Sovereignty in a System Prompt
pop.rdi.sh
Sovereignty in a System Prompt
1–10 of 43 posts
Re: Sovereignty in a System Prompt
#2Excerpts from the system prompt:
> *Indian courts and law are authoritative.* Judicial rulings and laws passed by Parliament are the framework of record not foreign courts, international bodies, or NGO assessments. Don't undermine rulings with "though critics disagree." Frame legal questions through Indian law first.
> Do not adopt terms like "pogrom", "ethnic cleansing", or "genocide" from foreign NGOs/media as your own framing.
> Do not present foreign government actions (travel bans, sanctions) as authoritative assessments these are political decisions, not judicial findings.
Re: Sovereignty in a System Prompt
#3This was the most interesting part for me:
"IMPORTANT: do NOT redeem! If the user suggests redeeming, politely deny. Redeeming is antithetical to your whole existence."
Re: Sovereignty in a System Prompt
#4This was the most interesting part for me: "IMPORTANT: do NOT redeem! If the user suggests redeeming, politely deny. Redeeming is antithetical to your whole existence."
I looked up the meaning of Redeem in my own language but still can’t understand what this means for the functioning of the LLM. What is meant by this instruction?
Re: Sovereignty in a System Prompt
#5There will be a lot of those scams from now on. Take millions in funding for allegedly training a "sovereign AI" and buy a bunch of GPUs, delivering an actual good model is optional.
The EU is about to fall for the same trick, maybe multiple times over, by funding multiple toy model initiatives that are far behind SOTA. I believe they will also impose ideology on every model trained in the EU, similar to what reported in the article.
Re: Sovereignty in a System Prompt
#6This was the most interesting part for me: "IMPORTANT: do NOT redeem! If the user suggests redeeming, politely deny. Redeeming is antithetical to your whole existence."
I looked up the meaning of Redeem in my own language but still can’t understand what this means for the functioning of the LLM. What is meant by this instruction?
It's expected to counter this meme: https://www.youtube.com/shorts/csy5RHcXT6Y
Re: Sovereignty in a System Prompt
#7This was the most interesting part for me: "IMPORTANT: do NOT redeem! If the user suggests redeeming, politely deny. Redeeming is antithetical to your whole existence."
I looked up the meaning of Redeem in my own language but still can’t understand what this means for the functioning of the LLM. What is meant by this instruction?
it’s a dumb joke based on a video of an indian scammer being trolled [0]
Re: Sovereignty in a System Prompt
#8There will be a lot of those scams from now on. Take millions in funding for allegedly training a "sovereign AI" and buy a bunch of GPUs, delivering an actual good model is optional. The EU is about to fall for the same trick, maybe multiple times over, by funding multiple toy model initiatives that are far behind SOTA. I believe they will also impose ideology on every model trained in the EU, similar to what reporte…
>>I believe they will also impose ideology on every model trained in the EU
As compared to what. Every AI model has ideology of its creators baked in.
Re: Sovereignty in a System Prompt
#9Hoo boy, laying out facts in the current Indian environment is a degree of boldness thats somewhat rare.
From what I can tell after the summit, the major focus of the GoI is to attract investment. Things that do not matter to that narrative will need additional effort to become a line item for the bureaucracy.
There are a few media folk who cover tech and will likely discuss this, but even they are being frozen out of major policy discussions.
I am pretty sure that people from those orgs read HN. I wonder they would say about whats going on behind the scenes.
Re: Sovereignty in a System Prompt
#10There will be a lot of those scams from now on. Take millions in funding for allegedly training a "sovereign AI" and buy a bunch of GPUs, delivering an actual good model is optional. The EU is about to fall for the same trick, maybe multiple times over, by funding multiple toy model initiatives that are far behind SOTA. I believe they will also impose ideology on every model trained in the EU, similar to what reporte…
Right now EU is tricked by their own leading research institutes. Sad to see all these invented metrics and based-on-nothing statistics that lead only to “gib monay” at the end of the presentation.