Hacker Newsnew | past | comments | ask | show | jobs | submit | Rekksu's commentslogin

He's also somewhat active on twitter: https://x.com/pinboard


What does "SF crime" have to do with the content that is expected on hackernews?


presumably they can do that without criminalizing speech


that's only specifically the first 10 amendments, which are generally referred to as the bill of rights as they were added to the constitution when it was ratified and cover most basic freedoms so they're taught in school

other rights-granting amendments are the post civil war ones which are slightly less well known but also covered in school


Nah I often see people talking about stuff like prop <some number> expecting everyone to know what that refers to.


Most Americans won't know those, because those are state-specific (usually for California; most states don't even have Propositions that citizens can vote on).

Every American knows what the 1st Amendment is, by contrast.


Can an LLM base64 encode an arbitrary string? I don't think so but conceivably the rules are learnable


Yes, it can. ChatGPT is already able to do it. It's good enough that you can then use ChatGPT to decode it which will fix small errors in the output assuming the input is normal words.


I have never seen an ML researcher claim that understanding the effect of specific training inputs on outputs is straightforward given the size of these LLMs. Most view it as a very difficult if not impossible problem.


And yet it's a major part of the overall concept of being responsible with our use of AIs. Throwing our hands up in the air and prematurely declaring defeat is not an option long term.

It's a non-starter for no other reason than potential copyright infringement means the government becomes involved, and they will stomp on the AI mouse with the force of an elephant - the opinions of amateurs and the anti-copyright movement notwithstanding.

As such, AI Observability is a problem that's both under active research, and the basis for B2B companies.

https://censius.ai/wiki/ai-observability

https://towardsdatascience.com/what-is-ml-observability-29e8...

https://whylabs.ai/observability

https://arize.com/


Observability is great but it doesn’t give granular enough insights into what is actually happening.

Given a black box you can do two things: watch the black box for a while to see what it does, or take it apart to see how it works.

Observability is the former. Useful in many cases, just not here.

If you want to know what LLMs are actually doing, you’ll need the latter. Looking at weight activations for example, although with billions of parameters that’s infeasible.


Those companies are not solving the problem you are describing


"No they're not" and "no it's not" (simplified from the actual response) are conversation enders, so I'll follow the lead and let this conversation end.

Have a great new years!


Well it helps to be accurate in your statements


Neither of those are verbatim copies, do you have an example of a verbatim (i.e. exact, non-transformative) replica?


Do you have a source for this claim? This isn't really how generative models work


https://news.ycombinator.com/item?id=34187038 points to the two specific claims WRT stable diffusion. The co-pilot ones are easily found. Quake 3's "fast inverse square root" is the best known instance.


Genuinely ludicrous to say that cryptocurrency has historically been talked up by Democrats; similarly, Republicans have historically been vastly more trusting of the financial industry


I mean ludicrous that since 2009 it's Democratic thinking to bail out the banks, sweep inflation concerns under the rug, spend more money, funding unprofitable tech companies etc. I never said Republicans aren't in on the deal, just that it's obvious one side leading the charge on those things.


Ignoring the factual errors in your post, it's really not addressing the point on cryptocurrencies.


It's an opinion about democratic thinking, and a general assessment in correlating level of education in finance/econ to party bases. I have no idea what "facts" you are referring to. Just wow.


The bank bailout was 2008, proposed and lobbied for by the Bush administration. It passed the House with support from 45% of Republicans and the Senate with support from 69% of Republicans. Generally considered bipartisan, it almost certainly saved the financial industry. It's not clear what you mean by "a general assessment in correlating level of education in finance/econ to party bases".

https://en.wikipedia.org/wiki/Emergency_Economic_Stabilizati...

https://clerk.house.gov/evs/2008/roll681.xml

https://www.govtrack.us/congress/votes/110-2008/s213


Yep, and so began the golden age of fraud we now experience, and the backdrop for this SBF story. 100% agree that the bailout was bi-partisan and necessary though.


> I mean ludicrous that since 2009 it's Democratic thinking to bail out the banks, sweep inflation concerns under the rug,

Again, more ludicrous tribalism. The Great Recession bailout was all orchestrated under Bush. Trump famously was hounding Powell to lower rates before the pandemic, which would have had even more disastrous inflation consequences, just to make the stock market go up.


I think your mistake is to look at bought-and-paid-for politicians, I'm talking about the general public. My guess is you don't know many republican voters and therefore unqualified to to make the comparison. Most democrats I know think the bailout "paid for itself" lol


> To ensure that we make progress on that roadmap, we are increasing our investment in product, engineering, and design, which means decreasing our spend on other ares of the company.

Underrated dynamic in the startup layoffs this year. Many software companies grew headcount rapidly in areas outside of prod / eng in 2021 and are now scaling back those roles while preserving the talent that was extremely difficult to hire.


Their career page really bears this out. Lots of open engineering positions.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: