-coding for individuals, though, can prove expensive, as the platforms charge for credits to use its services which, as we have seen, can be inefficient and frustrating. University computer science professors are having to adapt too, such as by insisting that students walk teachers through their code to make sure they actually understand how it works.

Some studies have found that, overall, vibe-coding does not really improve productivity. Aficionados, of course, just see this as teething problems in a new and emerging industry. The proof will emerge over time. Although, by then, some new breakthrough will no doubt have materialized, upending everything we thought we knew :(