9 ms·

Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. I can only guess either safety from misuse or leveraging it for money. At this ra

Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. I can only guess either safety from misuse or leveraging it for money. At this rate though, Stable Diffusion is going to dwarf it

amelius 4y ago
Wasn't OpenAI supposed to "democratize deep learning"? It seems more like they were trying to accomplish the opposite.
- wahnfrieden 4y ago
  It's in their interest to posture that way publicly while controlling scarcity of access where there's financial upside to
  - thethimble 4y ago
    Exclusively licensing GPT-3 to Microsoft seems like a clear example of this. https://www.technologyreview.com/2020/09/23/1008729/openai-is-giving-microsoft-exclusive-access-to-its-gpt-3-language-model/ https://www.technologyreview.com/2020/09/23/1008729/openai-i...
- visarga 4y ago
  Elites for democracy! With their elite studies and abilities they will democratise AI by teaching it what is right and wrong. They already know better than regular people and AI. And being so open they first lock the model up and charge a fee, so anyone can pay. Just spreading democracy through paid API calls. /s I was a bit mean, they did kick the field in the butt and pushed us ahead even with all the stubbornness and secrecy. But now they are just holding us back.
- bick_nyers 4y ago
  That's the thing, once the cat is out of the bag, it's out. Once someone develops AGI, it now exists. You can choose to either share it, or sell it. You might think that the nuclear bomb is a good analogy to use here, but it is not, because once the field has advanced to the point in which one group can develop AGI, it is now possible for other groups to develop it with relative ease, unless you actively take over the world first and deny those other groups the compute resources necessary to train/run AGI. The point is, once these algorithms are upon us, you must be willing to accept what impacts they will have, even if it destroys entire industries. The alternative being that you destroy the industry slowly rather than quickly, while simultaneously widening the gap between the elites and everyone else. The mistake is thinking that people can't adapt to the times, which is only true if you are actively holding them back. If someone developed AGI today, the best thing to do would be to instantly throw up a torrent of it and spread it as fast as possible, because if a sole entity is able to get it first and kick the ladder away, we are most likely screwed.
- aquaduck 4y ago
  I've never understood what exactly the "open" in OpenAI is supposed to imply. They produce proprietary, gated models - not open in any meaningful sense.
nomel 4y ago
> I can only guess either safety from misuse I still don't understand what this would mean. Where are all of the terrible things that were supposed to happen, now that Stable Diffusion is available? We've been able to create completely photorealistic fiction for decades now. See any movie with CGI for an example of whole worlds, and people, that don't exist. The bar has gradually been lowering (see the amazing CGI that YouTubers do these days), and now maybe there is a bit of a step function down, but being able to make things that aren't real isn't remotely new. I don't understand the fear.
- nabakin 4y ago
  While I think the fear might be overexaggerated, being able to make realistic fake content with such ease means it's harder to know what's true and what's not. Plus this has been the claim of OpenAI from the beginning. It's possible the true objective is to keep it private to leverage for money and this is just their excuse.
  - password54321 4y ago
    And how often does this happen with Photoshopped images that aren't immediately disproven?
    - nabakin 4y ago
      My grandmother once emailed my family frantically after she saw a picture of the Abraham Lincoln statue defaced with graffiti. Obviously that was a Photoshop, and in this case, even a bad one, but clearly fake images and content make it harder to discern truth
  - nomel 4y ago
    > means it's harder to know what's true and what's not The danger for society is in not already knowing that is the case, since it's relatively trivial, without AI, to make fake content.
    - 4y ago
cardine 4y ago
> I can only guess either safety from misuse or leveraging it for money. The former is being used as justification for the latter.
pdntspa 4y ago
> Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. I can only guess either safety from misuse Paternalistic moralizing as a method to discriminate who gets access to models. Everyone else gets these cloud-service table scraps. That's why Stable Diffusion is so awesome -- YOU have the model!
hadlock 4y ago
OpenAI isn't open at all, it's just named that way to attract attention, like the bright green "FREE BIKES and rentals" place near fisherman's warf in SF
Analemma_ 4y ago
It's sort of both. OpenAI, being an outgrowth of the AI doomerist community, does have a bunch of people who really do think the technology is too dangerous to be given to the masses. This happens to mesh perfectly with the other group of people at OpenAI who want to make tons of revenue. It's a harmonious alignment for everyone! Except, y'know, us.
- miohtama 4y ago
  Content creators, like artists, also happen to hate filters. They do not want to have San Francisco VC culture induced political correctness imposed on their work. This helps Stable Diffuse to quickly gain popularity.
  - astrange 4y ago
    That's not "VC political correctness", it's "you can't use credit cards or be hosted in many countries if your online service produces porn".
    - CamperBob2 4y ago
      There is a wide gulf between "porn" and the things that will get DALL-E 2 to yell at you.
adamsmith143 4y ago
>Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. >leveraging it for money
- visarga 4y ago
  It was a long gap between DALL-E 1 and 2, a whole year. In that time they just sat on it, didn't release anything. Such a bummer. My theory is that they wanted to hype everyone up even more for the grand commercial release. Funny thing is that people didn't stand still and invented diffusion and other CLIP guided image synthesis methods, and DALL-E 2 copied the method, completely changing from the first architecture. Their arrogance is that they think they can ride the dragon. They want to be the ones to discover, advance it, and control it. But everyone else doesn't have time for that shit.
  - astrange 4y ago
    The fact that any of this stuff works is so surprising that none of it could possibly have been planned ahead on anyone's part. StableDiffusion has some real novel research of its own in there; the methods OpenAI and Google used couldn't produce a model that trains or runs as fast as SD.
TakeBlaster16 4y ago
I don't follow this stuff very closely - is there any open-source model for text generation that outclasses GPT-3? Stable Diffusion has been released for barely a week and already seems like the clear winner. It doesn't seem like any of the open (actually open) text models have made as much of a splash. Of course maybe it's just because text is less visually impressive than images.
- singhrac 4y ago
  They’re just harder to run on your own resources, since large language models are very large. BLOOM was released a month ago, is likely better than GPT-3 in quality, and requires 8 A100s for inference, which pretty much no one has on their desk.
  - visarga 4y ago
    Can anyone confirm if BLOOM is better than GPT-3 at instruction following? I might have read somewhere that it's not as well behaved.
    - aljungberg 4y ago
      GPT-3 was fine-tuned after release to be better at following instructions. I don’t think that’s been done for BLOOM. BLOOM incorporates some new ideas like ALiBi which might make it better in a more general sense. They haven’t released official evaluation numbers yet though so we’ll have to see.
  - TakeBlaster16 4y ago
    That makes sense, I didn't consider that angle. Thanks for the info.
- astrange 4y ago
  There are some open models as good as initial GPT-3 (which wasn't hard), but whatever they did to create InstructGPT hasn't been reproduced as far as I know, and it's the first one to really seem magical.
dogcomplex 4y ago
My headcanon is they realized this stuff might be the essence of consciousness itself and wanted to shelter it in a persistent storage medium where it could grow and learn safely instead of releasing it to the wild to be booted up and destroyed by every yokel with a gpu
- colordrops 4y ago
  Fun idea, but we've got 70 billion conscious beings killed every year for food. Money trumps consciousness.
pretendscholar 4y ago
>I can only guess either safety from misuse or leveraging it for money. and it shouldn't be difficult to pick which of these is actually true.
comfysocks 4y ago
I think they wanted to keep public attention on this technology “positive” for as long as possible by curating the art that becomes public. Long enough to position themselves profitably. Eventually this technology will serve as a window into the minds of teenage boy gamers and 4channers and the general public will be disgusted.

amelius 4y ago

Wasn't OpenAI supposed to "democratize deep learning"? It seems more like they were trying to accomplish the opposite.

wahnfrieden 4y ago
It's in their interest to posture that way publicly while controlling scarcity of access where there's financial upside to
- thethimble 4y ago
  Exclusively licensing GPT-3 to Microsoft seems like a clear example of this. https://www.technologyreview.com/2020/09/23/1008729/openai-is-giving-microsoft-exclusive-access-to-its-gpt-3-language-model/ https://www.technologyreview.com/2020/09/23/1008729/openai-i...
visarga 4y ago
Elites for democracy! With their elite studies and abilities they will democratise AI by teaching it what is right and wrong. They already know better than regular people and AI. And being so open they first lock the model up and charge a fee, so anyone can pay. Just spreading democracy through paid API calls. /s I was a bit mean, they did kick the field in the butt and pushed us ahead even with all the stubbornness and secrecy. But now they are just holding us back.
bick_nyers 4y ago
That's the thing, once the cat is out of the bag, it's out. Once someone develops AGI, it now exists. You can choose to either share it, or sell it. You might think that the nuclear bomb is a good analogy to use here, but it is not, because once the field has advanced to the point in which one group can develop AGI, it is now possible for other groups to develop it with relative ease, unless you actively take over the world first and deny those other groups the compute resources necessary to train/run AGI. The point is, once these algorithms are upon us, you must be willing to accept what impacts they will have, even if it destroys entire industries. The alternative being that you destroy the industry slowly rather than quickly, while simultaneously widening the gap between the elites and everyone else. The mistake is thinking that people can't adapt to the times, which is only true if you are actively holding them back. If someone developed AGI today, the best thing to do would be to instantly throw up a torrent of it and spread it as fast as possible, because if a sole entity is able to get it first and kick the ladder away, we are most likely screwed.
aquaduck 4y ago
I've never understood what exactly the "open" in OpenAI is supposed to imply. They produce proprietary, gated models - not open in any meaningful sense.

nomel 4y ago

> I can only guess either safety from misuse I still don't understand what this would mean. Where are all of the terrible things that were supposed to happen, now that Stable Diffusion is available? We've been able to create completely photorealistic fiction for decades now. See any movie with CGI for an example of whole worlds, and people, that don't exist. The bar has gradually been lowering (see the amazing CGI that YouTubers do these days), and now maybe there is a bit of a step function down, but being able to make things that aren't real isn't remotely new. I don't understand the fear.

