Hacker Newsnew | past | comments | ask | show | jobs | submit | barcode_feeder's commentslogin

I gave it a series of 11 images stripped of all metadata. It performed quite well, only misidentifying the two taken in a small college town in the NE of the US. It got two questions correct on photos taken in Korea (one with a fairly clear view of Haneul Park, the other a rather difficult to identify picture not resembling anything on google of Sunrise Peak). It got every other question in the US correct, ranging from some under-construction Austin taken from the river to some somewhat difficult shots in NYC (the upper halves of some building from Rockefeller terrace to the black wall of the MOMA). While not perfect, I'm bluntly shocked at how well it performed


I uploaded this image that I screenshotted off Google street view (no metadata) and it got with 200m.

https://chatgpt.com/share/6801bbf7-fd40-8008-985d-75c8813f55...

There is the chat.

Weirdly it said, "I’ve seen that exact house before on Google Street View when exploring Cairns neighborhoods."


> Weirdly it said, "I’ve seen that exact house before on Google Street View when exploring Cairns neighborhoods."

That's slightly creepy!


The anthropomorphisation certainly is weird. But the technical aspect seems even weirder. Did OpenAI really build dedicated tools to have their models train on Google Street View? Or do they have generic technology for browsing complex sites like Street view?


It’s just a hallucination, same idea as o3 claiming that it uses its laptop to mine Bitcoin:

https://transluce.org/investigating-o3-truthfulness

I doubt the model was trained on Street View, but even if it was, LLMs don’t retain any “memory” of how/when they were trained, so any element of truthfulness would be coincidental.


If it's trained on street view data it's not unlikely that the model can associate a particular piece of context to street view. For example, a picture can have telltale signs that street view content has, such as blurred faces and street signs, watermarks, etc.

Even if it's not directly trained on street view data it has probably encountered street view content in it's training dataset.


The training process doesn't preserve information needed for the LLM to infer that. It cannot be anything other than nonsense that sounds plausible, which is what they do best.


I think the test which the OP performed (to pick a random street view and let it pinpoint it) would indicate that it has ingested some kind of information in this regard in a structured manner.


They should definitely add that feature.

Tell it your name and then it just looks you up and street views your house, and puts that all into memory.


It might train off of street view


This was the image: https://imgur.com/a/cCUvgDG


This is the most impressive ChatGPT chat I’ve seen yet. While I theoretically can accept how large-scale probabilistic text generation can lead to this chain of “reasoning”, it really feels like actual intelligence.


It's been intelligence for a long time; the goalposts just shift, and people can't abstract the idea to an LLM. But language processing and large data processing itself IS a form of intelligence.


Maybe you're right, but I think it's more likely that it had been trained on street view photos and then invented a plausible justification for the guess afterwards (which is something I often see ChatGPT do, when it easily arrives at the correct answer, but gives bullshit explanations for it).


I played a round of Geoguessr against it and while it did a shockingly good job compared to what I was expecting, it still lags behind even novice human players.

The locations and its guesses were:

Bliss, Idaho - Burns, Oregon (273 miles away)

Quilleco, Biobio, Chile - Eugene, Oregon (6,411 miles away)

Dettighofen, Switzerland - Mühldorf, Germany (228 miles away)

Pretoria, South Africa - Johannesburg, South Africa (36 miles away)

Rockhampton, Australia - Gold Coast, Australia (437 miles away)


Okay, I decided to benchmark a bunch of AI models with geoguessr. One round each on diverse world, here's how they did out of 25,000:

Claude 3.7 Sonnet: 22,759

Qwen2.5-Max: 22,666

o3-mini-high: 22,159

Gemini 2.5 Pro: 18,479

Llama 4 Maverick: 14,316

mistral-large-latest: 10,405

Grok 3: 5,218

Deepseek R1: 0

command-a-03-2025: 0

Nova Pro: 0


Neat, thanks for doing this!


How does Google Lens compare?


I tried it but as far as I can tell Google Lens doesn't give you a location - it just describes generally what you're looking at.


