Hi everyone!
I'm not sure if I'm doing this right, sorry in advance if overstepping! This is not a self promotion but I have published myself an article that discuss this alternative proposition for AI regulation that aims to update the Internet itself to better cope with AI services and products.
It's called RoboNet Artificial Media Protocol or RAMP for short, and basically proposes that AI internet traffic should be bounded to the RAMP protocol instead of the Internet's legacy HTTP, so that society may avoid being "victim" of AI revolution by having a proper strategy in place for today's and tomorrow AI challenges. The concept can be visualized as the attached images and the entire article is available at my personal substack page on this link: https://antoniomax.substack.com/p/ramp-robonet-artificial-media-protocol
RAMP creates not only this distinct regulatory instrument for most things AI, but it also enables conditions so that we may avoid impacting the HTTP Internet with most types of AI oriented policymaking, a much more powerful governance instrument that aims to make the Internet a more coherent and human place.
It may be a different idea but the article covers the topic in a very concise way. I personally like the idea that it does not have to be a product, a law, a business or another monopoly, it is merely an update to the Internet, one that can enable a better future for us all.
And nobody said we can't never update the Internet right? Specially when we need it to accommodate things that are beyond its initial design, such as the functional mass automation that AI enables. Apparently we can prevent AI misuse to be imposed on us, RAMP concept aims to address that!
update June 2023: the link for the article has been updated to its latest version.
- Prihlásiť sa na účely uverejňovania komentárov
Pripomienky
Hi Antonio, really liked this idea! Having a dedicated network protocol for AI can open doors to policy and regulation enforcement.
However, I think it still misses the part of solving one of the main concerns that GenAI introduces: humans miss-using (or abusing) AI...
For example, humans generating offensive or malicious content , than simply copy-pasting and posting as their own, or via HTTP app (unless I am missing something).
Still, a step in the right direction! Great job!
Uri
- Prihlásiť sa na účely uverejňovania komentárov
In reply to Hi Antonio, really liked… by Uri Rosenberg
Hello Uri,
Thank you for providing an example! I'll be back to that in a second, but if you allow me, I'd like to elaborate on why RAMP was created:
RAMP aims to be apolitical, free from monopolies providing certification schemes, vendor locked validation solutions and regulatory captured entities. This fact grants RAMP the same apolitical environment HTTP have, to offer AI provenance for ALL AI services, no matter its capabilities or geographical origins. This allows RAMP to be treated equally by both end users and service providers, removing the AI "problem ownership" famous internet companies hold today over how AI is consumed and regulated.
RAMP delivers AI content, its job is to say 100% backed by market standards that what it is being delivered as AI is AI without relying on any third party to do so. RAMP doesn't create another industry, another intermediary, RAMP *is* meant to be AI's delivery method and that's its proposition, to make AI delivery independent from corporations and politics, by providing an optimal regulatory scheme that as you mentioned, can better help enforcement.
I needed to say this as RAMP isn't meant to directly solve problems, its existence is meant to enable an independent provenance mechanism for AI, where solutions can be built over it. RAMP existence provides solutions, yes, but the fun happens when RAMP is used!
I took your example to the heart as many people have the same doubts as you have, RAMP is a completely new way to address these situations so I created an entire Annex for RAMP's article to demonstrate how RAMP may be used to provide safer provenance mechanisms for the specific use case of text generated by systems such as OpenAI's ChatGPT, as this is where most people's attention is right now.
There are two scenarios explored there:
- one, where as you say, a person copy/paste content from a GenAI service and post as their own. Spoiler alert: they got busted;
- another where a person generates malicious content with GenAI, writes it down in a piece of paper without copy/pastes of any kind, types it all in someone else's phone, and still gets found;
In the Annex I explore the underlying technologies required to make this work with a proven methodology and clear semantics, you can read it on RAMP's article page right under its Conclusion chapter. Thanks for stopping by!
https://antoniomax.substack.com/i/125560909/annex-ramp-and-genai-text-challenges
- Prihlásiť sa na účely uverejňovania komentárov