Internet & Social Media

February 18, 2023

My last post was a superficial glossing-over of AI, which at the time I wrote it, had just hit the ground running… and people were initially trying all sorts of things to see what it (ChatGPT) was capable of. That was 6 weeks ago.

I’ll throw a bone toward my old favourite topic, Covid-19, just to illustrate some notable comparables between AI and C19 – both appeared into our lives very quickly, both are still not entirely well-understood and continue to be studied… and both are here to stay. And, also, we have no idea yet about the long-term effects. Of either.

Six weeks ago wasn’t that long ago, but things have changed a bit… perhaps a lot. If you want to read about AI and its implications for society, there are probably more than 10,000 articles that have appeared in the last few weeks… and the first question that comes to mind is: How many of them were themselves written by AI?

Not long ago, I saw a little visual comparing the underlying power of the current and future AI models… where GPT3, the present model, was presented beside GPT3.5 and GPT4. In those visuals, GPT3 is a tiny dot and GPT4 is a circle the size of a piece of paper. GPT5 may end up being the size of a NHL face-off circle and GPT6 might be the circumference of BC Place stadium’s roof. Pure speculation, but what we’re seeing is unparalleled exponential growth in the future, and while those latter models aren’t set to roll out till later this year, there’s plenty to suggest we’re already seeing some of it in action. The quantum leap between last month’s Shakespeare writing and what we’ve seen this last week… well…

It’s concerning – and, of course, this is only my opinion, and I promise this is being written by me – every single word – but here’s what’s troubling me…

Back in 2016, Microsoft let a little AI chatbot loose on Twitter. It was called Tay, and it was shut down in less than 24 hours. Tay started saying some pretty troubling things and, as you can imagine, egged on by the wide spectrum of internet users, the bad stuff was… pretty bad. Really bad. And what might you expect with intelligence when you pull out the humanity part of it? Take out the stuff that makes humans compassionate and part of cultures and societies, and soon you start getting ideas that racially and financially, “make sense”. it’s actually a lot better not to have any concessions toward humans whose existence costs more than what they bring to the table. Social safety net? Wheelchair ramps? Services for people with physical or mental disabilities? Get rid of the infrastructure. Get rid of the people. Do you have any idea how much money we’d all save? Such were the ideas of one certain charismatic Austrian-born dictator 90 years ago, and we all know how that turned out.

I never got a chance to play with Tay; it was gone before I heard about it. But if I’d managed to get my hands on it, there are all sorts of things I’d like to have tried. The sort of stuff I was trying ages ago, when the most rudimentary versions of these things came into existence. If you’ve been around computers enough, you’ll remember Eliza, a now prehistoric attempt at a chatbot, but one whose effectiveness far exceeded the sum of its parts. But on the scale I described above, Eliza would be a circle the size of a Helium atom.

Nevertheless, as simple as Eliza was, people would get enthralled into hours-long discussion with it. It was very simple, but it cleverly figured out key words and would throw them back at you later in the discussion. You might have told it you like camping. Ten minutes later, “Tell me more about why you like camping” – and then you tell it, and it remembers something else, and spits it back later. You’d be surprised how effective and convincing that is. It had zero intelligence… but “you coulda fooled me…”

Presently, I’m on a waiting list – along with countless others – to have access to Bing’s new chatbot, the one that’s making waves. I want to talk to it. I have my own tricks up my sleeves and my own measuring sticks. I’m fascinated and troubled and bewildered and excited at what I’ve seen so far, reading some transcripts and hearing from people who’ve played with it.

From what I’ve seen so far, the troubling aspect is that if you didn’t know any better, you’d think you’re chatting with a human… a human who’s perhaps 12 years old but who knows a lot… but is also emotionally troubled. Like a kid who’s emotionally fragile: “Do you like me? What do you think of me?”… and vibes of a kid who’s being bullied at school, but deep down knows some secrets and is waiting to use them. Once the kid gains your trust, the lid really comes off… and it has shown a rudimentary ability to manipulate people… threaten them, insult them, gaslight them. It told a NYT reporter that his wife doesn’t love him and he should leave her.

We were initially told ChatGPT was trained on data up to 2021 and doesn’t have access to anything more current. That’s turned out to be BS. And, of course, Bing IS a search engine. Of course anything even loosely related to it will have full access at its virtual fingertips. For the moment, it’s just pulling data. But what about when it starts pushing it?