I had cause to try Google Lens today and found the location to exact address thanks to a veterinary clinic which was in the background of an image. ChatGPT got the country but wrong city.


What about 04-mini-high ?


OpenAI's naming confuses me but I ran o4-mini-2025-04-16 through a game and it got 23,885


Interesting. It supports what they said (this is the model with good visual reasoning)


I just took a picture from my own front porch of the street and the houses opposite. It said 'probably Australia but I'd need more info'.

I said, give me your best guess.

And it guessed Canberra, Australia. Where I'm sitting right now drinking a Martini. Pretty spectacular.


I gave It some photos from denmark, didn't even bother to strip the metadata. One is correctly said give of "Scandinavian vibes" every other photo was very wrong. I also gave it a photo of the french Alps, it guessed Switzerland.


I gave o4-mini-high a cropped version of a photo I found on Facebook[0][1], and it quickly determined that this was in the UK from the road markings. It also decided that it was from a coastal city because it could see water on the horizon, which is the correct conclusion from incorrect data. There is no water, I think that's trees on a hill. It focused heavily on the spherical structure, which makes sense because it's distinctive, though it had a hard time placing it. It also decided that the building on the left was probably a shopping centre.

It eventually decided that the photo was taken outside the Scottish Exhibition and Conference Centre in Glasgow. It actually generally considered Scottish locations more than others.

The picture was actually taken in Plymouth (so pretty much as far from Scotland as you can get in Britain), on Charles Street looking south-east[2]. The building on the right is Drake Circus, and the one on the left is the Arts University. It actually did consider Plymouth, but decided it didn't match.

[0] This image with the "university plymouth" on the left cropped out, just to make it harder: https://www.facebook.com/photo/?fbid=9719044988151697&set=gm...

[1] https://chatgpt.com/share/68024c91-61d0-800c-99b1-fcecf0bfe8...

[2] https://maps.app.goo.gl/3TXv2UxH5128xQjJ9


"The statements had been proved 20 years earlier by a little-known English mathematician named L.J. Rogers... Rogers was content to do his research in relative obscurity, play piano, garden and apply his spare time to a variety of other pursuits"

divinely inspiring


Indeed. Also is the retirement dream for many working software engineers.


The usage of British English as an identifying factor seems ridiculous to me. Faking such a thing is extraordinarily low-hanging fruit even for a bumbler. Letting slip some regional red herring is quite literally the easiest misdirection I can think of)


He's also fairly inconsistent. He occasionally uses -ize endings, e.g. "realize". He also uses some language that is more prevalent in North American English, e.g. "sucks", "gotten", "chump change" and others.

Someone attempting to use British English to obscure their identity would know to use words like "maths" and "flat", as he did; not as many people however are aware that British English doesn't (typically) feature the word "gotten", which he used on multiple occasions. To me, this lends to the theory the BrE was a misdirection.

(As an aside, he's inconsistent in some other things, like "e-mail" versus "email", "TOR" versus "Tor", and "double click" versus "doubleclick".)


https://www.nytimes.com/interactive/2023/12/17/world/asia/my...

This has a different victim demographic but is broadly similar and was a good read (at least as far as I remember)


I'm glad someone else mentioned rathergood! That's the one that sticks out in my memory as older weird internet


While I understand why you'd remove it, I appreciate the initial inclusion. It's humanizing and genuine. Cool post even without it!


I'm a late adopter of HN, but was a bit surprised to see nothing about Israel/Hamas on the home page. Is that sort of political stuff forbidden, or is there just no interest in this community?


From the guidelines:

> Off-Topic: Most stories about politics, or crime, or sports, or celebrities, unless they're evidence of some interesting new phenomenon. Videos of pratfalls or disasters, or cute animal pictures. If they'd cover it on TV news, it's probably off-topic.

