Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT now performs well at GeoGuesser

flausch.social

Post reply on HN
Print view View original

51–60 of 152 posts

Re: ChatGPT now performs well at GeoGuesser

#51

Post by sadeshmukh »

Kolya wrote: post #31
actuallyalys wrote: post #21

It wouldn’t shock me if multimodal LLMs were good at GeoGuesser [0], but if we’re being picky, it takes more than a few examples to demonstrate a game is “solved.” I also wonder what kind of data leakage might have been at play, like other people have suggested. To be clear, my point is not that this is unimpressive, just that this doesn’t demonstrate much. (Edit: I should have said, it doesn’t demonstrate what the t…

The examples are cherry-picked. I took a photo outside my office window in a built-up area, o3 thought for 5m 7s (!), and it got the location wrong by 40km. Doesn't look solved to me.

40 km is imo pretty impressive, but the 5 minute is really a killer. No use in real time applications

Re: ChatGPT now performs well at GeoGuesser

#52

Post by antonvs »

sadeshmukh wrote: post #51
Kolya wrote: post #31

Earlier quoted context omitted.

The examples are cherry-picked. I took a photo outside my office window in a built-up area, o3 thought for 5m 7s (!), and it got the location wrong by 40km. Doesn't look solved to me.

40 km is imo pretty impressive, but the 5 minute is really a killer. No use in real time applications

For real time applications, GPS seems like it'd be much more useful.

Re: ChatGPT now performs well at GeoGuesser

#53

Post by iambateman »

As the article notes, our threat model for who can identify where a picture was posted needs to change from “dedicated, skilled person” to “any creep with $20.”

That’s the point of the switch and it’s a big deal. We’re so used to posting pictures online…I’m just not sure it’s a good idea long-term.

Re: ChatGPT now performs well at GeoGuesser

#54

Post by avgDev »

MattGaiser wrote: post #25
dataviz1000 wrote: post #12

Earlier quoted context omitted.

I'm in Lima, Peru on vacation. Yes, it knows where I am.

I am sending it screenshots of my vacation pictures (i.e. without the metadata) and screenshots from GeoGuessr. It’s not doing well and is making guesses based on my location. So I sent it a photo from Warsaw and it guessed a place in my hometown of Calgary. Sent it a picture of Bangkok and it guessed Calgary in the summer or Hawaii or something like Singapore. It confidently stated that a photo from Versailles was H…

So much for a "Solved Problem".

Re: ChatGPT now performs well at GeoGuesser

#55

Post by SecretDreams »

Mistletoe wrote: post #24
colordrops wrote: post #20

Earlier quoted context omitted.

"The question of whether a computer can think is no more interesting than the question of whether a submarine can swim." - Edsger Dijkstra

Isn’t a human brain an organic computer? I find the human brain very interesting. The most interesting thing I know of in the universe.

I think growing a baby from nothing is pretty nifty, but brains are up there too.

Re: ChatGPT now performs well at GeoGuesser

#56

Post by swat535 »

AStonesThrow wrote: post #47
xnickb wrote: post #35

Earlier quoted context omitted.

Religious people claim it's merely a communication device, and the computer is elsewhere. Clearly one's interpretation is a function of the paradigm one exists in.

> Religious people claim it's merely a communication device, and the computer is elsewhere. No, that is a seriously weird and imaginary perversion that you've invented, unless you have some citation to a doctrine. Christians do not consider body organs as "devices" nor do our churches teach doctrine on "external computers", if you refer to the Holy Trinity as such? Are you thinking of Latter-Day Saints mythology, or…

Not to put words in the parent’s mouth, but I think they were pointing to the idea of the soul or consciousness, concepts that mean very different things in religious vs. scientific contexts. One sees it as an immaterial essence; the other as an emergent property of the brain.

Re: ChatGPT now performs well at GeoGuesser

#57

Post by bryanrasmussen »

AStonesThrow wrote: post #47
xnickb wrote: post #35

Earlier quoted context omitted.

Religious people claim it's merely a communication device, and the computer is elsewhere. Clearly one's interpretation is a function of the paradigm one exists in.

> Religious people claim it's merely a communication device, and the computer is elsewhere. No, that is a seriously weird and imaginary perversion that you've invented, unless you have some citation to a doctrine. Christians do not consider body organs as "devices" nor do our churches teach doctrine on "external computers", if you refer to the Holy Trinity as such? Are you thinking of Latter-Day Saints mythology, or…

I believe they are referring to the soul. The soul according to doctrine does not exist inside the body for it can exist after the body's destruction.

Re: ChatGPT now performs well at GeoGuesser

#58

Post by paulcole »

pelagicAustral wrote: post #41
layman51 wrote: post #26

Earlier quoted context omitted.

GeoGuessr is also not a “solved problem” in the sense that if you give the model a photo of an outdoor location that is not covered by Google Street View, then it will just make an educated guess which might still be many kilometers away.

I'm not ashamed to say that for about 6 months I played Geogessr thinking that you could not move around, but only look around... and I did pretty good in my mind.

Thats the only legit way to play IMO.

No moving, no panning, and no zooming is too hard. Moving is too boring. Pan and zoom is just the right balance.

Re: ChatGPT now performs well at GeoGuesser

#59

Post by paulcole »

Kolya wrote: post #31
actuallyalys wrote: post #21

It wouldn’t shock me if multimodal LLMs were good at GeoGuesser [0], but if we’re being picky, it takes more than a few examples to demonstrate a game is “solved.” I also wonder what kind of data leakage might have been at play, like other people have suggested. To be clear, my point is not that this is unimpressive, just that this doesn’t demonstrate much. (Edit: I should have said, it doesn’t demonstrate what the t…

The examples are cherry-picked. I took a photo outside my office window in a built-up area, o3 thought for 5m 7s (!), and it got the location wrong by 40km. Doesn't look solved to me.

That’s also not GeoGuessr.

Re: ChatGPT now performs well at GeoGuesser

#60

Post by bpodgursky »

sadeshmukh wrote: post #51
Kolya wrote: post #31

Earlier quoted context omitted.

The examples are cherry-picked. I took a photo outside my office window in a built-up area, o3 thought for 5m 7s (!), and it got the location wrong by 40km. Doesn't look solved to me.

40 km is imo pretty impressive, but the 5 minute is really a killer. No use in real time applications

The military/infosec uses of this are not real time. You can wait 5 minutes for a drone strike.
Post reply on HN