The Silicon Valley Bias in AI Access
The AI world is exploding. Projects and new startups seem to be popping up every single day and advancements in technology seem to be exponential. Are we in a golden age of AI technology? It would be difficult to argue that we aren't. It would also be difficult to argue that some of the best technology remains gatekept and locked by Silicon Valley to those in Silicon Valley ensuring that these companies have first-mover advantages and an unfair start on the competition.
If someone were to ask you to take part in a 100-meter running race where you got a prize at the end, you'd be inclined to take them up on that offer if you felt you were good at running. If someone asked you to take part in the same running race but you were to start at the start line and a few others were starting at the 70-meter mark, you would say that is silly and how could you ever win. Unfortunately, this is the state of play with some AI technology and in this article, we're going to discuss this very thing.
DALL-E or DALL-E 2 from OpenAI is an image generation model that has been available for those lucky enough to have got access in the early beta since March/April 2022. Early access was given to artists and influencers to make content on the technology and all generations had to be created within the OpenAI application. A popular early-access video of this technology can be seen here by popular YouTuber Marques Brownlee.
Whilst this may have seemed unfair for those on the waiting list and wanting access, I understand it. Marques has a large audience and it makes sense to leverage that audience to get the word out about DALL-E.
DALL-E has since been opened up to everyone, you can signup today and start using it within OpenAI. This then leads to the next question that most avid AI developers would have in mind, when can I put DALL-E into my project? When is an API available? You ask such questions on OpenAI's official forums and you are told that an API does not exist and that it hasn't been released yet. There is no waitlist, no way to inform them of your ideas or concepts for integrating DALL-E, and no process for taking this feedback.
Products are popping up using DALL-E and this must be due to the fact that they are testing the DALL-E API with OpenAI. I have questions however on how these products were chosen and why there was not a more transparent process to give users access. This brings back memories of the GPT-3 rollout where certain companies had legacy approval for certain features which made them ultra-desirable for acquisition because simply put, nobody else could do what they were doing and if you bought the company, you bought the legacy approval, and further shut down competition.
Looking at those who have this approval, one glaringly obvious point is that unless you are in Silicon Valley, you won't get access, and even if you didn't launch your product yet, just being in Silicon Valley and having the right connections is enough to get you access.
One of these companies is OpenArt.ai which has two founders from Silicon Valley and has been operating for about 3 months. Another is PlaygroundAI.com which has only recently launched with founders in Silicon Valley. Those who have access to this technology at this stage have a tremendous advantage over those who don't. Different rules seemingly apply for those in Silicon Valley regardless of if your company is established or even out of beta.
OpenAI seemingly continues to give advantages to their friends and those in Silicon Valley compared to the rest of the world, only giving others access when those who were in the right place and at the right status, regardless of if they had a business or not at the time have finished the race and are already taking a shower. This is unfortunately the Silicon Valley Bias in AI access and everybody seems to simply ignore the fact it happens, and it happens time after time.