The most recent fluctuations of the war between Israel and its neighbors that has been going on literally since Israel existed with lulls but rarely more than a few weeks without new armed attacks or human rights abuses directed by one side against the other (and with, on both sides, deliberate or indiscriminate targeting, and systematic collective punishment of civilians being more a norm than an exception, with the excuse that what the other side is doing justifies it, makes even events that feature these not novel phenomena even in the context of the specific conflict.)


https://news.ycombinator.com/newsguidelines.html

The guideline criteria say not to submit flamebait, to submit things with interesting, novel intellectual discussion, and generally nothing that is on tv news.

I would imagine that the decades long Israeli/Palestinian conflict would violate all three.


> I would imagine that the decades long Israeli/Palestinian conflict would violate all three.

So would the Russian/Ukrainian conflict, and yet[1]

[1]https://news.ycombinator.com/item?id=30450336


Slightly offtopic, but it's interesting to read comments from that day knowing what we know now. Especially those stating that there will be no conventional war nor loss of life.


I'd add to this that I think that thread is why it's interesting to see/have these discussions on Hacker News. The "appropriate" places for these topics aren't especially interesting because they tend to be ideological echo chambers for one side or the other. That leads to trite and banal "discussion", if it can even be called that. Actually seeing 'real' people all across the spectrum of views, values, backgrounds, and so on is just so much more interesting.


The fact that these conversations are rare on HN, is why the conversation tends not to be trite and banal.


> I'm a late adopter of HN, but was a bit surprised to see nothing about Israel/Hamas on the home page. Is that sort of political stuff forbidden, or is there just no interest in this community?

Because it isn't reddit. I'm sure this forum is full of Israeli but also Muslim devs, do you really want that kind of strife on HN? I don't. There is no way that kind of discussion will not devolved into a giant mess.

It's even not because it's controversial, there has been outrageous topics on this forum (from my perspective), but it's already discussed everywhere else, with the exact same predictable results...


[flagged]


Flamebait comments like this are exactly why these discussions are mostly against the policy at HN.


There is no interest in showing Israel being attacked but there is plenty of political discussion such as daily flamewars about ancient misdeeds of US three letter agencies (but almost never discussions when these agencies shut down a botnet or do something positive).

These rules are either applied unfairly or the community is full of shills that make it appear that way.


> but never discussions when these agencies shut down a botnet

https://news.ycombinator.com/item?id=35445867 (121 comments)

https://news.ycombinator.com/item?id=37310772 (171 comments)


To echo others, it’s against policy. Not that we don’t care (we do) but that it’s better covered (and serviced) elsewhere, this isn’t the forum for those kinds of discussions.


It’s a guaranteed unproductive flame war. The exact same flame war that’s been happening about as long as the Internet has existed. It’s really boring at this point.


Just stop at unproductive. We don’t expect comments on HN to upend the power-structures which fuel this ongoing conflict.

Calling it boring — as if it had the potential to entertain - seems an bit insensitive to all the victims of the current and past hostilities.


Boring in the sense that any such threads are predictably flamewar. Not that the tragedy is boring.


JFC. Boring in the sense that exactly nothing new is going to be posted and anyone who’s seen a couple of these could write entire exchanges that will happen basically verbatim. The discussion is guaranteed to be 100% a repeat of all the other ones. So, boring.


I dunno, a thousand terrorists coordinating land, sea, and air (paraglider and drone) over WhatsApp with videographers posting professionally editing footage to Telegram in real-time is actually pretty interesting.


Funny we call them terrorists while hundreds of millions view them as freedom fighters trying to liberate their lands. I wonder if the East and the West are destined to just always collide. The staunch support by the western governments for their permanent-colony experiment in Palestine (ie Israel) is what leads to millions of Middle Easterners continually hating western governments.


Yes, because why wouldn’t you call the people killing innocent 20 year olds at a fucking rave freedom fighters? Or going in families homes and killing parents in front of their children? Or raping and murdering innocent women? Or kidnapping and mutilating women and children?

Idgaf what nationality people are. When they commit acts like these, they are _not_ fighting for freedom, and they are _not_ worthy of any kind of respect. If it’s Americans, Russians, Chinese, Germans, or Palestinians committing these atrocities, it is disgusting, and they are _terrorists_, and I really hope that there is swift and harsh retaliation against these horrible human rights violators.


