Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
Kostchei
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
6 ms
·
31.
▲
by
Kostchei
2y ago
Using the real world- as in vision, 3d orientation, physical sensors and building training regimes that augment the language models to be multidimensional and check that perception, that is the next step. And there is very little shortage
32.
▲
by
Kostchei
2y ago
Developer. Prompt Engineer. Philosopher-Builder. (mostly) not programmer. The code part will get smaller and smaller for most folks. Some frameworks or bare-metal people or intense heavy-lifters will still do manual code or pair-programmin
33.
▲
by
Kostchei
2y ago
I mean fkcu me when they have those things, however, maybe they are just lazy and their judgement is fine, for a lazy intelligence. Inner-self thinks "why are these bastards asking me to do this? ". I doubt that is actually happen
34.
▲
by
Kostchei
2y ago
OP runs 300 essentially free forums. It's just too much.
35.
▲
by
Kostchei
2y ago
<insert manic laughter> I mean that might happen, but why get funding? Why not move to market immediately? Without debt. Get in the thick of it. Just do. You want a wage or do you want a product that does the thing. Because sometimes
36.
▲
by
Kostchei
2y ago
This my technique- test the output, make sure it works (in the sense of outputs) the way I want. Test the input edge cases, move on. Occasionally when I can't get it to do what I want the llm suggests things like logging of output betw
37.
▲
by
Kostchei
2y ago
verisimilitude, not coherency. The real world doesn't appear coherent, why would a fantasy one. The consistent application of style and rules- that gives verisimilitude, and that's what counts.
38.
▲
by
Kostchei
2y ago
They (mega) may have started to do this, but for several years I used to come across saved text files with lists of of mega links on most pedophiles computers. A different unit in the police sometimes checked the links and anecdotally they
39.
▲
by
Kostchei
2y ago
not to defend its "you must accept updates" insane /inane fail, but, the suite of crowdstrike inc falcon stuff we have enables the response side of EDR pretty well, and for a mixed windows, linux, mac shop, where we would lik
40.
▲
by
Kostchei
2y ago
As a user of Wiz (and I like what I get from them) ,I am relieved . 1. huge valuations are bad for customers- that investor cash has to come back from somewhere 2. Google has a habit of subsuming products into their stack and either sunsett
41.
▲
by
Kostchei
2y ago
This is the way. Robotics, self driving cars, mobile devices themselves, either phone or glasses.... They can generate the data needed for the next steps. (pun intentional) Text was easy, the process has been defined. Multi-modality impro
42.
▲
by
Kostchei
2y ago
he regularly does- and gets various threats from bad and dubious actors as a result. He and "have I been pwned" (troy hunt) are genuine positive actors with IRL stakes in the game.
43.
▲
by
Kostchei
2y ago
We have run that test.- generate random string(not by llm) names of values- ask the llm to do math (algebra) using those strings. Tests logic, 100% not in the data set GPT2 was like 50% accurate, now we up around the 90%.
44.
▲
by
Kostchei
2y ago
a generation of consultants is already doing that- look at the ruckus around PWC etc in Australia. Hell, look at the folks supposedly doing diligence on Enron. This is not new. People lie, fib and prevaricate. The fact the machines trained
45.
▲
by
Kostchei
2y ago
hmm. I ended up with a few 750+ line chunks of js, beyond the ability of chatgpt to parse back at me. So my go-to technique now is to break it into smaller chunks and make them files in a folder structure, rather than existing inside a si
46.
▲
by
Kostchei
2y ago
I speak English, not any code languages. I recently learned javascript this way - knowing English and other languages (Thai, Mandarin), I kept pasting my pseudo-code, asking GPT to convert it into javascript, and then iterating on that. Too
47.
▲
by
Kostchei
2y ago
This is my new capability- I am not a coder or a programmer and I can get things built, in my own time, at my own speed, solo. Would it be better code if someone with 3 years of university and 5 years of coding practice did it? Yes, very pr
48.
▲
by
Kostchei
3y ago
sure, but if the phone company says there was an sms, when you open the phone it looks like there is an sms, and cellebrite extracts one, which you then present in a nice power point to the court... then your accusation as to tapering with
49.
▲
by
Kostchei
3y ago
Depends what the incentives are. Some folks like the community- that doesn't change. A website like honestwargamer or goonhammer has a clan. Mostly because they are insightful and friendly. But if an AI scraped their content, it would
50.
▲
by
Kostchei
3y ago
Most of the arguments I hear from my mates-who-code about adding deterministic old school things like buttons and "ui" on top of LLMs is "sometimes it's faster and more specific"... Ok, sure, but go to voice to text
51.
▲
by
Kostchei
3y ago
Maybe, but I think it is more likely I will try a different type of project, a different stack. See if that is the only easy path. Try something with graphics (a visual map) or that uses the llm api (generate a narrative etc). But my mate w
52.
▲
by
Kostchei
3y ago
No different from "the team that built that is all gone, they left no doco, we assumed X, added the feature you wanted, but Y happened under load" , which happens a lot in companies pushing to market older than a minute. My defaul
53.
▲
by
Kostchei
3y ago
With no experience in java, no coding for 30years since pascal in high school, no previous use of git or github, no hands on experience of the azure stack... I stood up 4 static web apps that do things I want in my hobby in 4 weeks- the fi
54.
▲
by
Kostchei
3y ago
Are you kidding? "easy to stop"? When LLMs are integrated into law enforcement, banking, research, education, logistics... All these areas have people building backend systems leveraging the current llm tech and pipelines to plug
55.
▲
by
Kostchei
3y ago
yeh, this is my argument against people who say "LLM are dumb because x", er.. lets look at humans.. so maybe this is bottom 1% intelligence currently .. it's improving. What's the difference between a lot of personality
56.
▲
by
Kostchei
3y ago
No. Actual criminals are going much deeper than that. Custom hardware, custom apps, custom servers. Sometimes it works, sometimes they get this.
57.
▲
by
Kostchei
3y ago
because some of these networks- not the one you read about in the news, are LE resistant (servers in Panama etc). One or 2 are so tight that we (LE) can see use of the network as a marker of organised crime but not the messages. Reducing t
58.
▲
by
Kostchei
4y ago
Yeh- my feel is, language is the framework by which we developed reasoning and we used an organic NN to do it. And at scale an complexity approaching the human brain we get similar results
59.
▲
by
Kostchei
4y ago
Some folks grew up without practicing code- did business / operational /technical things for 20+ years. For someone like that, chatGPT is manna from heaven. 0>0> I point chatgpt at the general problem and ask about the tech.
60.
▲
by
Kostchei
4y ago
no reason not to be a dry drop. It would be frozen and no longer ... active... after a few minutes
More ›