• Instead of forming hypotheses, users asked the AI for ideas.
• Instead of validating sources, they assumed the AI had already done so.
• Instead of assessing multiple perspectives, they integrated and edited the AI’s summary and moved on.
This isn’t hypothetical. This is happening now, in real-world workflows.
"""
Amen, and OSINT is hardly unique in this respect.
And implicitly related, philosophically:
It is, but it adds disingenuous apologetic.
Not wishing to pick on this particular author, or even this particular topic, but it follows a clear pattern that you can find everywhere in tech journalism:
Some really bad thing X is happening. Everyone knows X is happening.
There is evidence X is happening, But I am *not* arguing against X
because that would brand me a Luddite/outsider/naysayer.... and we
all know a LOT of money and influence (including my own salary)
rests on nobody talking about X.
"Smartphones and social media are an essential part of modern life today... but"
That always sounds like those people who say "I'm not a racist, but..."
Sure, we get it, there's a lot of money and powerful people riding on "AI". Why water down your message of genuine concern?
OSINT only exists because of internet capabilities and google search - ie someone had to learn how to use those new tools just a few years ago and apply critical thinking
AI tools and models are rapidly evolving and more in depth capabilities appearing in the models, all this means the tools are hardly set in stone and the workflows will evolve with them - it’s still up to human oversight to evolve with the tools - the skills of human overseeing AI is something that will develop too
Offline version: https://www.kiwix.org
Is it unfair to make such demands for the inclusion of 101-level stuff in non-programming content, or is it unfair to give IT topics a pass? Which approach fosters a community of winners and which one does the opposite? I'm confident that you can work it out.
[1] https://www.tandfonline.com/doi/full/10.1080/16161262.2023.2...
- "Final FIXED & WORKING drawing.html" (it wasn't working at all)
- "Full, Clean, Working Version (save as drawing.html)" (not working at all)
- "Tested and works perfectly with: Chrome / Safari / Firefox" (not working at all)
- "Working Drawing Canvas (Vanilla HTML/JS — Save this as index.html)" (not working at all)
- "It Just Works™" (not working at all)
The last one was so obnoxious I moved over to Claude (3.5 Sonnet) and it knocked it out in 3-5 prompts.
If anything you’ve just pointed out how over reliance on AI is weakening your ability to search for relevant information
The "Pray Mr. Babbage..." anecdote comes to mind: https://www.azquotes.com/quote/14183
To pick an extreme example, programmers using a strongly typed language might not bother manually checking for potential type errors in their code and leave it to the type checker to catch them. If the type checker turns out to be buggy then their code may fail in production due to their sloppiness. However, we expect the code to eventually be free of type errors to a superhuman extent because they are using a tool that is strong to cover their personal weaknesses.
AI isn't as provably correct as type checkers, but they're pretty good at critical thinking (superhuman compared to the average HN argument) and human analysts must also routinely leave a trail of mistakes in their wake. The real question is what influence the AI has on the quality and I don't see why the assumption is that it is negative. It might well be; but the article doesn't seem to go into that in any depth.
Even if my prompt was low-quality, it doesn't matter. It's confidently stating that what it produced was both tested and working. I personally understand that's not true, but of all the safety guards they should be putting in place, not lying should be near the top of the list.
These tools are brand new and have proven kinks (hallucinations, for example). But instead of being, rightly, in my view, skeptical, the majority of people completely buy into the hype and already have full automation bias when it comes to these tools. They blindly trust the output, and merrily push forth AI generated, incorrect garbage that they themselves have no expertise or ability to evaluate. It's like everyone is itching to buy a bridge.
In some sense, I suppose it's only natural. Much of the modern economy sustains itself on little more than hype and snake oil anyway, so I guess it's par for the course. Still, it's left me a bit incredulous, particularly when people I thought were smart and capable of being critical seemingly adopt this nonsense without batting an eye. Worse, they all hype it up even further. Makes me feel like the whole LLM business is some kind of ponzi scheme given how willingly users will schill for these products for nothing.
I genuinely hope if you're a professional intelligence analyst it doesn't take a trained eye to distinguish Paris from Belgium. Genuinely every day there's articles like this. The post about college students at elite universities who can't read, tariff policy by random number generator, programmers who struggle to solve first semester CS problems, intelligence analysts who can't do something you can do if you play Geoguessr as a hobby. Are we just getting dumber every year? It feels like we're falling off a cliff over the last decade or so.
Like, the entire article boils down to "verify information and use critical thinking", you'd think someone working in intelligence and law enforcement which this author trains knows this when they get hired?
[0] https://samrawal.substack.com/p/the-human-ai-reasoning-shunt
Much good (and bad) sci-fi was written about this. In it, usually this leads to some massive conflict that forces humans to admit machines as equals or superiors.
If we do develop super-intelligence or consciousness in machines, I wonder how that will all go in reality.
For example, suppose a person shares a photo online, and your intelligence objective is to find where they are. In that case, you might use GPS coordinates in the photo metadata or a famous landmark visible in the image to achieve your goal.
This is just for others who are curious.
I doubt anyone can do it perfectly every time, it requires a posthuman level of objectivity and high level of information quality that hardly ever exists.
Assuming these professionals were great critical thinkers until the AI came along and changed that is a big stretch.
In my experience, the people who outsource their thinking to LLMs are the same people who outsourced their thinking to podcasts, news articles, Reddit posts, Twitter rants, TikTok videos, and other such sources. LLMs just came along and offered them opinions on demand that they could confidently repeat.
> The scary part is that many users still believed they were thinking critically, because GenAI made them feel smart
I don’t see much difference between this and someone who devours TikTok videos on a subject until they feel like an expert. Same pattern, different sources. The people who outsource their thinking and collect opinions they want to hear just have an easier way to skip straight to the conclusions they want now.
A con man often uses the illusion of confidence to gain trust, though that's not the only way. The reverse also works: gain their trust by seeming unconfident and incapable, and thus easily taken advantage of.
This seems contradictory to me. I suspect most experienced professionals start with the premise that the LLM is untrustworthy due to its nature. If they didn't research the tool and its limitations, that's lazy. At some point, they stopped believing in this limitation and offloaded more of their thinking to it. Why did they stop? I can't think of a single reason other than being lazy. I don't accept the premise that it's because the tool responded quickly, confidently, and clearly. It did that the first 100 times they used it when they were probably still skeptical.
Am I missing something?
That doesn't actually work though. Try to set it up and it just fails to download.
People also seem to be losing their ability to detect satire.
I'm concerned GenAI will lower creative standards too, that people will be fine with the sound of suno, or the look of Dall-E. How then would the arts evolve?
Most people don't actually critically evaluate LLMs for what they are, and actually buy into the hype that it's a super-intelligence.
Also, I won't remotely claim that it's the case here, but external pressures regularly push people into do the wrong thing. It doesn't mean anyone is blameless, but ignoring those pressures or the right (or wrong) stimuli makes it a lot harder to actually deal with situations like this.
Fair point. My intention isn't to be absolute, though. Even in a relative sense, I can't imagine a scenario where some level of laziness didn't contribute to the problem, even in the presence of external factors.
It seems like the author was eliminating laziness with their statement and instead putting the primary force on the LLM being "confident." This is what I'm pushing back against.
That's the main problem, it's becoming the standard in everything.
For example, I am learning Rust, for quite awhile now. While AI has been very helpful in lowering the bar to /begin/ learning Rust, it's making it slower to achieve a working competence with it, because I always seem reliant on the LLM to do the thinking. I think I will have to turn off all the AI and struggle struggle struggle, until I don't, just like the old days.
I don't really like the way LLMs code. I like coding. So I mostly do that myself.
However I find it enormously useful to be able to ask an LLM questions. You know the sort of question you need to ask to build an intuition for something? Where it's not a clear problem answer type question you could just Google. It's the sort of thing where you'd traditionally have to go hunt down a human being and ask them questions? LLMs are great at that. Like if I want to ask, what's the point of something? An LLM can give me a much better idea than reading its Wikipedia page.
This sort of personalized learning experience that LLMs offer, your own private tutor (rather than some junior developer you're managing) is why all the schools that sit kids down with an LLM for two hours a day are crushing it on test scores.
It makes sense if you think about it. LLMs are superhuman geniuses in the sense of knowing everything. So use them for their knowledge. But knowing everything is distracting for them and, for performance reasons, LLMs tend to do much less thinking than you do. So any work where effort and focus is what counts the most, you're better off doing that yourself, for now.
