alt Hacker NewsI predict in future the best frontier models will be gatekept solely to the wealthy.
Replies
If the frontier models will take as much money to train as they do now, there is no way the wealthy are able to afford their training just for their own consumption. Financing of this whole thing rests on the models being available to companies and consumers who are willing to pay astronomical (compared to other software) sums for it.
It's always been this way ever since the first industrial revolution.
I can very easily see a licensing requirement coming soon. Running a higher-grade AI will require a govt-issued license, which involves a six-month application process, explanations of why you need to run it, where it's going to be stored and who will have access to it, pretty much the same as non-USA countries deal with firearms.
Reasoning? More customers = more revenue, there's negative financial incentive behind restricting TAM. In the same way the iPhone isn't restricted to only the wealthy, that would significantly reduce total goods sold and thereby revenue. In addition that creates lower economies of scale, lower network effects etc.
Pay $1,000,000 per business function you want to build.
Businesses will gladly pay it.
Individuals will be locked out and unable to compete.
Oracle will be able to "prompt a new smartphone" for $500M or whatever and enter the market to compete with Apple and Google. You and I can't afford that and won't be able to compete.
Hyperscalers will hyper scale even faster.
They'll port Linux to Rust, remove the GPL, and have all kinds of new entirely proprietary OSes. They'll be attested, signed, and gradually we'll lose open hardware. Thin clients with binary blobs, highly encrypted, no control, only leased to us.
Within a generation nobody will be able to program or own devices that can program.
That's the scary scenario.
I'm praying that China survives this BS and remains the bastion of AI model openness and freedom of choice. Can't believe I just wrote that.
I realize these models are locked up pretty tight and terabytes in size, but in a future like that, I don’t see them not being leaked via an insider. The weights have to be loaded into VRAM at some point.