Skip to main content





You’ve granted some of the 25th’s drone team’s wishes. You’re helping them churn out birds at a heroic pace. byteseu.com/401455/ #Ukraine #Україна


An LLM will continue to engage in a “conversation” comprised of nonsense -- long past the point where a human would have abandoned the discussion as pointless.

This can be demonstrated experimentally by having an LLM-based chatbot “converse” with simpler text generation bots that would be inadequate to fool humans.

This article examines how a chatbot based on an open-source LLM (Llama 3.1, 8B version) reacts to attempts to get it to engage in endless exchanges with the following four basic text generation bots:

• a bot that simply asks the LLM the same question over and over
• a bot that sends the LLM random fragments of a work of fiction
• a bot that asks the LLM randomly generated questions
• a bot that repeatedly asks the LLM what it meant by its most recent response

conspirator0.substack.com/p/ba…

in reply to Chuck Darwin

Yeah. That's one of the points that make them so dangerous: no result is (almost) never an option. The AI will always "try" to produce a result ... even if it is utter hallucinated BS. That's how it works. To avoid that from happening, you would have to actively code "no-go scenarios" (like MS did with Bing CoPilot regarding questions an the 2020 presidential elections 🙄) ... Good luck with that: this Planet produces BS questions in exponentially higher magnitudes than programmers.
in reply to Chuck Darwin

I’ve done this by running experiments on laptop-based and local privateGPT, asking the same question 10 times in a row and never getting the same answer, including factually incorrect ones.

This is not a conversation with a start, context, purpose and a conclusion. This is a question response tool producing, not generating, any possible outputs from statistical word associations based on a specific vector space or corpus. It does not actually generate (cause or procreate) anything.

This entry was edited (4 months ago)






This entry was edited (4 months ago)


Last year, before I made a proper SDL renderer, #OpenCrystalCaves had a sprite viewer that showed the sprites as emoji. I've now resurrected that code and found a (cursed) use for it...

#SourceCodeSunday #CrystalCaves #DOS #OpenSource #GameDev






South Australia is aiming for 100% renewable energy by 2027. It’s already internationally ‘remarkable’ byteseu.com/401430/ #Australia


Venezuelan opposition candidate González has left the country for asylum in Spain


The departure of Edmundo González, who Venezuela’s opposition and several foreign governments consider the legitimate winner of July’s presidential race, was announced by Venezuela's vice president.

#news #npr #publicradio #usa
posted by pod_feeder_v2



After six years of development and $70M spent, Black Myth: Wukong is a hit whose initial sales run has outpaced Elden Ring and recent Zelda games.

This is in contrast to Sony’s Concord which was shit down within 2 weeks of launch after almost eight years of development.

AAA game development is now as expensive and hit-driven as movie making while taking even longer to bring to market. It’s no surprise the industry is in turmoil as this isn’t sustainable.

bloomberg.com/news/articles/20…

in reply to Dare Obasanjo

It's the dream of creating games as flashy and cinematic as hit movies that is unsustainable. It's the dream of raking in tens of millions or more in profit from games that is unsustainable. Creating fun games with a small team and good, fun mechanics is still very much sustainable.
in reply to Dare Obasanjo

The successes and failures aren't random, though. Black Myth: Wukong gave people what they wanted. Concord did not. One problem with the games industry is that executives are making decisions not based on what their customers want but rather on the monetization models that are most attractive to CEOs

Anyone who actually knew games knew a year ago that if properly executed, Black Myth: Wukong would be a hit and that Concord would fail.


in reply to Hacker News

@_L1vY_

1984: “ELIZA”, you’re kind of cute for a few rounds, but getting “what did you mean by ‘x’?” over and over again gets pretty old, pretty fast.
2024: Welcome back, “ELIZA”. You’ll be very helpful when trying to deal with pesky gen-“AI” bots.



in reply to Manni

Sadly, had you not mentioned where this was, It would be all but impossible to narrow it down.



I got the chance to #interview the #developer behind #junkstore - it gives you a way to play your redeeemed and purchased #epicgames and #gog games on your #steamdeck via a plugin.

The developer is so lovely, and quickly became one of my favorite projects on #Linux and #Steam.

reddit.com/r/SteamDeck/comment…

I hope someone here finds it interesting, I think the important thing in #gaming is to remember the people behind the apps and programs we love so much!