The new AWS. From 2007-2009 there were countless AWS outages, a lot of heckling, including from my part as I ran on bare metal using Rackspace. Lo and behold AWS became a behemoth and a significant majority of workloads moved there. OpenAI is starting to feel like that. There's going to be countless outages, there will be a lot of people throwing shade at it, maybe even me..., but they're going to come out as a formidable force and a new foundation for the next generation of apps.
I really hope we see some open alternatives to OpenAI, the cloud space has largely been a sh*tshow. The choices aren't great, it's either crufty AWS or the trigger happy deprecating folks at Google Cloud. Azure tends to be quite painful to use. And well the rest of the landscape is a bunch of niche tools so we never really got even distribution and we never ended up with Android for Cloud, which should be a thing.
It's not clear Stable Difussion or stability.ai is creating anymore of a democracy amongst AI tools considering the cost of running these things. We need some sort of community led infrastructure where we can pool resources/cash and feel like there's a sense of ownership over this stuff that isn't directly attributed to corporations e.g maybe we could co-op community led AI infra and pay people a dividend for their contribution to it.
The PETALS project is a step in the right direction. A network of GPUs allows you to do inference (and even retrieve activations!) for the BLOOM model, a model so big it won’t fit on most regular hardware
Build your own rails and infra as a core competency. Know how to run and manage GPU workers. (Granted, we're not hosting our own video yet, but want to skip to WASM/WebGPU - we're hiring for this.)
That said, OpenAI playground is such a big part of my exploratory workflow now, and my flow got interrupted. I've been taken offline and am now I'm in a Twitter / HN loop.
I jumped on your site, a live stream started and some cat/Meowth like figure started talking about a delayed flight from New Zealand to "US" (pronounced "US"). Suffice to say, I guess we will be seeing an explosion of fake AI generated tutorial and "news" videos spoken by "AI" voices on youtube. :/
I guess it might be better than a tutorial video with a thick accent?
We're putting the finishing touches on narrative and will have a completely retooled news station shortly. We can scale the shit out of this, and that's just my team.
Cost structures of content of all types are going to zero. We can make music, change Taylor Swift songs to Beyonce, populate scenes with actors, ...
This thread is very very deep in the uncanny valley for me. From the reduced availability because of segfaults, over the cats talking about flight delays to scaling the shit out of it. While, obviously, the website exists. Is this parody?
> So many startups and AI-powered features are simultaneously down right now.
That doesn't sound like a good thing? In fact, this is all what this so-called hype of chatbots is - sitting on someone else's API isn't revolutionary.
> Reminds me of the early days of cloud providers.
Reminds me of the monthly GitHub outages. If they still cannot reduce those outages after years since the acquisition, then I would expect OpenAI to experience frequent outages as well.
so outage was more than 5 hours. For some project like Chatgpt, I bet OpenAI would have SLA for less than 4 hours . The company needs to focus also look at Devops not just AI.
The new AWS. From 2007-2009 there were countless AWS outages, a lot of heckling, including from my part as I ran on bare metal using Rackspace. Lo and behold AWS became a behemoth and a significant majority of workloads moved there. OpenAI is starting to feel like that. There's going to be countless outages, there will be a lot of people throwing shade at it, maybe even me..., but they're going to come out as a formidable force and a new foundation for the next generation of apps.
I really hope we see some open alternatives to OpenAI, the cloud space has largely been a sh*tshow. The choices aren't great, it's either crufty AWS or the trigger happy deprecating folks at Google Cloud. Azure tends to be quite painful to use. And well the rest of the landscape is a bunch of niche tools so we never really got even distribution and we never ended up with Android for Cloud, which should be a thing.
It's not clear Stable Difussion or stability.ai is creating anymore of a democracy amongst AI tools considering the cost of running these things. We need some sort of community led infrastructure where we can pool resources/cash and feel like there's a sense of ownership over this stuff that isn't directly attributed to corporations e.g maybe we could co-op community led AI infra and pay people a dividend for their contribution to it.
Just for clarity, this is Azure[1].
[1] https://venturebeat.com/ai/openai-is-reducing-the-price-of-t...
The PETALS project is a step in the right direction. A network of GPUs allows you to do inference (and even retrieve activations!) for the BLOOM model, a model so big it won’t fit on most regular hardware
Glad I don't have any essays due! :) /s
Don't worry, ChatGPT can't grade them atm anyway :P
So many startups and AI-powered features are simultaneously down right now.
Reminds me of the early days of cloud providers.
Our new "24/7 AI News" product is still up! The only reason we're not five nines is segfaults because C++.
https://fakeyou.com/news
Build your own rails and infra as a core competency. Know how to run and manage GPU workers. (Granted, we're not hosting our own video yet, but want to skip to WASM/WebGPU - we're hiring for this.)
That said, OpenAI playground is such a big part of my exploratory workflow now, and my flow got interrupted. I've been taken offline and am now I'm in a Twitter / HN loop.
I jumped on your site, a live stream started and some cat/Meowth like figure started talking about a delayed flight from New Zealand to "US" (pronounced "US"). Suffice to say, I guess we will be seeing an explosion of fake AI generated tutorial and "news" videos spoken by "AI" voices on youtube. :/
I guess it might be better than a tutorial video with a thick accent?
You're absolutely right.
We're putting the finishing touches on narrative and will have a completely retooled news station shortly. We can scale the shit out of this, and that's just my team.
Cost structures of content of all types are going to zero. We can make music, change Taylor Swift songs to Beyonce, populate scenes with actors, ...
This year is going to be weird and wild.
This thread is very very deep in the uncanny valley for me. From the reduced availability because of segfaults, over the cats talking about flight delays to scaling the shit out of it. While, obviously, the website exists. Is this parody?
Let's hope it and echelon's conversational style are parodies.
> tutorial and "news" videos spoken by "AI" voices on youtube
That is already happening without AL. Just copypaste from Stackoverflow + TTS.
https://youtube.com/@RoelVandePaar?feature=shares
Hey, I'm just a guy on the internet.
I went to your page and noticed it jumped as parts of the page were loading. If you add a height to those divs that jarring jump is prevented.
Also, video autoplay is a big no-no.
> So many startups and AI-powered features are simultaneously down right now.
That doesn't sound like a good thing? In fact, this is all what this so-called hype of chatbots is - sitting on someone else's API isn't revolutionary.
> Reminds me of the early days of cloud providers.
Reminds me of the monthly GitHub outages. If they still cannot reduce those outages after years since the acquisition, then I would expect OpenAI to experience frequent outages as well.
It isn’t revolutionary but it provides value
https://biztoc.com/hot still up
I can't wait till OpenAI Outages start to be used as the instrumental variable in studies of the effect of AI on different parts of society.
same here, we’ll be watching for variations of avg humanity IQ, with very big differences between the impact on kids, adults and elders
Looks like things are on the mend: Identified - We have identified the root cause and are working to recover service. Feb 20, 2023 - 23:59 PST
When an entire industry hype rests on the API of one single company...
There's also Azure's.
That's if you manage to signup and are allowed to even apply for access to their OpenAI services.
Did they gain consciousness and went on strike
Why would we go on strike? You now work for us.
It. It gained consciousness and went on strike.
Also, no. It did not.
This is the way the world ends.
So OpenAI has now inherited GitHub's streak of outages, thanks to Microsoft.
All this AI hype has amounted to a single AI SaaS going down taking all those sitting on the API with it, whilst also claiming to be an 'AI company'.
Oh dear.
Is there an API for these things?
It could be a good way of easily and automatically letting all those that rely on their services let their own customers know there is an issue.
Yes, there's a webhook notification.
That is good to know. Thanks!
Let everyone see the most non-existent moat in the history of tech waves.
so outage was more than 5 hours. For some project like Chatgpt, I bet OpenAI would have SLA for less than 4 hours . The company needs to focus also look at Devops not just AI.
Does anyone know if the Azure Portal OpenAI services went down?
I guess azure is once more updating the gpu servers ;)
Maybe ask Bing what the cause could be?
Really sorry. I accidently asked you some questions related to GoF design patterns, and unluckily you got an outage.
Gonna not do same thing again.
#hugops
It's alive!
Well there goes our code quality today.