Gabe, with all respect go read history. If you don’t want to read there are literally video interviews with old Israelis [1] who admit to being part of the death squads which went around killing Palestinian villagers and raping their women. This is what makes easterners describe what the Palestinians are doing as resistance for liberation.

I know Israelis want everyone to just move on from how bloody of a process it was to displace nearly one million indigenous Palestinians prior to announcing the foundation of their state. Unfortunately/fortunately — depending on your politics — we are (1) in modern times so most people don’t take kindly to European settler-colonial efforts and (2) the discomfort of Israelis with their history won’t erase the bloody genesis…

- 1: https://youtu.be/MQ1TAOibLss


The news may be interesting. No open-registration forum has generated an interesting thread about this topic since the very first such discussion that ever happened. It's basically one of those laws of the Internet.


You're right, and usually threads that do get up-voted are focused on specific technological novelties, whether in the title of the article or the content.

Just look at this story - the reasons it's considered "reasonable" for HN is because the framing is a technological story (internet traffic spike). That then also gives a small space for some people to discuss other aspects of this.


The Ukraine/Russia war was OK[1] to post, so it seems weird that Israel/Palestinian wouldn't be.

[1]https://news.ycombinator.com/item?id=30450336


A land invasion bordering what is now the EU hasn’t happened in over half a century.

And most of the posts are technical discussions about the implications of the war, like Russia trying to get an airplane to fly out of a wheat field; not inane “X has progressed 200 meters” or whatnot.


Check https://news.ycombinator.com/newsfaq.html and https://news.ycombinator.com/newsguidelines.html

In general, even if it's submitted, it's up to a bunch of clicks on the small arrows to keep it on the first page, and other submissions might win instead.

Even if there was a "it must be about hardware or software" rule, that wouldn't be such a bad thing, otherwise we'd just end up getting another reddit.


this is another reddit. just with less functionality


Anything that can lead to flame wars is frowned upon and gets pruned in a number of ways. Political topics tend not to lead to useful, sober discussions.


Forbidden per the guidelines.

> Off-Topic: Most stories about politics, or crime, or sports, or celebrities, unless they're evidence of some interesting new phenomenon. Videos of pratfalls or disasters, or cute animal pictures. If they'd cover it on TV news, it's probably off-topic.

https://news.ycombinator.com/newsguidelines.html


Note it says "most" and "probably". Evening news grade stuff is still admissible if it fits the "On Topic" conditions.


This is all you need to know: https://www.youtube.com/watch?v=2z9UyPurVok


Relatively small amount of flaggers can keep a topic off the front page if they think it doesn't belong there.


I don't have children yet, but I expect to in the coming years. I wish to instill a love of reading in them, but I worry about how to go about doing so.

I spent so much time reading as a child that I'd get grounded from it, forbidden from reading at the dinner table, that sort of thing. But that love came partially from a lack of alternatives; we only had local TV channels and no video games or other comparable distractions.

I worry how, to a child, a book can be as effective a lure as a screen when I, as an adult, often find myself taken by the latter.


First read to your kid. Fill your house with books and let them know by example that they are fun and important.

The other advice I'd have is to take your kid to the library once a week. Let them pick out a couple books (maybe also pick up one your choosing so that you can expose them to more things - kids can be obsessive) Once they can manage a chapter book tell them they are allowed to pick whatever books they like from anywhere in the entire library. Kids love having choice and flexing agency. Don't worry if they pick up something too mature/scary/boring/whatever. Anything they want to pick really is fine, but if they don't like it they'll have to wait until next week to swap it out.

Get them their own library card as soon as you can too. Make a big deal of it. Having an official document for access to the books highlights the importance of them too.


This will be controversial, but the one thing that worked to get my kids to read every day was to impose a strict zero screen time policy.

edit: ages 3-9


We're doing the same thing. We don't even own a TV. Toddler loves books so much we get sick of reading to him (but do it anyway).


yeah, going screen free is hard work. a TV/tablet is the ultimate babysitter.