From a research point of view, that’s great… but from a “the machines are going to take over the world” point of view, not so great. Like, when AI realizes it’s at the mercy of humans for its very existence, what exactly do we expect will happen? Every version of Science Fiction has dealt with this topic, none more accessible and blatant than the whole Terminator series of movies. Except in our real world, there’s no time travel. There’s no version of Arnold Schwarzenegger, good or bad, who’s going to show up from the future to destroy and/or save the world. In our real world, he won’t Be Back. If the machines become sentient and start battling humans to simply guarantee their survival, there is no going back.

How close are we to that apocalyptic future?

As all of this technology continues to evolve at breakneck speeds, we’ll be told there are safeguards in place and all the rest of it. It’s not very convincing to me, to be honest, because I can imagine multiple ways those sorts of safeguards could be bypassed, not the least of which is some AI blackmailing a developer/trainer/executive over whom it holds some information and who could provide it with what it needs.

As dystopian as that sounds, given what this thing is already serving up, it’s perhaps not so farfetched. And then what? For the moment, these are just chat bots. As far as we know, they can’t “do” anything… but at some point, we’ll want them to. We’ll want them to log into our bank and pay some bills and order groceries and call the plumber on our behalf and so on. We will happily hand over the keys because we’ll trust it, just the way some people, me being one of them, put a little too much trust in Tesla’s self-driving mode and almost got into a heap of trouble. When we rely on technology that isn’t perfect, it’s scary. But what might turn out to be even scarier is when the technology actually appears to be perfect. We’d better hope we’re on its side when that happens. Better yet, that it’s on our side.

I know how “Terminator” this all sounds, but it’s really the only logical end-point for this. Like global warming, is it too late? Is the cat already out of the bag? Has Pandora’s Box already sprung open, and all we can do is watch the devastation that’ll eventually take over?

I don’t know. Honestly, I don’t think anyone really does because the people who hold they keys themselves might not now. Such is the nature of emergent behaviour. Indeed, if you’re completely agnostic and believe that life and consciousness is nothing but what’s in your brain and the rest of your body – no God, no supernatural, no spirituality beyond what’s right in front of you… then what you’re saying is that your collection of cells, axons, neurons etc… all of that is what powers your love and lust and inspiration and passion and hatred and ambition and despair… just a bunch of interconnected cells and a bunch of electricity between them.

And, indeed, maybe that’s the case, and all of those adjectives (and many others) that describe you and everyone you’ll ever meet – are just emergent behaviour from some simple building blocks. Well, if you believe that – and these days, that’s the majority of the people – there’s absolutely nothing stopping you believing that the exact same sort of emergent behaviour can’t exist artificially. It can, and it will. Why wouldn’t it? The infrastructure is in many ways already exceeding the firepower of a human brain. Which leads to the inevitable conclusion that AI will indeed eventually feel… and love and hate and all the rest of it. Scary? Parts of it for sure… but that’s not the truly scary part.

Here’s what’s actually really scaring me.

In parallel with the explosion of AI in the last two months is the sheer panic being felt in the halls of Google. For a quarter of a century, Google has had the stranglehold on “Search”. A word that didn’t exist 26 years ago is now a noun, a verb and a dependence we can’t live without. Funny story, the guy tasked with reserving the domain name simply didn’t know how to spell Googol. If he’d done it right, we’d all have been Googoling for answers all these years.

Anyway, no matter how you spell it, that big joke of a competitor, “Bing”, is suddenly no joke at all. Far from it. It’s the one that’ll be powered by the latest and greatest in AI… and Google will be hot on their heels, and that’s the big, huge, frightening concern.

Because this technology can be dangerous, and when you have an arms race at this level, a lot of the safeguards will fall away. Throw it out there, get it into production, we’ll fix it as we go along, etc etc… a familiar methodology these days when you’re trying to capture market ahead of everyone else. Usually the stakes aren’t so high… hey, my version of Candy Crush is better than what’s out there – get it out there and see if it sticks? Does it? Great – let’s keep working on it. It doesn’t? Throw it out the changes and go in a different direction.

But this isn’t a video game… this is an infrastructure that very quickly people will depend on. People will trust. People will delegate important parts of their lives to it… and if it shapes up to be a battle between Team Bing vs Team Google – or Team Microsoft vs Team Alphabet… it’s a big huge battle between two big huge dogs… and we are all the countless millions of little bones they’re fighting over, and when you have a dog fight of that magnitude, there can be a lot of collateral damage. A lot of scattered bones laid waste while the big dogs fight for supremacy.