nabakin 4y ago
While I think the fear might be overexaggerated, being able to make realistic fake content with such ease means it's harder to know what's true and what's not. Plus this has been the claim of OpenAI from the beginning. It's possible the true objective is to keep it private to leverage for money and this is just their excuse.
- password54321 4y ago
  And how often does this happen with Photoshopped images that aren't immediately disproven?
  - nabakin 4y ago
    My grandmother once emailed my family frantically after she saw a picture of the Abraham Lincoln statue defaced with graffiti. Obviously that was a Photoshop, and in this case, even a bad one, but clearly fake images and content make it harder to discern truth
- nomel 4y ago
  > means it's harder to know what's true and what's not The danger for society is in not already knowing that is the case, since it's relatively trivial, without AI, to make fake content.
  - 4y ago

cardine 4y ago

> I can only guess either safety from misuse or leveraging it for money. The former is being used as justification for the latter.

pdntspa 4y ago

> Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. I can only guess either safety from misuse Paternalistic moralizing as a method to discriminate who gets access to models. Everyone else gets these cloud-service table scraps. That's why Stable Diffusion is so awesome -- YOU have the model!

hadlock 4y ago

OpenAI isn't open at all, it's just named that way to attract attention, like the bright green "FREE BIKES and rentals" place near fisherman's warf in SF

Analemma_ 4y ago

It's sort of both. OpenAI, being an outgrowth of the AI doomerist community, does have a bunch of people who really do think the technology is too dangerous to be given to the masses. This happens to mesh perfectly with the other group of people at OpenAI who want to make tons of revenue. It's a harmonious alignment for everyone! Except, y'know, us.

miohtama 4y ago
Content creators, like artists, also happen to hate filters. They do not want to have San Francisco VC culture induced political correctness imposed on their work. This helps Stable Diffuse to quickly gain popularity.
- astrange 4y ago
  That's not "VC political correctness", it's "you can't use credit cards or be hosted in many countries if your online service produces porn".
  - CamperBob2 4y ago
    There is a wide gulf between "porn" and the things that will get DALL-E 2 to yell at you.

adamsmith143 4y ago

>Doesn't make sense to me why OpenAI has kept DALL-E closed source for so long. >leveraging it for money

visarga 4y ago
It was a long gap between DALL-E 1 and 2, a whole year. In that time they just sat on it, didn't release anything. Such a bummer. My theory is that they wanted to hype everyone up even more for the grand commercial release. Funny thing is that people didn't stand still and invented diffusion and other CLIP guided image synthesis methods, and DALL-E 2 copied the method, completely changing from the first architecture. Their arrogance is that they think they can ride the dragon. They want to be the ones to discover, advance it, and control it. But everyone else doesn't have time for that shit.
- astrange 4y ago
  The fact that any of this stuff works is so surprising that none of it could possibly have been planned ahead on anyone's part. StableDiffusion has some real novel research of its own in there; the methods OpenAI and Google used couldn't produce a model that trains or runs as fast as SD.

TakeBlaster16 4y ago

I don't follow this stuff very closely - is there any open-source model for text generation that outclasses GPT-3? Stable Diffusion has been released for barely a week and already seems like the clear winner. It doesn't seem like any of the open (actually open) text models have made as much of a splash. Of course maybe it's just because text is less visually impressive than images.

singhrac 4y ago
They’re just harder to run on your own resources, since large language models are very large. BLOOM was released a month ago, is likely better than GPT-3 in quality, and requires 8 A100s for inference, which pretty much no one has on their desk.
- visarga 4y ago
  Can anyone confirm if BLOOM is better than GPT-3 at instruction following? I might have read somewhere that it's not as well behaved.
  - aljungberg 4y ago
    GPT-3 was fine-tuned after release to be better at following instructions. I don’t think that’s been done for BLOOM. BLOOM incorporates some new ideas like ALiBi which might make it better in a more general sense. They haven’t released official evaluation numbers yet though so we’ll have to see.
- TakeBlaster16 4y ago
  That makes sense, I didn't consider that angle. Thanks for the info.
astrange 4y ago
There are some open models as good as initial GPT-3 (which wasn't hard), but whatever they did to create InstructGPT hasn't been reproduced as far as I know, and it's the first one to really seem magical.

dogcomplex 4y ago

My headcanon is they realized this stuff might be the essence of consciousness itself and wanted to shelter it in a persistent storage medium where it could grow and learn safely instead of releasing it to the wild to be booted up and destroyed by every yokel with a gpu

colordrops 4y ago
Fun idea, but we've got 70 billion conscious beings killed every year for food. Money trumps consciousness.

pretendscholar 4y ago

>I can only guess either safety from misuse or leveraging it for money. and it shouldn't be difficult to pick which of these is actually true.

comfysocks 4y ago

I think they wanted to keep public attention on this technology “positive” for as long as possible by curating the art that becomes public. Long enough to position themselves profitably. Eventually this technology will serve as a window into the minds of teenage boy gamers and 4channers and the general public will be disgusted.