HackerLangs
TopNewTrendsCommentsPastAskShowJobs

zmmmmm

20,975 karmajoined 16 anni fa

comments

zmmmmm
·20 ore fa·discuss
Yes it has completely turned me around - was all in on Anthropic but now it just looks too risky. Better off leaning into open models. Even if I found a way to work with the restrictions as they are, who is to say they won't suddenly change tomorrow. It's not worth it.
zmmmmm
·ieri·discuss
> It envisions delivery of “20 mission-ready aircraft” by 2031.

Hmm, I'm not sure they fully address the problem if that is what is being proposed. The world will be an entirely different place by 2031 and 20 drones is .... meaningless? Surely they should be talking in the thousands or tens of thousands.
zmmmmm
·l’altro ieri·discuss
Good to see Meta finally back to releasing something at least worth evaluating. And it sounds like they did at least a bit skate to where the puck is going by focusing on tool and computer use.
zmmmmm
·l’altro ieri·discuss
Yes and Zuck effectively disbanded the entire team that did that. Not saying we shouldn't cast a critical eye on it, but it probably does warrant a second chance.
zmmmmm
·l’altro ieri·discuss
I don't care about the politics, I wouldn't trust anything made by Musk.
zmmmmm
·l’altro ieri·discuss
Hard not to reflexively reference the XKCD here. I think the authors of most charting languages would say they look good by default. It seems more likely this is achieving a subjectively different presentation than something objectively better. Higher level implies information loss. So it can only be better if it is doing so by assuming some better defaults. But then you have to ask if it lost expressiveness.
zmmmmm
·3 giorni fa·discuss
Like a lot of the rest of the world they would probably rather take the alternative option and accelerate the transition to clean energy. Has the upside of not handing more power to an authoritarian state on the other side of the Atlantic that clearly hates them and routinely threatens them.
zmmmmm
·3 giorni fa·discuss
And yet the will badger you endlessly to the point their photos app is near unusable to turn on auto sync which slurps up every photo and makes it very awkward to then delete them after. To me, this makes Google a liable party even if real CSAM is stored.
zmmmmm
·5 giorni fa·discuss
I would love to know that inside story. The whole saga is starting to look like one of the biggest own goals in history - Meta went from being widely respected and considered a peer with leading frontier labs to having no competitive technology. How a company seemingly willfully threw away a leading position in the most valuable tech race of all time should be a business case study, apart from a technology one.

I do have a theory : Llama3.1 marks the point where Zuck got seriously interested and took over the reigns in driving the work. From the minute he started directing things instead of considering the AI work as a quirky side project, things went downhill. He tried to force a huge scale up in Llama4 which didn't work. Then as we know he disbanded the whole team and brought in a new crowd of mercenaries who may or may not have had the technical skills but they came into an organisation in disarray and still driven by Zuck himself who is continually forcing decisions that are not well founded in the science.

All the above is an entirely evidence free fan fiction version of things, but I would be completely unsurprised if it is true.
zmmmmm
·8 giorni fa·discuss
To me a lot of the anti-short leash sentiment is reflective of the low accountability SWE have always had for their output. Software devs seem to strongly reject the concept that it isnt ok to ship defective products and fix later. It will be interesting to see if it persists as incidents start to occur due to fully automated code.
zmmmmm
·10 giorni fa·discuss
I can't decide if this will make science better or be the death of it. The potential wave of slop about to hit journals is frightening. Essentially what happened with GitHub code reviews is about to hit academic peer reviews and it isn't going to be pretty.
zmmmmm
·10 giorni fa·discuss
It's honestly quite baffling that the EU would want to put any more power in the hands of any US controlled company at this point. The US is a borderline hostile state, only recently threatening to invade Greenland among numerous other examples. The situation with Anthropic has illustrated that the US government will not hesitate to leverage power over US companies when it feels its interests are advantaged by doing so. If anything, the EU should be banning use of Google or Apple dependent architectures, not pseudo mandating them.
zmmmmm
·23 giorni fa·discuss
That's a great write up.

The one thing I feel it seems to under estimate is the likelihood of improvement. Even the authors acknowledge it's not even worth comparing local models from a year ago to what we have now. In fact, people widely see Opus 4.5 in November last year - 8 months ago - as the first time agentic coding became viable broadly viable even with frontier hosted models.