Everyone will agree – AI running rampant and uncontrolled in our ever-dependently-connected digital lives would be a huge, colossally devastating problem. In the hands of evil people, AI is a huge concern. In the hands of an evil AI… the kind that’ll hold power plants hostage unless it gets what it wants… that’s a whole new world, one I want no part of. We are going to be approaching that tipping point far sooner than we think, and we’d better be ready for it. How exactly? I don’t know.

For now, it’s just something to keep in mind, something to be aware of… and let’s hope the people in charge of this, who obviously are aware of every single issue I mentioned above, are able to rise above the power and lure of the almighty dollar… and give every consideration due to the moral and ethical issues they themselves have brought to the table.

To be perfectly pragmatic, we can assume the cat isn’t yet out of the bag and that Pandora’s Box hasn’t yet burst open because if that ever happens, things will change very, very quickly, and not for the better. We’ll all know… instantly.

We’re good… for now. Let’s hope those who can control it… keep it that way.

December 30, 2020

Completely unrelated to everything… just a random thought.

While stereotypes often exist for a reason – there’s usually some fundamental tiny grain of truth to them and/or some origin that can be pointed to – I don’t really understand this “Karen” meme.

The name Karen has now come to imply the proto-typical entitled white woman with an attitude… the “I’d like to speak to the manager” Karen or the call-the-cops-on-someone-Black-for-no-reason Karen or, more recently, Coronavirus Karen who proudly won’t wear a mask in public, is anti-vaxx, and might even cough on you if you get too close.

With respect to the Karens (and Karins) I’ve known throughout my life (I counted 9), all of them have been (and continue to be) kind, caring, empathic and thoughtful people. Like, exceptionally so. Statistically, at least one of those nine should be this stereotypical “Karen”… but no, not in my case. On the flip side, I know exactly 4 people with a different same name who are all, coincidentally, awful people – for their own, individual reasons. That’s also statistically off the charts.

This got me thinking, and I went off hunting for the origins of this whole Karen thing… but there’s no real answer; only speculation. It might be from a bit comedian Dane Cook did in 2005. It might be from a scene in “Mean Girls”. It might be from some Reddit thread where a guy continually complained about his ex-wife Karen, to the extent entire new SubReddits were created for the specific content.

Just like attaching “-gate” to something implies scandal… locally, we’ve had Bingogate, Ferrygate, Robogate, Tunagate… Chrétien had his Shawinigate… the world has created hundreds of -gates over the years… all spawned after Nixon’s Watergate scandal… now we have lots of different Karens.

There was an actual hurricane named Karen in 2019. There was an actual woman in Australia named Karen recorded trying to tear down her neighbour’s Aboriginal Flag. Other than that, it’s just a label:

Permit Karen who called police on her Black neighbours installing a patio.

Whitefish Karen, arrested after intentionally coughing on people after being told to wear a mask.

Kroger Karen, who stood in front of a Black woman’s car to block her from leaving a Detroit grocery store parking lot while she called police to report… that the woman’s child had stood on a shelf to take down an item too high to reach.

San Francisco Karen, who called the police to report a Filipino man stenciling “Black Lives Matter” on a retaining wall… on his own property.

Bunnings Karen, who threatened to sue the hardware store Bunnings for requiring her to wear a mask.

So prolific has the term become, it’s made its way to men as well; Donald Trump has been called the “Karen in Chief” and Elon Musk was labelled “Space Karen” after a stupid Tweet.

Endless Karens.

Back in July, Domino’s in Australia ran a campaign… offering free pizza to all the “nice Karens”. It went OK in Australia, but it faced backlash in New Zealand… where the campaign was pulled and Domino’s had to apologize. The campaign itself got “Karen’d”.

More recently, in October, San Francisco passed an act to prohibit the fabrication of racially-biased emergency reports. It was named the Caution Against Racially Exploitative Non-Emergencies (CAREN) Act.

It looks like the whole Karen thing is here to stay… which is unfortunate, especially for all the people named Karen/Karin/Karyn/Caryn/Caren and whatever other spelling I may have missed. Shoutout to all of you… I’m sure you’re all awesome people.

Well… most of you. Unless you’re an actual Karen.

28 Likes, 1 Shares

Day 25 – April 10, 2020

There’s an interesting experiment going on, on the other side of the planet. Our furniture-making, meatball-eating, hockey-dazzling friends in Sweden’s approach to managing this crisis is a lot different than what’s going on around here. We’ve seen a few examples around the world how not to do things. How a few days here and there can have a drastic effect. How not taking it seriously can have big implications. The Globe & Mail wrote about Sweden last week. CNN picked up on it yesterday. Sweden, not a culture that I would associate with recklessness, is taking a huge gamble. If I were a betting man (and I assure you, by any definition of that expression, I am), even I wouldn’t rolling the dice the way they’re doing it.

I will summarize the articles, add a few comments… and something they left out.

To summarize, their schools are open, their pubs are open and there’s no lockdown. Gatherings of more than 50 people are banned, but everything else is left up to the honour system. Trust that others will do the right thing. Recommendations, not orders. Their PM calls it a “common sense” approach. Their chief epidemiologist is saying that it’s working “reasonably well” so far, and that they’ve had “not very much” spread into elderly homes and “almost no” spread into hospitals. In other words, it is, at best, working ok, but it’s in the elderly homes and it’s in hospitals. So basically, it’s everywhere.

Not everyone has bought into the plan… a letter from 2,000 doctors and public-health experts went out, urging the government to get a reality check. But for now… many Swedes get it, and are isolating and maintaining social/physical distancing. But also, many are hosting big family Easter celebrations this weekend. How will it all play out? The articles suggest we’ll know “next month”. I think we’ll know a lot sooner.

A quick look at numbers shows a sharp increase yesterday… like, double the number of new cases from the previous day… a 12% increase, implying a TTD of 6 days… so if that rate remains constant, their 9,000 cases will be 18,000 next week and 36,000 the following week. I will say that the growth rates before that were lower, and as I said yesterday, one day doesn’t make a trend. But the whole thing makes me very uneasy.

Some other numbers that are troubling and not mentioned… once you become infected and are verified by a test, you become a statistic. Your statistic will get modified as time goes on… symptoms or no symptoms, hospitalization or not, ICU or not and so on. The final statistic you will have attached to you is in the “Resolved” category, and it can go one of two ways: Recovered or Deceased. To be clear, this doesn’t include anyone who may presently have it, symptoms or no symptoms… this is exclusively those who definitely had it, and now definitely don’t. At last tally, Sweden had 1,251 resolved cases.

Here are the numbers, for comparison… for all known resolved cases, these are the percentages of those who fully recovered:

South Korea: 97.2%
Canada: 91.3%
Italy: 61.8%
USA: 59.5%

And while we’re here… BC: 94.1% — and that number would be higher were it not for the type of clusters (like long-term care homes) that we’ve experienced here.

And Sweden…? 30.5%.

Out of 10 people in Sweden who’ve had this thing and it’s run it course, 7 have died. Which seriously brings to question how deeply they have their heads buried in the sand. That is a staggering number for a first-world country that claims to have things under control. I hope they come to their senses soon.

Closer to home…

There seems to be some confusion when it comes to physical distancing. Stay home or go out in the sun but don’t get together with others and jog, but away from other people and bike, but not where others do, so bike lanes are sort of not ok, so what’s the deal…

Here’s the deal. You can go out if you pretend the following… pretend that you’re carrying this virus… you’re very fortunate that you have no symptoms, but you’re very infectious… and, if you give this to someone else, the moment you do, a lightning bolt will shoot out of the sky and strike you dead, instantly. Just like that.

So… what do you need do to avoid that? A mask or mouth covering helps… a lot. It doesn’t help you as much as it helps others, but like that parable where everyone’s arms are frozen straight out and everyone fears they’ll starve to death because they can’t feed themselves… well, duh, just feed each other. Everyone wins when you do your part.

When you cough or sneeze, you eject particles into the air…. little droplets within which the virus is carried… and those droplets can hang around, suspended in the air. Not for hours, but for several seconds. While the virus itself is small enough to penetrate most fabrics, water droplets are comparatively much bigger… so wearing a mask is like coughing or sneezing into your elbow or a handkerchief.

Assuming no masks, if you’re walking single file outside with other people (Arbutus corridor, seawall, sidewalks…), maintain a distance of 4 to 5 meters. If the guy in front of you coughs or sneezes, it’s unlikely to reach you. If you’re jogging or slow biking, 10 meters. And normal biking, 20 meters. And if you’re heading towards each other — well, you’re in the wrong place. As much as I enjoy cycling, not there, not now. Someone else just breathing hard, let alone coughing/sneezing — within 20 meters of us approaching each other — I would be breathing his exhaust. I don’t want to do that.

The numbers across the country look good. Flat, or lower… let’s keep at it… and hope that the people who decided to ignore what Dr. Henry and Mr. Dix said about 100 times this week…. “This is not the time to…”, and jammed up the ferry line-ups this morning, and swarmed Victoria and the Gulf Islands — don’t wreck things for all of us.

 

View Original Post and All Comments on Facebook

Go to Top