First Impressions with GPT-4V(ision) • Hacker News Board

Re: First Impressions with GPT-4V(ision)

#91

Post by justlikeyou » Wed, Sep 27, 2023, 2:14 PM UTC

jihadjihad wrote: post #85

mbb70 wrote: post #37

The "Why is this image funny?" test reminds me of https://karpathy.github.io/2012/10/22/state-of-computer-visi... In 10 years we went from "SoTA is so far from achieving this I don't even know where to start" to "That'll be $0.0004 per token and have a nice day"

Has anyone tried GPT-4V on that image?

Note: I had to ask it why people in the photo are laughing.

In the image, Barack Obama, the former U.S. President, seems to be playfully posing as if he's trying to add weight while another official, who appears to be former UK Prime Minister David Cameron, is standing on a scale. Obama's gesture, where he's putting his foot forward as though trying to press down on the scale, suggests a playful attempt to make Cameron appear heavier. The lightheartedness of such a playful gesture, especially in the context of world leaders typically engaged in serious discussions, is a break from formality, which is likely why others in the vicinity are laughing. The scene captures a candid, informal moment amidst what might have been a formal setting or meeting.

Re: First Impressions with GPT-4V(ision)

#92

Post by apexalpha » Wed, Sep 27, 2023, 2:15 PM UTC

cs702 wrote: post #10

Sure, there are a few edge-case failures and mistakes here and there, but I can't help but be in awe . AWE. Let me state the obvious, in case anyone here isn't clear about the implications: If the rate of improvement of these AI models continues at the current pace, they will become a superior user interface to almost every thing you want to do on your mobile phone, your tablet, your desktop computer, your car, your…

Correct, this will be the successor to the GUI.

Re: First Impressions with GPT-4V(ision)

#93

Post by m3kw9 » Wed, Sep 27, 2023, 2:15 PM UTC

cs702 wrote: post #10

Sure, there are a few edge-case failures and mistakes here and there, but I can't help but be in awe . AWE. Let me state the obvious, in case anyone here isn't clear about the implications: If the rate of improvement of these AI models continues at the current pace, they will become a superior user interface to almost every thing you want to do on your mobile phone, your tablet, your desktop computer, your car, your…

Gpt4 was awesome, this looks like it does not disappoint based on the progress we expect of OpenAI.

Re: First Impressions with GPT-4V(ision)

#94

Post by apexalpha » Wed, Sep 27, 2023, 2:17 PM UTC

hypertele-Xii wrote: post #89

op00to wrote: post #65

Earlier quoted context omitted.

Until the AI mainframe runs on your $device

GPT-3 requires 700 gigabytes of GPU RAM. I'm looking at my cheapest computer components retailer listing a 48 gigabyte GPU at $5k. So to run the previous generation of GPT would cost me about $70k right now. When do you think I can expect to run GPT-4 on my consumer $device? :)

You are now where the GUI was in in 1986 or so.

Re: First Impressions with GPT-4V(ision)

#95

Post by zarzavat » Wed, Sep 27, 2023, 2:17 PM UTC

troupo wrote: post #60

dottjt wrote: post #18

Earlier quoted context omitted.

Oh wow, I'm completely fucked as a front end developer.

Has AI been able to figure out conflicting and quickly changing product requirements and implement them on top of existing technical debt? You'll be fine.

People keep saying "ah but it can't do X!". So what? Most of us have multiple decades until we can retire. This AI is getting better every few months.

It will be able to do it even faster, better and more cheaply than a human can.

Re: First Impressions with GPT-4V(ision)

#96

Post by boplicity » Wed, Sep 27, 2023, 2:18 PM UTC

layer8 wrote: post #28

cs702 wrote: post #10

Sure, there are a few edge-case failures and mistakes here and there, but I can't help but be in awe . AWE. Let me state the obvious, in case anyone here isn't clear about the implications: If the rate of improvement of these AI models continues at the current pace, they will become a superior user interface to almost every thing you want to do on your mobile phone, your tablet, your desktop computer, your car, your…

I don’t know, I hate the idea of having to hold a natural-language conversation with a computer in order to make use of its functionality. It feels like being one of those Futurama heads in a jar that can’t do anything by themselves.

Visual input is not "natural language." The implications of that could, especially in the long-term, could be incredibly far-reaching, and possibly even terrifying.

Re: First Impressions with GPT-4V(ision)

#97

Post by zerojames » Wed, Sep 27, 2023, 2:19 PM UTC

Usu wrote: post #9

I'd be interested in knowing how good it is at solving visual captchas, do we foresee a huge rise in automated bypasses?

We have just added a section on this! TL;DR: GPT-4V isn't great at this task at the moment :)

Re: First Impressions with GPT-4V(ision)

#98

Post by m3kw9 » Wed, Sep 27, 2023, 2:21 PM UTC

I’m just imagining a mode where OpenAI calls it “App Mode” where you say what you want say “a dog themed cute calculator app with units conversions”, and it will generate the UI for a working app. You add these into a widget like place. The OpenAI AppStore will carry these apps. Although in the beginning the apps would be simple but I do see potential

Re: First Impressions with GPT-4V(ision)

#99

Post by M4v3R » Wed, Sep 27, 2023, 2:21 PM UTC

hypertele-Xii wrote: post #89

op00to wrote: post #65

Earlier quoted context omitted.

Until the AI mainframe runs on your $device

GPT-3 requires 700 gigabytes of GPU RAM. I'm looking at my cheapest computer components retailer listing a 48 gigabyte GPU at $5k. So to run the previous generation of GPT would cost me about $70k right now. When do you think I can expect to run GPT-4 on my consumer $device? :)

Where did you get the 700 gigabytes figure from? I don't think OpenAI even released the model size, although it's considered to be 175B parameters. Given how well quantization works at these sizes you would need less than 200 GB of GPU memory to run it.

Re: First Impressions with GPT-4V(ision)

#100

Post by BoxOfRain » Wed, Sep 27, 2023, 2:28 PM UTC

thelittleone wrote: post #16

cs702 wrote: post #10

Sure, there are a few edge-case failures and mistakes here and there, but I can't help but be in awe . AWE. Let me state the obvious, in case anyone here isn't clear about the implications: If the rate of improvement of these AI models continues at the current pace, they will become a superior user interface to almost every thing you want to do on your mobile phone, your tablet, your desktop computer, your car, your…

Agree and the next big step may well be human computer interface. Speech is starting point for input. At some point output will change also and if think it out longer term perhaps a future where instead of reading information we install knowledge, including the stored memory of actual experience. If I want to do pottery, I could think this, download the experience and then be competent at it.

The idea of giving the adtech industry direct access to my brain is genuinely horrifying, not to mention Aldous Huxley would have some reasonable objections to everything being this easy all of a sudden.