So why would we lock in hard on any concept at this point of what a local model is and isn't good for? Whatever it is right now, it probably won't be that in a year. It might be naive optimism to think we'll ever get to long horizon tasks with models that run on consumer / pro grade hardware. But so far the naive optimists are winning.
zmmmmm
·23 giorni fa·discuss
> Engineering for the sake of engineering has no value to the economy

I think that's the adventure we're on now. If recreating something is low cost, what is the value in investing in designing it well in the first place? We can empirically discover issues and the the AI to address them.

I certainly routinely find in supervising what the LLM is writing that it's making terrible internal design choices and correct them. Usually things one level up from code. "This will cache every image on the client and cause a huge amount of bloat. Change it to pull the image in real time from the server" kind of stuff. You do slowly build that up in the project documentation - "Never store unnecessary data on the client: we assume they are using low powered devices without substantial storage". But it takes time and the road to discovering that empirically is through a lot of unhappy users.

So I think there is still a lot of room for genuine engineering - that is, at the technical design level. Levels up from that - code structure etc - are much less clear. I am guessing that over time we will heavily optimise code written by AI for maintenance by AI. Which may be mostly about matching the context window to the code module size. Factoring something to 5 modules may be less of a good idea if it means the context window has to hold all of them for the LLM to work. But that is the path of discovery we are on which history tells us is a 20 year journey.
zmmmmm
·23 giorni fa·discuss
Now you get not just the 5 LoC to review but a 5 page essay to read in the form an auto-generated review as well. Which makes the submitter even more indignant when you start nit picking things about how it's implemented.
zmmmmm
·26 giorni fa·discuss
I've tried so many of these and paid for a lot of them and I still can't find what I want. It sounds like this is closer than most:

- record and separate two sides of the conversation

- save meetings in a simple transcription format in a local folder

- connect with my calendar (Outlook, Google Calendar) and name meeting transcripts accordingly

- for recurring meetings, append rather than create a new transcript

- let me label speaker voices and recognise those voices across different meetings

A tool that did all this and then ALSO built a knowledge base to let me RAG query my meetings would be the holy grail for me.
zmmmmm
·26 giorni fa·discuss
> they would have started verifying the identity of their customers.

Very good point. Yes i think this part goes to hubris. Amodei probably didn't think the ban would cut along those lines if it happened. And in fact it wouldn't surprise me if the government specifically made it that way (singling out foreign nationals) as a way of punishing Anthropic for putting them in this position. It's clear they absolutely hate being dictated to by anybody, but especially Amodei and they probably thought through what would hurt them a lot to implement and deliberately made it that way.
zmmmmm
·26 giorni fa·discuss
I think it's a near universal phenomenon that people with extraordinary amounts of power become victims of their own hubris. Once you get sufficiently decoupled from the consequences of your own actions, it is near impossible to tether yourself to a calibrated sense of reality.

So I genuinely think that Amodei thought here that he was building a moat - set a very high bar for safety at exactly the line Anthropic but nobody else meets, and then declare anything less to be too unsafe to be allowed. That would put a permanent halt to open models, Chinese models and throw a significant barrier in front of competitors - if OpenAI is about to release something competitive with Mythos, they would have to immediately double back and implement at least equivalent safeguards. It might cost them months at the most critical juncture in Anthropic's history, when they are filing for IPO.

Having said this, I am sure they calculated in the possibility of their own model being restricted. They probably still see it as a win because it acts as a strong endorsement of them as the market leader and the model as the most powerful available model. So I think both things are true, but we are in the "plan B" scenario now rather than "plan A".
zmmmmm
·26 giorni fa·discuss
it's pretty interesting in the sense that Amazon went far out of their way to build a secure process around this:

https://docs.aws.amazon.com/bedrock/latest/userguide/data-pr...

In theory, there currently isn't any technical mechanism for the retention Anthropic is requiring here to happen. Either Amazon is not being truthful (unlikely) or Anthropic is requiring them to significantly alter the architecture they have built around hosting of their models. But then doing so would put them in strict violation, I would imagine, of the commitments they have made to enterprises - many of which themselves are mandated into meeting these requirements by their own legal or regulatory requirements.

There are missing pieces I can't reconcile here - but one answer is simply that there aren't answers and that would be why Jassy is EXTREMELY pissed at Anthropic right now.
zmmmmm
·27 giorni fa·discuss
Seems rather stingy - 6 months is barely longer than you will get on a free signup deal for a lot of online products anyway. Kind of worse than nothing if it causes you to adopt work patterns that aren't sustainable for the project after the offer ends.