Both Washington and Beijing now treat AI less like software and more like oil or nuclear technology. Once that happens, export controls are almost inevitable. The real question is whether open weight AI make those controls increasingly irrelevant over time.
Open weight AI only allows you to tune the parameters and run it locally, where open source AI also gives you the training data and stuff that was used to make the model itself.
That distinction is totally correct in principle, but in practice almost no frontier model releases its training data because of licensing, privacy, and sheer scale. Today, Open Weight AI has become the common compromise: useful for research and local use, even if it falls well short of the traditional open source definition.
For me, I’m going to be one of those hardliners who refuses to use closed source AI no matter how much better than open source it is.
Back when I was first learning about computers and operating systems and software, I did not care so much about open source or understand it and why it was important, so therefore I’ve made decisions that now I have trouble getting away from some closed source programs, and I refused to do the same thing with AI that I did back then.
Oh yeah, sure, and you can absolutely run open-weight models on your own hardware, which is definitely an improvement from having to use an API from a company. But for me personally, I won’t touch even open-weight models because they’re not fully open source, and fuck that.
Both Washington and Beijing now treat AI less like software and more like oil or nuclear technology. Once that happens, export controls are almost inevitable. The real question is whether open weight AI make those controls increasingly irrelevant over time.
Open wait AI? No. Open source AI? Yes.
Open weight AI only allows you to tune the parameters and run it locally, where open source AI also gives you the training data and stuff that was used to make the model itself.
That distinction is totally correct in principle, but in practice almost no frontier model releases its training data because of licensing, privacy, and sheer scale. Today, Open Weight AI has become the common compromise: useful for research and local use, even if it falls well short of the traditional open source definition.
Look at OLMO from AI2
There are lots of awesome OSAI! But today, they are completely outperformed by OWAI and closed AI. My guess: that won’t change in future.
For me, I’m going to be one of those hardliners who refuses to use closed source AI no matter how much better than open source it is.
Back when I was first learning about computers and operating systems and software, I did not care so much about open source or understand it and why it was important, so therefore I’ve made decisions that now I have trouble getting away from some closed source programs, and I refused to do the same thing with AI that I did back then.
I can totally understand that. Personally, however, it’s particularly important to me at the moment to switch from the cloud to local hardware.
Oh yeah, sure, and you can absolutely run open-weight models on your own hardware, which is definitely an improvement from having to use an API from a company. But for me personally, I won’t touch even open-weight models because they’re not fully open source, and fuck that.
For me, it’s open source or GTFO.