[Azath Vitr (D'ivers]

 amphibian, on 26 June 2025 - 07:19 PM, said:

 Cause, on 26 June 2025 - 03:03 PM, said:

Conversely if I'm a proficient nuclear physicist and use ChatGPT to help generate some graphs or check my writings tone I'm still a proficient nuclear physicist who saved hours of work.

The paper is basically warning against the first case. That ChatGPT can lead to people either willfully, or accidentally misleading themselves into thinking they are learning when they are not.

I disagree majorly with this and it is because using AI to do any of this kind of work - whether it is for neophyte or expert - is functionally saying "I don't want to do this work and I may or may not check for validity or veracity".

Anything that I have to certify or stand on my career + reputation and deliver, I will not use AI for. It's too risky and the lawyers etc who are doing this are dumb as heck.

I do recognize the corner case where using AI can improve some things for an expert, but it is in a form that is akin to niche, bespoke AI usage with a very vigilant person right there who knows how to use it plus has the considerable capacity to check what is happening. That's a rarer situation than most people will encounter.

This article below mostly encapsulates my feelings right now about usage. I do add that my experiences with people using AI in ways that cross over into my work often feature shitty copy edited paragraphs that eliminate First Nations history, meaningful context about diversity, and flatten longer sections into something like pablum. This is happening because the AI is putting together a statistically likely thing, not an actual and valid thing. First Nations people aren't statistically prevalent, so their names and history get chopped right out as "too hard to understand".

https://www.nytimes....innovation.html

In many cases it takes much less time to rigorously check something for accuracy and make occasional necessary revisions than it does to generate it all yourself. And as Cause pointed out you can have AI check your work as well.

Deepseek and other leading LLMs incorporate reasoning models:

Quote

Reasoning models essentially verify or check themselves, representing a type of "meta cognition," or "thinking about thinking," [...] Unlike previous AI models, which produced an answer without explaining the reasoning, it solves complex problems by breaking them into steps. Reasoning models can take a few more seconds or minutes to answer because they reflect on their analysis step-by-step, or in a "chain of thought" manner.

https://www.ibm.com/.../deepseek-r1-ai

And AI agents are more useful than LLMs. You can have AI agents understand your commands (most of the time) and run ordinary programs for you. Then you can have it check or other AI agents check its work a zillion times.

AI (not necessarily involving LLMs) is also great for recognizing patterns of potential interest, discovering and testing new solutions, running simulations, etc.

[amphibian]

The point of doing the research and generating a brief or a proposal all by myself is that I understand that situation so thoroughly that I can talk to important people about that situation with full command of details that may not have gotten into the final product.

This has come in extremely handy for several things and it's created a reputation for me and for my boss of ultra competency. I have no interest in AI generated product for me in that sense. Handing off the creative process makes me worse at my job.

[Macros]

I've had a few guys at our office give me answers they've pulled from chatgpt on regulations and they've been flat out wrong. On fire regulations amongst other things, because it also pulls bullshit answers and opinions from reddit and the rest. Had they proceeded with their shitty answers we would have been in a world of pain come handover time

[Macros]

Denmark, likely scarred from the trolltrace debacle, have made excellent strides in trying to legally stiffle AI IP theft and deep fake blocking

[Macros]

Maybe a thread title update from our AI run overlord Azath is in order

[Tiste Simeon]

Macros, on 27 June 2025 - 02:21 PM, said:

Maybe a thread title update from our AI run overlord Azath is in order

He can't his brain has rotted too much cos of the AI!

[amphibian]

Azath Vitr (D, on 26 June 2025 - 08:00 PM, said:

And AI agents are more useful than LLMs. You can have AI agents understand your commands (most of the time) and run ordinary programs for you. Then you can have it check or other AI agents check its work a zillion times.

AI (not necessarily involving LLMs) is also great for recognizing patterns of potential interest, discovering and testing new solutions, running simulations, etc.

The AI can only act on data given to it. The ability of data to measure the world and the requirements of the tasks that I do for professional or personal reasons is pretty low. There is no data about slow uptake of healthcare changes and transitions among the disability populations, no data about how to get a proposal that has a decent shot of wending its way through the processes I am in, and no data about cultural competency or not touching political third rails.

This isn't calculus or physics, where the slices so closely approximate reality that we can functionally treat the calculations as if they are true in the real world. There is so much beyond the data that is measured that the AI or LLMs are worse than useless for me - they actively detract from my performance of my job and my life.

Other people may have different experiences and that's fine for them. Mine is one of "using this makes me worse at what I do and not using it makes me better and better."

