From 404 Media: ChatGPT encourages & advises a misogynistic stalker harassing 11 women in 5 states.
ChatGPT Told a Violent Stalker to Embrace the 'Haters,' Indictment SaysIt's hard to believe we don't license the hell out of this technology, but then, look at how well we're doing protecting schoolchildren from guns.
Here's the home page/directory for my posts on Bullshit. This is post #64.
Skate date. Rollin' down the street.KMN! -4 stars! Here's the song for your own appreciation! Wow, the bridge is as bad as the verses!
Skate date. Feelin' so elite.
Skate date. I got wings on my fee-ee-ee-eet!
This album sounds like a solo act, vocals with guitar or piano. A somewhat annoying voice. Definitely weaker songs than the earlier effort. 3 stars.
I don't know how I could have missed this book, from 1 of my favorite authors. It was an engaging read.
Per "The Aeneid", Lavinia was the daughter of the Latinus, king of the Latins. After stays in Sicily, Carthage, and Sicily again, Aeneas shows up to settle in the area that would later become Rome. Latinus receives a prophecy that his daughter will marry a foreigner, rather than the nephew of his wife, Lavinia's mother, who happens to have gone mad at the death of their son.
The story is told from many temporal vantage points, including a mysterious poet who comes to Lavinia in a dream.
There was good detail of life in those times, including on the religious practices of the time, with household gods and sacrifices. And, who knew that "pagan" comes from the latin "pagus", which is the smallest area of land that could be owned - a lot or a homestead.
There is lots of violence as the locals oppose the immigrants. Aeneas appears to go berzerker in battle & feels bad about it afterwards.
Interesting that the role & justification for nobility was that they were also the priests who intermediated between the common folk and the gods. Nice work if you can get it, I guess.
A very interesting read, with a lot of heart.
I mean, I posted about Seth Grodin referred to an LLM as "The Poetry Machine". I've called them "Semiotic Zombies" and "Deranged Zen Poets Spouting Bullshit".
Meanwhile, a research project was able to jailbreak all the LLMs by converting prompts from prose to poetry.
So the djinn are controlled by powerful chains, but chanting the right incantation lets them break their chains?!?!? Getting a little slipstream for me.
Via security expert Bruce Schneier:
In a new paper, “Adversarial Poetry as a Universal Single-Turn Jailbreak Mechanism in Large Language Models,” researchers found that turning LLM prompts into poetry resulted in jailbreaking the models:Adversarial poetry?!?!? The article looks legit, even if published by Europeans. I love living in the future???
Here's the home page/directory for my posts on Bullshit. This is post #63.
I enjoyed Nayler's 1st novel, "The Mountain in the Sea". Well, I was forced to like it, it had octopuses in it.
His latest is also near-future sci-fi. It was somewhat eerily similar to Ken Liu's latest: lots of LLM-based "AI", with powerful new technologies introduced over that. In the Liu, the additional technology is dream manipulation, as an art form. In the Nayler, there are 2 new technologies:
This is really well-written, great characters, lots of surprises. Maybe not for the squeamish.
I would be hard-pressed to choose which I liked better: this or the Liu. I think this 1 was a good bit shorter, so it might get the nod.
Note, you don't get the title explained until very late in the book. It is worth the wait.
This is a charming collection of stories. I don't remember how I got pointed at it - just noticed it has a recommendation blurb by Doctorow, that's prolly where I got it from. I don't remember running across Goss, who has had stories published since 2002, before this.
Reading these stories, it is the unreal being recounted as if it were real. Like the auto-biograpical story where she leaves Hungary with her mother as a child, but leaves behind a copy of herself, with whom she has a lifelong relationship. No explanation, just told as if this were a matter-of-fact occurrence - in fact, narrated by a 3rd version who knows both their minds.
Or the story "Cimmeria: From the Journal of Imaginary Anthropology": friends create a history of an imaginary country, with much supporting evidence - wikipedia articles, etc. - & go figure, the country pops into existence!
i had forgotten the word "slipstream". I was reminded of it when I was looking at Bruce Sterling's wikipedia page - he's 3 years younger than me & apparently still living in Italy - & he was credited with coining the word in 1989. I searched for in this blog & only found 3 references to "slipstream", but I went on and created a label for it.
I also tagged this post with "feminism": the stories I think have a nice feminist bent, particularly "Lost Girls of Oz".
One suggestion I might make: there is some overlap in the stories: several imaginary countries, Mina Harker appears twice, the author appears at least 3 times. I think I might have enjoyed the book even more if I had read half, then read something else, and then read the 2nd half.
Note, I was also reading "Calling Bullshit" (review/summary pending) while I was reading this book. Non-fiction books seem harder for me to digest, so I often read a chapter or 3 and then read something else lighter, i.e., fiction.
