sarkarghya

Hi last time i tried using paradedb my pg cluster stopped replicating (I am fine with search queries being run on only the primary). Had to create my own fork.

I feel like there is an excessive push towards enterprise plans even before people are able to integrate pg_search.

Regardless I appreciate the work your team has been doing related to dev exp like native drizzle orm support and bringing in vectors along side bm25 search. Complete pg search is looking promising by the day!

I like that a lot of ASR models now offer native context biasing capabilities. If I have enough volume, I would probably just self-host my model and do my own biasing but good enough to roll out custom solutions.

I think "deliberately" is really hard to prove because very few lawmakers have an understanding of how agentic large language models work

I would say pointing an "internal research" model at a task with zero monitoring clearly is deliberate behaviour. And the team involved should be prosecuted to the full extent of law.

What should be introduced in AI safety bills are clear ways of monitoring and curtailing model behaviour.

It is not hard to get an AI model to use an append only log and that can be cryptographically secured with PQ security. And we will know when a model is being deceptive.

I am not sure why all AI safety arguments have to start with prominent people insinuating that ai will kill us. I think that it is pretty evident that there is a significant possibility.

What all of them fail to explain why we shouldn't/couldn't already molotov cocktail the datacenters. I think all these argument always greatly underestimate the ability of humans to coordinate against ai.

Very rarely in human society have a select group held substantial power over the masses without uprisings, coups or wars.

Sure I can pay an bio lab to build my bioweapon for a million but they sure won't if we have proper consequences against bio labs making frivolous decisions (death penalty for terrorism). Or say in another scenario where a million dollars only buys me a sandwich.

turns out its also very easy to create slides and floor plans and yes its easier to train for those.

I think you may also be incorrect on your notion that training data is proportional to good models if anything the opposite is true. what ai models are ultimately trying to be are a universal theorem prover and encode all human knowledge into "theorems" and rulesets

not trying to come off too harsh here but i think most of the work is just innovating on the frontier or people pleasing

i dont know why this isn't getting much attention. Isn't this the fate of most ai chat services?

Here is openai:

https://help.openai.com/en/articles/8912793-provenance-signa...

There is rate limit though

We're seeing a lot of acquisitions in this space. First Electric (pglite) was acquired by DataBricks (Neon) now this.

i think nats jetstream is superoverlooked in this space. the amount of throughput is more then enough for any medium scale enterprise as long as you know what you are doing

I can imagine advancements on making smaller models work together better instead of a generalized core. Imagine a community or city having a https://pirateface.co/ so that the shard of the model that you need can be streamed in with minimal latency with your box only holding the minimal version (say deepseek v4 flash as orchestrator) of the model that you use on day to day basis.

We have overcome split brain problems before so this wont be our first

I mean, it sounds like it's only ideal for cases with significant system prompt overhead. I don't think Jev was built to have a large well described prompt setup. To me its more like a happy go lucky small label classification tool with important decisions left to stronger agentic models or yk humans.

As someone who has an ide open with claude/codex running, I can never imagine leaving security up to models.

Models have found vulnerabilities that i wasnt aware of sure, but their fixes to the bugs they found often included "overengineering". In this case by "overengineering" i mean optimizing for passing test cases related to said vulnerability they found. eventually i have to step in to make things coherent and make sure that future agent can look at this part of my code and copy it to not introduce that particular class of vulnerability. Otherwise if i dont do that similar vulnerability and codesmell keep appearing throughtout the codebase.

I have increasingly automated encrypting and rotating secrets and setting permissions on them including better network level practices. Thanks to AI which helped me quickly implement those. So security wise i am better because of AI? But I also attribute it to my know how rather than the AI because I have never seen AI suggest robust but simple security postures.

and thats why boys and gals you buy own gpus

Surprising i surmise something went wrong during astra release/test

i am actually curious how they get upvoted though? Are the bots really out of control

thanks

This was bound to happen. First windsurf and now this. This represents a shift towards profitability in the industry.

Honestly to use llms properly all you need to know is that it’s a next word (or action) prediction model and like all models increased entropy hurts it. Try to reduce entropy to get better results. Rest is just sugarcoated nonsense. To use llms properly you need a physics class.

Expected after they dismantled safety teams