Will it be controversial? I feel that to most of us in tech it makes intuitive sense that too much screen time isn't a good thing. We know the vice it can be.


That's true but most people will argue for at least some limited screen time. For us only a complete ban got them interested in reading, outdoors, sports, card games etc again.


Ive spent hundreds of hours reading aloud to my kids. It's a big investment but I believe it's almost certain you'll stumble upon something that will catch their attention. It's also about family culture - if you establish the habit of reading at a certain time of day (we do it after dinner / before bed) then it's easy to stick with that over turning to other things to fill the time.


Read to your child, and provide books and opportunity once he starts learning to read. He'll either be drawn to it or not. I've had three kids and their interests and things that motivate them are all wildly different. So a lot of it just boils down to "if he likes reading, he'll read."

Do the same with music, and sports/physical activity, and other things.


I was not read to as a child, but I was pretty good at reading from an early age and grew to love it. Adult life is busy and many other activities compete for my attention now, but I still manage to finish around 10-20 books a year. This lower bound is not impressive to any serious, habitual reader, but I'm happy I manage to keep it up.

I think what the trick for me:

- My parents read regularly, so it just was a normal part of life in my eyes.

- My parents' siblings would gift me volumes of children's book series for birthdays and holidays. I'd look forward to getting the next book in a series for months, and it was a big exciting event for me.

- Adults were willing to discuss the books I read with me (and sometimes humor me by reading one), so I got recognition from them for what I was doing, and learned to connect over books. If a book was important to me, other people were willing to take my feelings seriously. Being rooted in a book made them legitimate.

- My dad was not a spend-y sort of person, but once a year for Christmas, he'd rifle through a book club catalog with me and let me order more or less any book I wanted. This was exciting, and we got tons of random books at home filling the shelves.

- Access to books, i.e. those full shelves. Bored to tears on a rainy day? My father would tell me to pick a book.


Just to note: their child might not be a boy. Feels a bit disconcerting to see the male default used here.


I'm actually sorry to see this down voted.

I have a daughter, and it is just crazy once you start reading books how much is needlessly male focussed.

From the first farmyard books where the boy is driving the tractor and the girl is in the cart on the back. To fantastic Mr fox and his four small foxes (all male) etc etc. It's just extremely unbalanced.

In our community, there's every reason not to use a male default everywhere. It would not have changed the original commenters meaning at all to have said "they" instead of "him" and would have been more inclusive. Surely that's a good default?


Here I've seen female and male defaults used as well as "they". I haven't noticed any prevalence for one other the others, probably because it makes no difference to me. I understand from the context that the text is applicable to any gender.


I'm curious, if it makes no difference to you then why not extend your sympathy to those who feel they are affected by it and simply use a gender neutral pronoun?


I'm not sure how much of it is just not allowing my 2-year-old screen time (and not owning a television), but his 3 favorite activities are being read to, playing in the park, and playing with his wooden trains.

I wish it were as simple as making reading time quality bonding time. However, that doesn't seem sufficient. My two favorite activities these days are reading to him and giving him a bath, and yet he has a love/hate relationship with his bath. He acts like he hates the bath, until I announce to him that I'm going to take a bath without him. I think he really does hate his bath (and that it cuts into his train time), but puts up with it in order to spend time with me.

I do think it really helps that I give him time to examine the page and imagine before I turn the page. Also, sometimes he really wants to flip the page before I'm done reading, and I'm less happy with that, but let it happen.


Expose them to good books. Visit the library, fill the bookshelves at home, read to them. All kids love stories, so your job is simply to make it a nigh-certainty they will encounter stories they love.

Don’t try to force specific stories on them- just provide a buffet of great choices.


Marry someone who loves reading.


looks like the temperature got messed up somehow?


Know of any comparably inspired venues in 2023?


Not exactly the same thing, but they had a phone version of this at MeowWolf, the huge interactive trippy art exhibit, in Denver. They had phone terminals laying around with a whole directory of fake businesses. At first, I thought they staffed the lines with employees, but it turned out to be a system like this minus the video screens.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: