I think it might be more the opposite!
They could be staying away to not intrude into Microsoft's "turf". MS is also pushing lots of AI integrations.
17,811 karma · joined June 4, 2016
I think it might be more the opposite!
They could be staying away to not intrude into Microsoft's "turf". MS is also pushing lots of AI integrations.
Sorra, Atlas, buying Sky.
There have been signs that Apple is going with Anthropic instead of OpenAI for integrating AI features.
This may partially be an aggressive strategic push to take the wind of out such a potential partnership.
Sora 4.5 tends to randomly hallucinate odd/inappropriate decisions and goes to make stupid changes that have to be patched up manually.
Install an empty SIGINT signal handler (without SA_RESTART), then run the loop.
When the thread should stop:
* Set stop flag
* Send a SIGINT to the thread, using pthread_kill or tgkill
* Syscalls will fail with EINTR
* check for EINTR & stop flag , then we know we have to clean up and stop
Of course a lot of code will just retry on EINTR, so that requires having control over all the code that does syscalls, which isn't really feasible when using any libraries.
EDIT: The post describes exactly this method, and what the problem with it is, I just missed it.
* This looks relatively convincing.
https://avherald.com/h?article=52e80701
https://x.com/oeingoboeing/status/1979588783982133360
* They obviously would have cleaned up the wounds... this can be a few hours old , surface wounds build crusts quickly
I just think that effects of the AI bubble bursting would be at most a symptom or trigger of much larger geopolitical and financial shifts that would happen anyway.
The US admin has been (almost desperately) trying to prop up markets and an already struggling economy. If it wasn't AI, it could have been another industry.
I think AI is more of a sideshow in this context. The bigger story is the dollar losing its dominant position , money draining out into Gold/Silver/other stock markets, India buying oil from Russia in Yen, a global economy that has for years been propped up by government spending (US/China/Europe/...), large and lasting geopolitical power balance shifts, ...
These things don't happen overnight, and in fact over many years for USD holdings, but the effects will materialize.
Some of the above (dollar devaluation) is actually what the current admin wanted, which I would see as an admission of global shifts. We might see much larger changes to the whole financial system in the coming decades, which will have a lot of effects.
An AI bust would take the stock price down a good deal, but the stock gains have been relatively moderate. Year on year: Microsoft +14%, Meta +24%, Google +40, Oracle +60%, ... And a notable chunk of those gains have indirectly come from the dollar devaluing.
Nvidia would be hit much harder of course.
There is a good amount of smaller AI startups, but a lot of the AI development is concentrated on the big dogs, it's not nearly as systemic as in dot com, where a lot of businesses went under completely.
And even with an AI freeze, there is plenty of value and usage there already that will not go away, but will keep expanding (AI chat, AI coding, etc) which will mitigate things.
2008 was a systemic breakdown rippling through the foundations of the financial system.
It would lead to a market crash (80% of gains this year were big tech/AI) and likely a full recession in the US, but nothing nearly as dramatic as a global systemic crisis.
In contrast to the dot com bubble, the huge AI spending is also concentrated on relatively few companies, many with deep pockets from other revenue sources (Google, Meta, Microsoft, Oracle), and the others are mostly private companies that won't have massive impact on the stock market.
A sudden stop in AI craze would be hard for hardware companies and a few big AI only startups , but the financial fallout would be much more contained than either dot com or 2008.
It is slower, but the results are much more often correct and it doesn't rush into half-baked solutions/dumb approaches as eagerly.
I'd much rather wait 5 minutes than have to clean up manually or try to coax a model into doing things differently.
I also wouldn't be surprised if the slowness was partially due to OpenAI being quite resource constrained. They are repeatedly complaining about not having sufficient compute.
Bigger picture: I think all the AI coding environments are incredibly immature. There are many improvements to be unlocked.
This is a on-demand managed container service with a convenient API, logging, global placement in 300+ locations, ...
AWS Lambda is probably closer in terms of product match. (sans the autoscaling)
Depending on what you do , Sandbox could be roughly on par with Lambda, or considerably cheaper.
The 1TB of included egress alone would be like 90$ on AWS.
Of course on lambda you pay per request. But you also apparently pay for Cloudflare Worker requests with Sandbox...
I reckon ... it's complicated.
Ideas, concepts, implementation plans are first written down as a proposal, which is read by others and discussed online. Meetings are only required if there are blockers to resolve, or differences in opinion.
The "parted ways" phrasing would probably annoy me quite significantly. We didn't "part ways", they fired me.
A lot of large companies put their servers into Equinix data centers.
Rising tide lifts all boats.
Specs for other languages are also for a specific version/snapshot.
It's also a specific version of a compiler that gets certified, not a compiler in perpetuity, no matter what language.
It seems like OpenAI is trying to turn Sora into a social network - TikTok but AI.
The webapp is heavily geared towards consumption, with a feed as the entry point, liking and commenting for posts, and user profiles having a prominent role.
The creation aspect seems about as important as on Instagram, TikTok etc - easily available, but not the primary focus.
Generated videos are very short, with minimal controls. The only selectable option is picking between landscape and portrait mode.
There is no mention or attempt to move towards long form videos, storylines, advanced editing/controls/etc, like others in this space (eg Google Flow).
Seems like they want to turn this into AITok.
Edit: regarding accurate physics ... check out these two videos below...
To be fair, Veo fails miserably with those prompts also.
https://sora.chatgpt.com/p/s_68dc32c7ddb081919e0f38d8e006163...
https://sora.chatgpt.com/p/s_68dc3339c26881918e45f61d9312e95...
Veo:
https://veo-balldrop.wasmer.app/ballroll.mp4
https://veo-balldrop.wasmer.app/balldrop.mp4
Couldn't help but mock them a little, here is a bit of fun... the prompt adherence is pretty good, at least.
NOTE: there are plenty of quite impressive videos being posted, and a lot of horrible ones also.
It seems like OpenAI is trying to turn Sora into a social network - TikTok but AI.
The webapp is heavily geared towards consumption, with a feed as the entry point.
The editing aspect seems very secondary.
Generated videos are very short, with minimal controls. The only selectable option is picking between landscape and portrait mode.
There is no mention or attempt to move towards long form videos, storylines, advanced editing/controls/etc, like other products in this space.
Seems pretty clear they want to turn this into AITok.
Case in point: I'm seeing much more success in LLM driven coding with Rust, because the strong type system prevents many invalid states that can occur in more loosely or untyped languages.
It takes longer, and often the LLM has to iterate through `cargo check` cycles to get to a state that compiles, but once it does the changes are very often correct.
The Rust community has the saying "if it compiles, it probably works". You can still have plenty of logic bugs of course , but the domain of possible mistakes is smaller.
What would be ideal is a very strict (logical) definition of application semantics that LLMs have to implement, and that ideally can be checked against the implementation. As in: have a very strict programming language with dependent types , littered with pre/post conditions, etc.
LLMs can still help to transform natural language descriptions into a formal specification, but that specification should be what drives the implementation.
So please wait a bit - subsequent runs will be fast, since compiled Python will be cached.
We unfortunately didn't get the final release out quite in time...
curl https://get.wasmer.io -sSfL | sh -s "v6.1.0-rc.5"
wasmer run python/pythonCodex / Jules etc make this pretty easy.
It's often not a sustainable pace with where the current tooling is at, though.
Especially because you still need to do manual fixes and cleanups quite often.