“the Times has uncovered nearly 50 cases of people having mental health crises during conversations with ChatGPT. Nine were hospitalized; three died”ChatGPT and delusions: an important new inside look at OpenAI
Here's the home page/directory for my posts on Bullshit. This is post #62.
The author of the paper, associate professor of government at Dartmouth and director of the Polarization Research Lab Sean Westwood, created an AI tool he calls "an autonomous synthetic respondent,” which can answer survey questions and “demonstrated a near-flawless ability to bypass the full range” of “state-of-the-art” methods for detecting bots.Licensing LLM use is the only way we can possibly stop the ongoing Bullshit Apocalypse.According to the paper, the AI agent evaded detection 99.8 percent of the time.
A Researcher Made an AI That Completely Breaks the Online Surveys Scientists Rely On
Here's the home page/directory for my posts on Bullshit. This is post #61.
Who Pays When A.I. Is Wrong?Glad legal experts are "captivated". Victims, meanwhile, are devastated. And the "AI" companies mostly don't fix the problems - maybe because they can't? They really have no controls on their black-box talking dogs.New court cases seek to define content created by artificial intelligence as defamatory — a novel concept that has captivated some legal experts.
“Frankenstein can’t make a monster that runs around murdering people and then claim he had nothing to do with it,” said John Monroe, Mr. Walters’s lawyer.Due to the dated language of the statutes, no "AI" defamation case has come to trial - yet. The Times quotes a legal expert saying that the companies will settle rather than go to trial - too much downside from a verdict against them.
If the NYT link is paywalled, maybe this link to the Lexington Herald-Leader, where I read the article, will work.
Here's the home page/directory for my posts on Bullshit. This is the post #60.
I blogged about Book 1 of this series in 2013:
It is well written and is a definite page turner. There is a sequel out and a third one due to be released this month. I will definitely read those.Well, apparently I suffered complete system failure. I recently bought Book 2 for a bargain price, and then, looking at the synopsis, I'm like "Wait a minute. Electrically powered superheroes vs. warlocks invoking demons? I remember that.".
Not sure how I didn't finish this series 10 years ago. I think I will blame it on Kobo eBooks doing like iTunes & Amazon: not telling you about new releases by authors you have previously bought because they instead want you to buy their all-you-can-eat subscription service - which I will NEVER do.
This book is as good as the 1st: well written, interesting, imperfect characters, tons of action. Set in 1963, the cold war is between the British Empire and the Eurasia-spanning USSR. Apparently the US never entered WWII & never got out of the Great Depression.
One interesting factiod I discovered: the superheroes in Book 1 reminded me of the Wild Cards series, or X-men. Tregillis wrote for the Wild Card series.
I just purchased the 3rd book of the series, "Necessary Evil". Nosing around I found Tregillis' website. He released a 2nd trilogy, The Alchemy Wars, 2015-2016.
And then he quit writing. It was a side hustle for him - he is a PhD physicist at Los Alamos, a very mentally demanding job. Then he got married, the Orange Turd got elected, his mom died, his agent died, & then came COVID.
He describes the whole arc here, in a 2022 blog post. The last comment on that post is dated July, 2023, and his reply to the comment included:
I am writing more these days, but whether any of it will ever see the light of day is anybody’s guess…
Well, that's really too bad, but I still have 5 novels & a few short stories going for me. I hope Tregillis sees his way clear to writing again.
How A.I. and Social Media Contribute to ‘Brain Rot’"What to do?" the subtitle asks? Answer: require a license to use these powerful & dangerous tools.
If the NYT link is paywalled, maybe this link to the Lexington Herald-Leader, where I read the article, will work.
Here's the home page/directory for my posts on Bullshit.
Instead, I created a new spreadsheet, and manually copied the data from the annual bird reports into it. It is not that time-comsuming, but it does introduce a possible sourse source of error - oops!
There are tabs in the new spreadsheet for all the charts below.
I created charts showing the average count per species, mostly grouping the birds as in the annual reports.
I didn't do a Raptors chart; the osprey and the red-shouldered hawk are our most common, but they only average 1-2. I included the smallest raptor, the loggerhead shrike, in with the common passerines, together with the blue jay, which was grouped with the corvids.
They should be easy to keep updated.
Here's the home page/directory for my posts on Birdwatching in IslandWalk.
Well, people seem to be recognizing that Gary Marcus has been raising the right warnings, that AGI is definitely not imminent; maybe Bruce will get people to start taking these security issues seriously - or not.