He's talking specifically about OSINT analysts. Are you saying these people were outsourcing their thinking to podcasts, etc. before AI came along? I have not heard anyone make that claim before.
The real power I've found is using it as a tutor for my specific situation. "How do list comprehensions work in Python?" "When would I use a list comprehension?" "What are the performance implications?" Being able to see the answers to these with reference to the code on my screen and in my brain is incredibly useful. It's far easier to relate to the business logic I care about than class Foo and method Bar.
Regarding retention, LLMs still doesn't hold a candle to properly studying the problem with (well-written) documentation or educational materials. The responsiveness however makes it a close second for overall utility.
ETA: This is regarding coding problems specifically. I've found LLMs fall apart pretty fast on other fields. I was poking at some astrophysics stuff and the answers were nonsensical from the jump.
> You upload a protest photo into a tool like Gemini and ask, “Where was this taken?”
> It spits out a convincing response: “Paris, near Place de la République.” ...
> But a trained eye would notice the signage is Belgian. The license plates are off.
> The architecture doesn’t match. You trusted the AI and missed the location by a country.
Okay. So let's say we proceed with the recommendation in the article and interrogate the GenAI tool. "You said the photo was taken in Paris near Place de la République. What clues did you use to decide this?" Say the AI replies, "The signage in the photo appears to be in French. The license plates are of European origin, and the surrounding architecture matches images captured around Place de la République."
How do I know any better? Well, I should probably crosscheck the signage with translation tools. Ah, it's French but some words are Dutch. Okay, so it could be somewhere else in Paris. Let's look into the license plate patterns...
At what point is it just better to do the whole thing yourself? Happy to be proven wrong here, but this same issue comes up time and time again with GenAI involved in discovery/research tasks.
EDIT: Maybe walk through the manual crosschecks hand-in-hand? "I see some of the signage is in Dutch, such as the road marking in the center left of the image. Are you sure this image is near Place de la République?" I have yet to see this play out in an interactive session. Maybe there's a recorded one out there...
There is a whole training section right there like you just didn’t feel like clicking on it
It's not? Why not? It's a "wake-up call", it's a "warning shot", but heaven forbid it's a rant against AI.
To me it's like someone listing off deaths from fentanyl, how it's destroyed families, ruined lives, but then tossing in a disclaimer that "this isn't a rant against fentanyl". In my view, the ways that people use and are drawn into AI usage has all the hallmarks of a spiral into drug addiction. There may be safe ways to use drugs but "distribute them for free to everyone on the internet" is not among them.
A director of NSA, pre 9/11, once remarked that the entire organization produced about two pieces of actionable intelligence a day, and about one item a week that reached the President. An internal study from that era began "The U.S. Government collects too much information".
But that was from the Cold War era, when the intelligence community was struggling to find out basic things such as how many tank brigades the USSR had. After 9/11, the intel community had to try to figure out what little terrorist units with tens of people were up to. That required trolling through far too much irrelevant information.
The idea that humans in general actually do any thinking is demonstrably false.
No one ever considers that maybe they all did lower our attention spans, prevent us from learning as well as we used to, etc. and now we are at a point we can't afford to keep losing intelligence and attention span
Of course, every ranter wants to be seen that way, and so a protest that something isn't a rant against X is generally a sign that it absolutely is a rant against X that the author is pre-emptively defending.
I think my biggest concern with AI is its biggest proponents have the least wisdom imaginable. I'm deeply concerned that our technocrats are running full speed at AGI with like zero plan for what happens if it "disrupts" 50% of jobs in a shockingly short period of time, or worse outcomes (theres some evidence the new tariff policies were generated with LLMs.. its probably already making policy. But it could be worse. What happens when bad actors start using these things to intentionally gaslight the population?)
But I actually think AI (not AGI) as an assistant can be helpful.
Or what?
Screaming "no one is evil, its just markets!" probably helps people who base their lives on exploiting the weak sleep better at night.
Speaking of Wisdom and a different "AGI", I think there's an old Dungeons and Dragons joke that can be reworked here:
Intelligence is knowing than an LLM uses vector embeddings of tokens.
