sunami-ai

Ouch. Yes. I agree.

What I'm asking really is what is the benefit of TS if we could use decent and improving coding AIs to help us code in Rust, even if we are new to it, or never used it, and compile to WASM. Much faster execution per my experience. I mean not even comparable.

I do all my work in Rust now via o3-mini-high and convert to WASM... JS just for DOM and event handling. What is the point of building these CPU-bound functions in TS. Why not Rust->WASM?

Answer: The British.

[flagged]

My view is that those who designed CSS suck when it comes to designing intuitive systems.

I would not advocate constraint solvers, in the future I hope ViTs are so cheap to run that they can infer the right layout of things at any orientation and size in single digit milliseconds, solving the layout problem for good =)

Read the reasoning after clicking in the stats or swipe to next page and click on the donuts. For in-house legal teams not for mobile users. Read the statements and reasoning. Tell me if the reasoning is wrong. One of the top 100 lawyers in US looked at it and told us he likes how it balances the intent of the law and no gripes with reasoning. Not using a reasoning model unde the hood. We built one for legal. It means that gen AI is useful and LLMs can be made to reason with inference time scaling. It’s not rocket science but also not easy. What hype? We think gen AI is actually under hyped.

Sorry it’s for corp legal users, so desktop only Not for mobile and never tested on Firefox.

This is all BS tbh. People don’t know how to use the current gen AI take to do very useful reasoning.

I keep posting our work as an example and NO ONE here (Old HN is dead) has managed to point out any reasoning issues (we redacted the in-between thinking most recently like the thinking traces that people were treating as the final answer)

I dare you to tell me this is not useful when we are signing up customers daily for trial:

Https://labs.sunami.ai/feed

I found Gaussian Processes with the right kernel to be very powerful with even just a few data points and a very small set of parameters. I don't know if I was using it correctly tbh, but it worked out great in predicting values that I could not predict so accurately. I used it as a predictable yet non-linear process to tweak the input in a computer vision task. The proof was literally in the pudding.

The sensitivity can be turned up or down. It's why we are asking for input. If you're talking about the Disney EULA, it has the context that it is a browsewrap agreement. The setting for material omission is very greedy right now, and we could find a happy middle.

A former attorney general is taking it for a spin, and has said great things about it so far. One of the top 100 lawyers in the US. HN has turned into a pit of hate. WTF all this hate for? People just really angry at AI, it seems. JFC, Grow up.

LLMs don't reason the way we do, but there are similarities at the cognitive pre-conscious level.

I made a challenge to various lawyers and the Stanford Codex (no one took the bait yet) to find critical mistakes in the "reasoning" of our Legal AI. One former attorney general told us that he likes how it balances the intent of the law. Sample output (scroll and click on stats and the donuts on the second slide):

Samples: https://labs.sunami.ai/feed

I built the AI using an inference-time=scaling approach that I evolved over a year's time, and it is based on Llama for now, but could be replace with any major foundational model.

Presentation: https://prezi.com/view/g2CZCqnn56NAKKbyO3P5/ 8-minute long video: https://www.youtube.com/watch?v=3rib4gU1HW8&t=233s

info sunami ai

She is a hardcore human.

Lawyers are not all bad, as I'm finding out.

We met some amazing human lawyers on our journey so far.

LOL, says the one with hurt ego.

Yes, it used to work with just one post, that I get normal responses, not pure hate. HN has fallen to haters.

I built this AI tool to spot "bugs" in legal agreements, which is harder than spotting errors in research papers because the law is open textured and self contradicting in many places. But no one seems to care about it on HN, Gladly, our early trial customers are really blown away by it.

Video demo with human wife narrating it: https://www.youtube.com/watch?v=346pDfOYx0I

Cloudflare-fronted Live site (hopefully that means it can withstand traffi): https://labs.sunami.ai/feed

Free Account Prezi Pitch: https://prezi.com/view/g2CZCqnn56NAKKbyO3P5/

I spent $10 on a bad apple and it got $0 back. Then I go and sell that bad apple for $10. Whomever bought it is giving me back my money. This is kindergarten level arithmetic.

Part of me thinks that that before they gave the $130M they knew exactly how they'd get it back (thru some pre-arranged M&A) if it doesn't work out. Or at least that would be the smart thing to do.

Making Transformers the same cost as CNN's (which are used in character-level ocr, as opposed to image-patch-level) is a good thing. The problem with CNN based character-level OCR is not the recognition models but the detection models. In a former life, I found a way to increase detection accuracy, and, therefore, overall OCR accuracy, and used that as an enhancement on top of Amazon and Google OCR. It worked really well. But the transformer approach is more powerful and if it can be done for $1 per 1000 pages, that is a game changer, IMO, at least of incumbents offering traditional character-level OCR.

Older coders/technical folks tend to have more wisdom than raw compute (relative to younger coders who may have more raw compute than distilled wisdom.) Wisdom takes a more reliable and more efficient path than raw compute.

Both raw compute and wisdom will be eventually replaced by AI, but "deep wisdom" is largely held in the body, in the way we react viscerally to things, which AI as it is envisioned today does not factor in at all. So we still have a refuge in the wisdom stored in our body memory.

Eric Schmidt is too old to reason correctly.

Funny you say that. Back in former life, we built a distributed alarm and monitoring system for AT&T central offices with no single point of failure. So it's like HVAC can be taken offline easily at critical facilities, with backups and backups of backups.

Maybe you should prototype and get an investor?

This showing up here at the same time articles about the push for tourism in Saudi Arabia showing on major news outlets.

Brain organpids do not mimic the brain, unless you are blind and cannot google. They are the brain.

Have you heard of brain organoids?

A Google search yields "three-dimensional tissues that mimic the human brain and are grown in a lab"

The distinction between natural and what is man-made (aka artificial) is itself artificial.

We are learning how to recreate nature, whether from silicon or in 3D tissue or ab initio (google "synthetic life")

Any technology that's advanced enough is indistinguishable from magic, or something along these lines...

Anything a single human can do, reasoning wise, AI will eventually be able to do.

Anything emerging out of a collective of humans interacting and reasoning (or interacting without reasoning or flawed reasoning) the AIs (plural) will eventually be able to do.

Only thing is machine kind does not need sleep, does not get tired, etc, so it will fail to fully emulate human behavior, with all the pros and cons of that for us to benefit from and deal with.

I'm not sure what is the point of a theoretical discussion beyond this.

Worst part is that the patterns of implementation won't be consistent across the pieces. So debug a whole codebase that was authored with LLM generated code is like having to debug a codebase where ever function was written by a different developer and no one followed any standards. I guess you can specify the coding standards in the prompt and ask it to use FP-style programming only, but I'm not sure how well it can follow.