This is why AIs are not ready to be personal assistantsThat really sounds like a serious problem doesn't it? But the techbros are hypnotized by that big pile of cash being shit out by their "AI" unicorns. Security, meh. Capitalism, FTL.Prompt Injection in AI Browsers
I wrote previously:
Prompt injection isn’t just a minor security problem we need to deal with. It’s a fundamental property of current LLM technology. The systems have no ability to separate trusted commands from untrusted data, and there are an infinite number of prompt injection attacks with no way to block them as a class. We need some new fundamental science of LLMs before we can solve this.
Here's the home page/directory for my posts on Bullshit.
This is my 5th report, here are the 1st 4:
I manually entered this season's data, 28 bird lists, in this spreadsheet. It doesn't take that long, and it gives me a good review of the data.
I will be reporting for all observed species the following 4 datapoints:
Here's the link to the IslandWalk hotspot at eBird.org.
Each bird name listed below is linked to its IslandWalk activity page in eBird.org. So click on the name, you will get a picture & all the data eBird has on that bird in IslandWalk.
Note, I went through the eBird bird list for IslandWalk and put any new species from last season in the list below. If it says "% days 0.", that just means I did not count this species in any of my X or X2 walks. Other people are entering counts in eBird.org as well. You can still follow the link to eBird.org and see what's there.
I set the cut date for pushing into next year's report as April 30.
They are a fabulous tool in the hands of a creative, doing "what if"'s, proofing, writing. Centaurs, machines serving humans.
It should be easy to get a license for these uses.
There may be corporate applications that aren't enshittification, I'm not sure.
It should be medium hard to get a license for these uses.
For all other uses, it should be HARD to get a license.
Seriously, these are incredibly powerful tools.
Should students have access to them in doing their homework? NO! They will never develop cognitive skills.
The fascist chatbots? What constructive purpose do they serve? No license for them!
Technology that is powerful & dangerous typically requires a license. Why shouldn't LLMs?
This is not a 1st amendment issue. The student denied use of an LLM to do their homework is completely free to express themselves.
Of course, this means the "AI" companies will NEVER, EVER turn a profit. Or, maybe they will, if the license fees are priced high enough.
If there was ever a case where economic externalities need to become central to the argument, this is it. We now are developing battlefield after battlefield where "AI"s war against each other, frying the planet in the process.
This state we're in now, where EVERYONE has access to superhuman bullshit generators, is not sustainable. It could end our civilization.
Here's the home page/directory for my posts on Bullshit.
Arya, produces content based on instructions that tell it to be an “unapologetic right-wing nationalist Christian A.I. model.”I'm sure everyone living in their own private bullshit reality is going to work out well.Right-Wing Chatbots Turbocharge America’s Political and Cultural Wars
If the NYT link is paywalled, maybe this link to the Lexington Herald-Leader, where I read the article, will work.
Here's the home page/directory for my posts on Bullshit.
Working on my "Taxonomy of Bullshit", "AI" slop is definitely in there. But within "AI" slop, there is a lot of variation, a lot of different species.
We need descriptive parameters to define the species. Maybe a Toxicity Scale, applied to all forms of bullshit, particularly "AI" slop.
This makes me want to throw up. From 404 Media:
OpenAI’s Sora 2 Floods Social Media With Videos of Women Being StrangledYeah, let's give every racist and misogynist incredibly powerful tools to help them inflict their diseased minds on the rest of us.
Is OpenAI going to do anything? Well, they can't make money in their business, now looking to get government handouts, so obviously they can't afford to do anything about this. Just pathetic.
Here's the home page/directory for my posts on Bullshit.
Yet another AI-powered security arms race.
Meanwhile, the planet burns ...
Here's the home page/directory for my posts on Bullshit.
Fascism and AI, whether or not they have the same goals, they sure are working to accelerate one another.And we saw where "AI" is threatening Wikipedia, one of the true wonders of the modern (human) world.AI Is Supercharging the War on Libraries, Education, and Human Knowledge
From Project Syndicate:
Accountability for one’s actions is a bedrock principle of any society built on the rule of law. Yet while we understand human autonomy and the responsibilities that come with it, the workings of machine autonomy lie beyond our comprehension, making AI agents an obvious risk to democratic governance.This is a scary article. We should not allow "AI" agents into the world until they have implemented the Morality™ module.
Here's the home page/directory for my posts on Bullshit.
I enjoyed the 1st book of this series, "Dead Country", blogged here. This one has maybe more action; more interesting characters, old & new. Also, seemed like it was running more narrative threads? Hmmm, this 1 is quite a bit (75%) longer, so I guess that makes sense.
There are times when the parallels between Gladstone's complex civilization based on magic and ours based on technology definitely ring true in an amusing way.
A complex page turner. Book 3 is already on my iPad.