Wisdom is knowing LLMs shouldn't be used for business rules.
At individual perspective - AI is useful as a helper to achieve your generative tasks. I'd argue against analytic tasks, but YMMV.
At the societal perspective, e.g. you as individual can not trus anything the society has produced, because it's likely some AI generated bullshit.
Some time ago, if you were not trusting a source, you could build your understanding by evaluating a plurality of sources and perspectives and get to the answer in a statistical manner. Now every possible argument can be stretched in any possible dimension and your ability to build a conclusion has been ripped away.
A further extension to the AI "conversation" might be: "What other locations are similar to this?" And "Why isn't it those locations?" Which you can then cross reference again.
Using AI as an entry point into massive datasets (like millions of photos from around the world) is actually useful. Correlation is what AI is good, but not infallible, at.
Of course false correlations exist and correlation is not causation but if you can narrow your search space from the entire world to the Eiffel tower in Paris or in Vegas you're ahead of the game.
A few thousand years of pre-LLM primary sources remain available for evaluation by humans and LLMs.
For example, human and biological based goals are around self-preservation and propagation. And this in turn is about resource appropriation to facilitate that, and systems of doing that become wealth accumulation. Species that don't do this don't continue existing.
A different branch of evolution of intelligence may take a different approach, that allows its affects to persist anyway.
You can look to the prohibition period for historical analogies with alcohol, plenty of enterprising humans there.
Every work had multiple versions. All versions were different. Some versions were diametrically opposed to others.
Have a look at Bible scholarship to see just _how_ divergent texts can become by nothing more than scribe errors.
Reality/truth/history has always been an expensive pursuit in the face of evolving pollutants.
"But the tradecraft is slipping. Analysts are skipping the hard parts. They’re trusting GenAI to do the heavy cognitive lifting, and it’s changing how we operate at a foundational level."
Next we're going to be hearing about how participation trophies and DEI are also contributing to this imagined "problem."
If I can solve two problems in a near constant time that is a few hours, what is the value of solving the problem which takes days to reason through?
I suspect that as the problem spaces diverge enough you’ll have two skill sets. Who can solve n problems the fastest and who can determine which k problems require deep thought and narrow direction. Right now we have the same group of people solving both.
> The study revealed a clear pattern: the more confidence users had in the AI, the less they thought critically
And the study didn't even checked that. They just plotted the correlation between how much user think they rely on AI vs how much effort they think they saved. Isn't it expected to be positive even if they think as critically.
[1]: https://www.microsoft.com/en-us/research/wp-content/uploads/...
• Instead of validating sources, they assumed the AI had already done so.
• Instead of assessing multiple perspectives, they integrated and edited the AI’s summary and moved on.
These are point against certain actions with a tool not against the tool.
AI is for the starting point not the final result.
AI must never be the last step but it often is because people trust computers especially if they answer in a confident language.
It's the ELIZA effect all over again.
The false promises of the AI companies and the false expectations of the management and users.
Had it just recently for a data migration where the users asked if they still need to enter meta data for documents they just could use AI to query data that was usually based on that meta data.
They trust AI before it's even there and don't even consider a transition period where they check if the result are correct.
Like with security convenience prevails.
If a human generates a story containing Count Dracula, that doesn't mean vampires are real, or that capabilities like "turning into a cloud of bats" are real, or that the algorithm "thirsts for the blood of the innocent."
The same holds when the story comes from an algorithm, and it continues to hold when story is about a differently-named character named "AI Assistant" who is "helpful".
Getting people to fall for this illusion is great news for the companies though, because they can get investor-dollars and make sales with the promise of "our system is intelligent", which is true in the same sense as "our system converts blood into immortality."
If we were headed straight to the AGI era then hey, problem solved - intelligent general machines which can advance towards solutions in a coherent if not human like fashion is one thing but that's not what AI is today.
AI today is enormously unreliable and very limited in a dangerous way - namely it looks more capable then it is.
https://news.ycombinator.com/item?id=43303755
I'm proud to see it evolving in the wild, this version is better. Or you know it could just be in the zeitgeist.
Hopefully narrowed by team, role and task..
“It’ll change everything!” they said, as they continued to put money in their pockets as people were distracted by the shiny object.
People is why we can't have anything nice. It sucks.
I have medical reasons to take opioids, but in the eyes of people, I am a junkie. I would not be considered a junkie if I kept popping ibuprofen. It is silly. Opioids do not even make me high to begin with (it is complicated).
It does not help that his examples of things an imaginary LLM might miss are all very subjective and partisan too.
One of the famous Greek philosophers complained that books were hurting people's minds because they no longer memorized information, so this kind of complaint is as old as civilization itself. There is no evidence that we would be on Mars by now already if we had never invented books or television.
The kids will be alright.
Here's the CIA's perspective on this subject.[1] The US intelligence community has a generative AI system to help analyze open source intelligence. It's called OSIRIS.[2] There are some other articles about it. The previous head of the CIA said the main use so far is summarization.
The original OSINT operation in the US was the Foreign Broadcast Monitoring Service from WWII. All through the Cold War, someone had to listen to Radio Albania just in case somebody said something important. The CIA ran that for decades. Its descendant is the current open source intelligence organization. Before the World Wide Web, they used to publish some of the summaries on paper, but as people got more serious about copyright, that stopped.
DoD used to publish The Early Bird, a daily newsletter for people in DoD. It was just reprints of articles from newspapers, chosen for stories senior leaders in DoD would need to know about. It wasn't supposed to be distributed outside DoD for copyright reasons, but it wasn't hard to get.
[1] https://www.cia.gov/resources/csi/static/d6fd3fa9ce19f1abf2b...
[2] https://apnews.com/article/us-intelligence-services-ai-model...
Maybe what I'm getting at is this [0] poem of Taylor Mali. Somehow we all lost our nerve to challenge really, really bad things, wrapping up messages in tentative language. Sometimes that's a genuine attempt at balance, or honesty. But often these days I feel an author is trying too hard to distance themself from ... from themself.
It's a a silly bugbear, I know.
[0] https://taylormali.com/poems/totally-like-whatever-you-know/
Gell-Mann Amnesia. Attention span limits the amount information of information we can process and with attention spans decreasing, increases to information flow stop having a positive effect. People simply forget what they started with even if that contradicts previous information.
> If I can solve two problems in a near constant time that is a few hours, what is the value of solving the problem which takes days to reason through?
You don't end up solving the problem in near constant time, you end up applying the last suggested solution. There's a difference.
This sort of lazy thinking doesn't miss a beat when it comes to take the opinions of an LLM at face value.
Why not? It sounds mostly the same. The motivations to believe AI, is exactly the same as the motivation to believe government officials and journalists.
If your LLM + pre-prompt setup sounds confident with every response, something is probably wrong; it doesn't have to be that way. It isn't for me. I haven't collected statistics, but I often get decent nuance back from Claude.
Think more about what you're doing and experiment. Try different pre-prompts. Try different conversation styles.
This is not dismissing the tendency for overconfidence, sycophancy, and more. I'm just sharing some mitigations.
Besides "OSINT" has been busy posting scareware for years, even before "AI".
There's so much spam that you can't figure out what the real security issues are. Every other "security article" is about "an attacker" that "could" obtain access if you were sitting at your keyboard and they were holding a gun to your head.
Maybe the article addresses that, I'm not permitted to read it, likely because I'm using IPv6.
Forensic Architecture is a decent counterexample, however. They've been using machine learning and computer synthesis techniques for years without dropping in quality.
Russia, North Korea and handful of other countries were spared, likely because they sided with the US and Russia at the UN General Assembly on Feb 24 of this year, in voting against “Advancing a comprehensive, just and lasting peace in Ukraine.” https://digitallibrary.un.org/record/4076672
EDIT: Found it: https://nitter.net/krishnanrohit/status/1907587352157106292
Also discussed here: https://www.latintimes.com/trump-accused-using-chatgpt-creat...
The theory was first floated by Destiny, a popular political commentator. He accused the administration of using ChatGPT to calculate the tariffs the U.S. is charged by other countries, "which is why the tariffs make absolutely no fucking sense."
"They're simply dividing the trade deficit we have with a country with our imports from that country, or using 10%, whichever is greater," Destiny, who goes by @TheOmniLiberal on X, shared in a post on Wednesday.
> I think they asked ChatGPT to calculate the tariffs from other countries, which is why the tariffs make absolutely no fucking sense.
