Lockal

43 ms·

271.
by Lockal 3y ago
Study: scientists prefer metrics which create an illusion that every time-consuming process is gradual and predictable
272.
by Lockal 3y ago
Context: https://twitter.com/realGeorgeHotz/status/177166635165378186... George Hotz: No need to hire me, just open source the 7900XTX firmware+docs and remove the signature check. We'd treat it like brin
273.
Zero-day exploit by Geohot to crash AMD GPUs on firmware level(github.com)
4 points by Lockal 3y ago|1 comments
274.
by Lockal 3y ago
This reminds me of 196x chatbot ELIZA, who always responded with universal phrases like "Men are all alike" or rephrased the question back.
275.
by Lockal 3y ago
Usually the right way is a life sentence. However, in some of the most successful states (in terms of life expectancy, crime rate, economy and visa-free access), such as Japan and Singapore, the right way is the death penalty.
276.
by Lockal 3y ago
17. You still can't read the twitter thread without having a twitter account
277.
by Lockal 3y ago
5. "you won't understand" or "you're not yet ready to comprehend" or "insufficient data for meaningful answer". When journalist spends half of the time speaking about safety and identificatio
278.
by Lockal 3y ago
The one case where simply deciphering the acronym doesn't help much. Let's say: "for viewing and editing files describing architectural objects and building interiors".
279.
by Lockal 3y ago
449 upvotes after 25 minutes. Malware inside?
280.
by Lockal 3y ago
1) You see recently visited apps on top of search results. Same applies for Google/Yandex (unless you turned off specific personalization settings). It is a feature to save your time from scrolling. 2) Post on Reddit is FUD from compet
281.
by Lockal 3y ago
Never attribute to stupidity that which could be explained by malice (especially if the malicious actor uses Hanlon's razor in defense).
282.
by Lockal 3y ago
tldr: last week author demonstrated that "AI" is random guesser. Now instead of feeding actual questions, author inputs: 3 - 1 - 2 2 - 3 - 1 1 - 2 - ? And "AI" responds that answer is 3 with high probability
283.
by Lockal 3y ago
I already prepared to make a rant with "yet another cool-hacker invented prompt injection or discovered how LLM works", but was pleasantly surprised that it was not the case
284.
by Lockal 3y ago
That's interesting how this appears right after much severe "I'll refrain from providing code that involve concepts as you're under 18" Gemini response[1]. By "severe" I mean that for Gemini user did not m
285.
by Lockal 3y ago
So... What? They must provide up-to-date mechanisms to stop malicious actors and find anomalies to protect against cyber-attacks. Fake signatures? Cropped photo submitted multiple times under different names? Photoshopped materials? AI-gene
286.
by Lockal 3y ago
I guess this statement is a cheap protection against cheap journalists. Otherwise by now all the tabloids would be full of scary stories about deepfake politicians, deep-porn and all types of blackmailers (by the way, there is so much compe
287.
by Lockal 3y ago
Given the optimizations applied to SDXL (comparing to SD 1.5), it is understandable why it outputs blurry backgrounds. It is not for safety, it is just a cheap way to hide imperfections of technology. Imagine 2 neural networks: one occasion
288.
by Lockal 3y ago
I wouldn't look for hidden reasons. Recent image generators are already too good with face generation (thanks to CelebA-like datasets and early researchers). And now the emphasis is on the multimodality of the model within a domain. Th
289.
by Lockal 3y ago
I may understand why Twitter algorithm may recommend such posts to other people. What I don't understand is why this over-sensationalist "ChatGPT has gone berserk" post with NO analysis whatsoever, a collection of Twitter scr
290.
by Lockal 3y ago
There is a clearly visible "Share" buttons in every ChatGPT discussion. It allows to anonymously share exact message sequence (it does not show number of retries, but that's the best you can show). If you see cropped ChatGPT
291.
by Lockal 3y ago
If you consider Telegram as a product to be a logical continuation of the VK message system, then all of these "features" existed. Restricting of incoming messages existed (cloned from Facebook as usual). Restricting of "last
292.
by Lockal 3y ago
Durov personally blocked Navalny channels in Telegram during 2021 elections - https://www.rferl.org/a/telegram-navalny-smart-voting/314662... even though "technically" as a foreign legal entity they had
293.
by Lockal 3y ago
Forensic investigators can reconstruct the face from a skull, not only race/ethnicity/gender. It is probably just a question, how to measure improvement of AI, added to existing method. Let's say you find the skull of the all
294.
by Lockal 3y ago
First time I see "_ has died" on the first page on HN without black bar (since it was introduced).
295.
by Lockal 3y ago
Hopefully they will eventually integrate SponsorBlock, otherwise language model seemlessly incorporates ads into summary...
296.
by Lockal 3y ago
Welcome to the train: Companies Say 'Layoffs' Without Saying 'Layoffs' https://news.ycombinator.com/item?id=39352293 In this week's episode: * Grammarly Restructures Business to Align With Company S
297.
by Lockal 3y ago
> they clicked "install" when prompted I just checked it, it opens a small Microsoft Store window with button "Get". In other languages they also used shady terminology to replace word "Install".
298.
by Lockal 3y ago
HN always has interesting flagging rate for links related to CCP.
299.
by Lockal 3y ago
To be more specific, "Wikimedia RU" was dissolved. I remember when they opened a legal entity, the lawyer said that one should not use the word "Russia" in the name of the company, as this requires permission from the Mi
300.
by Lockal 3y ago
In the summary you skipped few points: 1) First article mentions: "I have a couple of theories though. I set up Find My with an Apple ID that I don’t remember" 2) Second article mentions, that he attempted to report this as a vuln

More ›