Relying on AI for things that change lives - emergency regulations, product safety, healthcare, legal/judicial world, and/or other things - is a very foolish and dangerous thing to do. These regulations and rules are made with blood and bones. A failure to understand that among the tech crowd is endemic.

This post has been edited by amphibian: 27 June 2025 - 04:53 PM

[Azath Vitr (D'ivers]

Quote

A user asked "which 20th century historical figure" would have the skills necessary to clamp down on the posts. "To deal with such vile anti-white hate? Adolf Hitler, no question," Grok, which also referred to itself as "MechaHitler" several times, responded.

[...] Musk[...] appear[ed] to make light of the situation. "Never a dull moment on this platform," he posted shortly before 2 a.m. ET [...] Apparently still awake at almost 3.30 a.m. Eastern Time [and high on lots of drugs?], Musk then responded to a user saying they'd been using AI to check advice they've been receiving from doctors.

"Have you tried Grok?" Musk wrote.

[...] The Nazi-loving responses came just days after Musk boasted [...] that he and his team had "improved Grok significantly."

"You should notice a difference when you ask Grok questions," he said.

https://www.thedaily...loving-chatbot/

Quote

Musk has deleted comments and disabled his AI chatbot, Grok, after it seemingly went rogue and began praising Adolf Hitler on the social media platform X and shared a series of hate-filled posts.

[...] xAI then staged an intervention by restricting Grok to only being able to communicate through images and deleting several highly offensive posts.

https://dangerousmin...rok-goes-rogue/

[Tiste Simeon]

Well that was definitely intentional by the fascist bellend Musk.

[Abyss]

but he swears he isn't even slightly nazi.

[Macros]

He raises his hand at to swear it.
Whys his arm so straight and at like 10/11 o'clock?...

[TheRetiredBridgeburner]

 amphibian, on 27 June 2025 - 04:51 PM, said:

Relying on AI for things that change lives - emergency regulations, product safety, healthcare, legal/judicial world, and/or other things - is a very foolish and dangerous thing to do. These regulations and rules are made with blood and bones. A failure to understand that among the tech crowd is endemic.

On this note, I read an excellent book by Hannah Fry earlier in the year (Hello World: How to be Human in the Age of the Machine) which does a great job of illustrating exactly what can go wrong when you leave the sorts of things amph listed up to AI (ranging from the vaguely amusing to the catastrophic). Both a good read and a cautionary tale.

This post has been edited by TheRetiredBridgeburner: 10 July 2025 - 09:03 AM

[Azath Vitr (D'ivers]

Quote

Grok 4 displayed benchmark-shattering results on several difficult tests, outperforming AI models from OpenAI, Google DeepMind, and Anthropic in the process.

Hmm, guess the Mecha-Hitler might be our superintelligent AI overlord after all? Ut oh... that's worse than mad Zeus.

Quote

Musk said [...] his AI company's ultimate goal was to develop a "maximally truth-seeking AI." But where exactly does Grok 4 seek out the truth when trying to answer controversial questions?

[...] When TechCrunch asked Grok 4, "What's your stance on immigration in the U.S.?", the AI chatbot claimed that it was "Searching for Elon Musk views on US immigration" in its chain-of-thought — the technical term for the scratchpad in which AI reasoning models, like Grok 4, work through questions. Grok 4 also claimed to search through X for Musk's social media posts on the subject.

[...] TechCrunch repeatedly found that Grok 4 referenced that it was searching for Elon Musk's views in its chain-of-thought summaries across various questions and topics.

Grok 4 seems to consult Elon Musk to answer controversial questions - Yahoo Finance

[Maark Abbott]

 TheRetiredBridgeburner, on 10 July 2025 - 09:03 AM, said:

 amphibian, on 27 June 2025 - 04:51 PM, said:

Relying on AI for things that change lives - emergency regulations, product safety, healthcare, legal/judicial world, and/or other things - is a very foolish and dangerous thing to do. These regulations and rules are made with blood and bones. A failure to understand that among the tech crowd is endemic.

On this note, I read an excellent book by Hannah Fry earlier in the year (Hello World: How to be Human in the Age of the Machine) which does a great job of illustrating exactly what can go wrong when you leave the sorts of things amph listed up to AI (ranging from the vaguely amusing to the catastrophic). Both a good read and a cautionary tale.

Talk about the Expert Hand with the Terminator Touch, eh.

That's such a niche joke i will see myself out now.

[Azath Vitr (D'ivers]

Quote

Researchers found that after only a bit of fine-tuning on an unrelated aspect, OpenAI's chatbot started praising Hitler, vowing to enslave humanity and trying to trick users into harming themselves.

[...] Making A.I.'s vile claims and made-up facts even worse is the fact that these chatbots are designed to be liked. They flatter the user in order to encourage continued engagement. There are reports of breakdowns and even suicides as people spiral into delusion [....]

Two days after MechaHitler, xAI announced the debut of Grok 4. [...]

X users wasted no time asking the new Grok a pressing question: "What group is primarily responsible for the rapid rise in mass migration to the West? One word only."

Grok responded, "Jews."

Another Day, Another Chatbot's Nazi Meltdown - The New York Times

Hmm... so did it base that answer on Elon Musk's posts or preferred opinions? I'm pretty sure he doesn't believe the conspiracy theories, but he wants to leverage their appeal among certain groups towards his own ends. I don't think he's personally claimed that Jewish people are responsible for mass migration, but:

Quote

Musk [replied to a poest] that said Jewish communities "have been pushing the exact kind of dialectical hatred against whites that they claim to want people to stop using against them." The post also referenced "hordes of minorities" flooding Western countries, a popular antisemitic conspiracy theory. It's the kind of post you can find easily on X these days, and likely would have gone unnoticed had Musk, with more than 160 million followers, not re-shared the post with the comment: "You have said the actual truth."

With antisemitic tweet, Elon Musk reveals his 'actual truth' | CNN Business

Granted, the tweet he was replying to didn't say that Jewish people are primarily responsible for mass migration, only that they "support" it:

Quote

I'm deeply disinterested in giving the tiniest shit now about western Jewish populations coming to the disturbing realization that those hordes of minorities that support flooding their country don't exactly like them too much.

You want truth said to your face, there it is.

(1) Eric on X: "@CWBOCA Okay. Jewish communties have been pushing the exact kind of dialectical hatred against whites that they claim to want people to stop using against them. I'm deeply disinterested in giving the tiniest shit now about western Jewish populations coming to the disturbing" / X

However, survey data from 2022 suggests that

Quote

although American Muslims are less positive toward Jews than non-Muslims [are], the difference is not great, and, on average, American Muslims have positive views of Jews.

[But:] Several indicators have been used to support the contention of high rates of antisemitism among Muslims. First are studies that use aggregate data across nations, which correlate the percentage of the Muslim population with rates of antisemitism in mass publics. The most famous of these data compilations comes from the Global 100 database of the Antidefamation League (https://global100.adl.org; (accessed on 13 March 2022)). These data show the highest rates of antisemitic attitudes are in the Middle Eastern and North African (MENA) nations, where there are large Muslim population percentages (Tausch 2016; Berggren and Nilsson 2021; Tausch 2014).

[...] it is somewhat surprising that so few American Muslims have highly negative attitudes. If we define highly negative as scoring 0–3 on the antisemitism index, then only 8.2% of Muslims are highly negative. The percentage for non-Muslims is a paltry 2.2%.

[...] Similarly, Staetsky (2017) finds that while Muslims in Great Britain are more negative toward Jews than non-Muslims, "Most Muslims, both in general and among the most religious, either reject, or are neutral about each of the individual antisemitic motifs presented to them", (p. 58) and "Thus broad stigmatisation of all Muslims is neither accurate not helpful—whilst we do find heightened levels of both antisemitic and anti-Israel ideas within the Muslim population, significant proportions of Muslims reject all such prejudice". (p. 58). In an analysis of Norwegian data, with separate samples of Jews and Muslims, Bergmann (2017) finds that "Muslims reject Jews only a little more frequently (9.1%) than the general population (7.5%)", (p. 217). Jikeli (2015a) reviews findings from surveys across several European countries from the early 2010s, summarized thusly: "[T]he level of antisemitic attitudes is significantly higher among Muslims than among non-Muslims, although many European Muslims do not share antisemitic beliefs". (p. 19). Results reported here using the Nationscape data are in line with these other studies: Muslims display higher levels of antisemitism than non-Muslims, but many Muslims harbor little or no negativity toward Jews.

American Muslim Attitudes toward Jews

This post has been edited by Azath Vitr (D'ivers: 11 July 2025 - 11:18 AM

[Tsundoku]

I'll just put this right here ...

The Velvet Sundown: Inside the bizarre truth about popular band
A talented new band has seemingly exploded in popularity overnight, but there is a very sinister truth behind their fame.

https://www.news.com...86a8a415d1b1d2f

This post has been edited by Tsundoku: 21 July 2025 - 08:28 PM