> They're simply dividing the trade deficit we have with a country with our imports from that country, or using 10%, whichever is greater. https://t.co/Rc45V7qxHl pic.twitter.com/SUu2syKbHS
> — Destiny | Steven Bonnell II (@TheOmniLiberal) April 2, 2025
He attached a screenshot of his exchange with the AI bot. He started by asking ChatGPT, "What would be an easy way to calculate the tariffs that should be imposed on other countries so that the US is on even-playing fields when it comes to trade deficit? Set minimum at 10%."
"To calculate tariffs that help level the playing field in terms of trade deficits (with a minimum tariff of 10%), you can use a proportional tariff formula based on the trade deficit with each country. The idea is to impose higher tariffs on countries with which the U.S. has larger trade deficits, thus incentivizing more balanced trade," the bot responded, along with a formula to use.
John Aravosis, an influencer with a background in law and journalism, shared a TikTok video that then outlined how each tariff was calculated; by essentially taking the U.S. trade deficit with the country divided by the total imports from that country to the U.S.
"Guys, they're setting U.S. trade policy based on a bad ChatGPT question that got it totally wrong. That's how we're doing trade war with the world," Aravosis proclaimed before adding the stock market is "totally crashing."
They are much better at fractally subdividing and interpreting inputs like a believer of a religion, than at deconstructing and iteratively improving things like an engineert. It's waste of token count trying to have such discussions with an LLM.
Sometimes this is just sloppy methodology. Other times it is intentional.
Mere observation of others has shown me the decadence that results from even allowing such "tools" into my life at all.
(who or what is the tool being used?)
I have seen zero positive effects from the cynical application of such tools in any aspect of life. The narrative that we "all use them" is false.
Ask on a Wednesday. During a full moon. While in a shipping container. Standing up. Keep a black box on your desk as the sacred GenAI avatar and pray to it. Ask while hopping on one leg.
Techne is the Greek word for HAND.
There's definitely a metaphor to be made for trolling for data, that GP could have been intentionally making. I've certainly seen that idiom used before, although it could have been an eggcorn [2] for trawling.
[0] https://en.wikipedia.org/wiki/Trolling_(fishing)
[1] https://en.wikipedia.org/wiki/Troll_(slang)#Origin_and_etymo...
Since then, it's become clear that much life in the deep sea is anaerobic, doesn't use phosphorus, and may thrive without sunlight.
Sometimes anthropocentrism blinds us. It's a phenomenon that's quite interesting.
But all the examples feel like people are being really lazy, e.g.
> Paste the image into the AI tool, read the suggested location, and move on.
> Ask Gemini, “Who runs this domain?” and accept the top-line answer.
Yes, thars a part of why AI has its bad rep. It has uses to streamline workflow but people are treating it like an oracle. When it very very very clearly is not.
Worse yet, people are just being lazy with it. It's the equi talent of googling a topic and pasting the lede of the Wikipedia article. Which is tasteless, but still likely to be more right than an unfiltered LLM output
There are actually export statistics (obviously errors, possibly fraud) for these islands. Someone probably stuck the numbers in a formula without digging a little deeper.
Or if not, then what, is it not true that both substances and AI can be used responsibly, and irresponsibly?
"People is why we can't have anything nice. It sucks." is also true, applies to many things, just consider vending machines alone, or bags in public (for dog poop) and anything of the sort. We no longer have bags anymore, because people stole it. A great instance of "this is why we can't have nice things". Pretty sure you can think of more.
Make the down-votes make sense, please.
(I do not care about the down-votes per se, I care about why I am being disagreed with without any responses.)
... or just to know what they seem to be thinking, which is also important.
The short answer is: you can know for a fact that it _isn't_ thinking more carefully because LLMs don't actually think at all, they just parrot language. LLMs are performing well when they are putting out what you want to hear, which is not necessarily a well thought out answer but rather an answer that LOOKS well thought out.
These people absolutely were reading Reddit comments from a year ago to help them parse unfamiliar jargon in some document they found or make sense of what's going on in an image or whatever.
Basically it's a response to regulatory reality, little different from soy wire insulation in automobiles. I'm sure they'd love to deliver pure opium and wire rodents don't like to eat but that's just not possible while remaining in the black.
The LLM-only AI just hands you a fully-formed opinion with always-plausible-sounding reasons. There's no cognitive prompt to make you consider if it's wrong. I'm actually deliberately cultivating an instinctive negative distrust of LLM-only AI and would suggest it to other people because even though it may be too critical on a percentage basis, you need it as a cognitive hack to remember that you need to check everything coming out of them... not because they are never right but precisely because they are often right, but nowhere near 100% right! If they were always wrong we wouldn't have this problem, and if they were just reliably 99.9999% right we wouldn't have this problem, but right now they sit in that maximum danger zone of correctness where they're right enough that we cognitively relax after a while, but they're nowhere near right enough for that to be OK on any level.
I've encountered enough instances in subjects I am familiar with where the "I'm 14 and I just googled it for you" solution that's right 51% of the time and dangerously wrong the other 49 is highly upvoted and the "so I've been here before and this is kind of nuanced with a lot of moving pieced, you'll need to understand the following X, the general gist of Y is..." type take that's more correct is highly downvoted that I feel justified in making the "safe" assumption that this is how all subjects work.
On one hand at least Reddit shows you the downvoted comment if you look and you can go independently verify what they have to say.
But on the other hand the LLM is instant and won't screech at you if you ask it to cite sources.
2. Most analysts in a formal institution are professionally trained. In Europe, Canada and some parts of the US it's a profession with degree and training requirements. Most analysts have critical thinking skills, for sure the good ones.
3. OSINT is much more accessible because the evidence ISN'T ALWAYS controlled by a legal process so there are a lot of people who CAN be OSINT analysts or call themselves that and are not professionally trained. They are good at getting results from Google and a handful of tools or methods.
4. MY OPINION: The pressure to jump to conclusions in AI whether financially motivated or not comes from perceived notion that with technology everything should be faster and easier. In most cases it is, however, just as technology is increasing so is the amount of data. So you might not be as efficient as those around you expect, especially if they are using expensive tools, so there will be pressure to give into AI's suggestions.
5. MY OPINION: OSINT and analysis is a Tradecraft with a method. OSINT with AI makes things possible that weren't possible before or took way too much time for it to be worth it. Its more like, here are some possible answers where there were none before. Your job is to validate it now and see what assumptions have been made.
6. These assumptions have existed long before AI and OSINT. I seen many cases where we have multiple people look at evidence to make sure no one is jumping to conclusions and to validate the data. MY OPNION: So this lack of critical thinking might also be because there are less people or passes to validate the data.
7. Feel Free to ask me more.
class ParsedAddress(BaseModel):
street: str | None
postcode: str | None
city: str | None
province: str | None
country_iso2: str | None
{
"street": "Boulevard",
"postcode": 12345,
"city": "Cannot be accurately determined from the input",
"province": "MY and NY are both possible in the provided address",
"country_iso2": "US"
Sure, I can spend 2 days trying out different models and tweaking the prompts and see which one gets it, but I have 33 billion other addresses and a finite amount of time.
The issue occurs in OSINT as well: A well structured answer lures people into a mental trap. Anthropomorphism is something humans have fallen for since the dawn of mankind and is doing so yet again with AI. The thought that you have someone intelligent nearby with god-like abilities can be comforting but... Um... LLMs don't work like that.
But AI will review it and then you only have to .. oh
But AI will review AI and then you .. oh ..
What Dutch OSINT Guy was saying here resonates with me for sure - the act of taking a blurry image into the photo editing software, the use of the manipulation tools, there seems to be something about those little acts that are an essential piece of thinking through a problem.
I'm making a process flow map for the manufacturing line we're standing up for a new product. I already have a process flow from the contract manufacturer but that's only helpful as reference. To understand the process, I gotta spend the time writing out the subassemblies in Visio, putting little reference pictures of the drawings next to the block, putting the care into linking the connections and putting things in order.
Ideas and questions seem to come out from those little spaces. Maybe it's just letting our subconscious a chance to speak finally hah.
L.M. Sacasas writes a lot about this from a 'spirit' point of view on [The Convivial Society](https://theconvivialsociety.substack.com/) - that the little moments of rote work - putting the dishes away, weeding the garden, the walking of the dog, these are all essential part of life. Taking care of the mundane is living, and we must attend to them with care and gratitude.
Malaise exists at an individual level, but it doesn't transform into social malaise until someone comes in to exploit those people's addictions for profit.
2. While the question of "is the AI thinking" is interesting, I think it is a malformed question. Think about it: how do you make progress on that question, as stated? My take: it is unanswerable without considerable reframing. It helps to reframe toward something measurable. Here, I would return to the original question: to what degree does an LLM output calibrated claims? How often does it make overconfident claims? Underconfident claims?
3. Pretending requires at least metacognition, if not consciousness. Agree? It is a fascinating question to explore how much metacognition a particular LLM demonstrates.
In my view, this is still a research question, both in terms of understanding how LLM architectures work as well as designing good evals to test for metacognition.
In my experience, when using chain-of-thought, LLMs can be quite good at recognizing previous flaws, including overconfidence, meaning that if one is careful, the LLM behaves as if it has a decent level of metacognition. But to see this, the driver (the human) must demonstrate discipline. I'm skeptical that most people prompt LLMs rigorously and carefully.
4. It helps discuss this carefully. Word choice matters a lot with AI discussions, much more than a even a relatively capable software developer / hacker is comfortable with. Casual phrasings are likely to lead us astray. I'll make a stronger claim: a large fraction of successful tech people haven't yet developed clear language and thinking about discussing classic machine learning, much less AI as a field or LLMs in particular. But many of these people lack the awareness or mindset to remedy this; they fall into the usual overconfidence or lack-of-curiosity traps.
5. You wrote: "LLMs are performing well when they are putting out what you want to hear."
I disagree; instead, I claim people, upon reflection, would prefer an LLM be helpful, useful, and true. This often means correcting mistakes or challenging assumptions. Of course people have short-term failure modes, such is human nature. But when you look at most LLM eval frameworks, you'll see that truth and safety matter are primary factors. Yes-manning or sycophancy is still a problem.
6. Many of us have seen the "LLMs just parrot language" claim repeated many times. After having read many papers on LLMs, I wouldn't use the words "LLMs just parrot language". Why? That phrase is more likely to confuse discussion than advance it.
I recommend this to everyone: instead of using that phrase, challenge yourself to articulate at least two POVs relating to the "LLMs are stochastic parrots" argument. Discuss with a curious friend or someone you respect. If it is just someone online you don't know, you might simply dismiss them out of hand.
The "stochastic parrot" phrase is fun and is a catchy title for an AI researcher who wants to get their paper noticed. But isn't a great phrase for driving mutual understanding, particularly not on a forum like HN where our LLM foundations vary widely.
Having said all this, if you want to engage on the topic at the object level, there are better fora than HN for it. I suggest starting with a literature review and finding an ML or AI-specific forum.
7. There is a lot of confusion and polarization around AI. We are capable of discussing better, but (a) we have to want to; (b) we have to learn now; and (c) we have to make time to do it.
Like I wrote in #6, above, be mindful of where you are discussing and the level of understanding of people around. I've found HN to be middling on this, but I like to pop in from time to time to see how we're doing. The overconfidence and egos are strong here, arguably stronger than the culture and norms that should help us strive for true understanding.
8. These are my views only. I'm not "on one side", because I reject the false dichotomy that AI-related polarization might suggest.
Now that we have thinking models and methodology to train them, surely before long it will be possible to have a model that is very good at the kind of thinking that an expert OSINT analyst knows how to do.
There are so many low hanging fruit applications of existing LLM strengths that have simply not been added to the training yet, but will be at some point.
In particular LLMs seem particularly good at passing the initial smell test, which I'd imagine is first line of defense for most on determining whether to trust info. And unless it is something critical most people probably wouldn't deem looking at sources worth while.
Lately I've been running many queries against multiple LLMs. Not as good as organic thinking but comparing two does at least involve a bit of judgement as to which set of info is superior. Probably not the most eco friendly solution....
A better example would have been the complaint tablet to Ea-nāṣir. We're pretty sure it's real; there might still be people alive that remember it being discovered. But in a hundred years, people with gen AI have created museums of fake artifacts but plausible, can future people be sure? A good fraction of the US population today believes wildly untrue things about events happening in real time!