"Agents" can't think and LLMs aren't sentient. They aren't suited to be your coworker, but they also aren't suited for generation computational tasks. The chat interface is all that there is and their behavior in chat is not deterministic or bounded enough to be useful in most applications. They mimic tokens in reply to the tokens you give them, and that is all.
You know what's a bad idea from an engineering (that thinky thing we used to do as part of building software) perspective?
Building a dependency on an expensive remote API into your system.
This isn't just me bloviating, I've been down this road before. In my case I had a project using LLMs to automatically edit videos provided by Hollywood content owners. It seemed like a decent application, but LLMs are structurally unsuited for dealing with user data like this. The way that the prompt is evaluated means there is no separation between system and user input, so once you start dealing with a wide variety of topics you pretty quickly run into walls.
One example - ChatGPT refusing to summarize and pick a top segment from a news program because it contained references to a murder-suicide, and both murder and suicide are included in the many prohibited topics that are filtered in ChatGPT replies. This was through their API, not the regular user interface, so it is in theory as unrestricted as access gets. But because the LLM cannot be trusted to behave properly around the topic, they have to filter anything which touches it.
Structurally, I don't see a way this can be overcome - LLMs by design mix the entire prompt together, it's not like a parameterized SQL query where you can isolate the user and system data. That means that a long or bold enough user input is often enough to outweigh the system prompt, and that causes the LLM to veer into unpredictable territory.
> I half expect the entire Trump family to move to Dubai in 2029.
Russia, maybe Israel. Not Dubai. Dubai will remain too closely tied to the next administration in the US without a major change in our energy supply. But yes I think it is highly likely that many of the criminals in this administration and the trump family will flee the country and take their pilfered millions with them once they are out of power.
Ok. Since this is a very precedented and normal action, please link me to an article or press release from when Obama, Clinton, Bush 1, Bush 2, or Carter decided the fire the entire NSF oversight board.
Yeah that is concerning. Glad the US doesn't have any sort of credit scoring system that might make it hard to get out of poverty. That would be really scary. Imagine if you had to pass a credit check to get an apartment!
Senate Democrats have made at least 5 attempts to stop this. Every Democrat voted in favor except Fetterman. Every republican opposed the attempt except Ron Paul.
The Republicans are entirely responsible for the war in Iran - they started it and have opposed every attempt to reign the administration in. Don't play this "both sides" game when one side is clearly causing the issue here.
I mean. The side voting while knowing they can't stop it that way, the definition of optics, thought this was worth it to not cater to their anti-war voters, no?
Both side bad yeah, you seem to think the solution is still to cling to the lesser of two evils. Hell, dem presidential candidates were praising Trumps Iran policy up until the bombs started falling. There is no evidence they _actually_ care about a war with Iran outside of, just as you do, saying OH this is THEIR doing.
Look, I would love to reply to this but you don't even articulate a coherent position here.
It seems like you trying to claim that the democrats actively opposing and trying to stop the war are somehow just as culpable as the republicans who are actively supporting the war just because the democrats don't have enough power to actually stop it? I want to be generous in my interpretation here, but I can't make heads or tails of your statement.
No-one supports the war now. But Palestine just had to go? It can't be about _war_ or loss of life, and the people caring can get f-ed we don't need your votes.
But sure one side is extremely against anything like this but unfortunately only get to demonstrate it when in opposition and unable to do anything.
The difference might be there in regards to Iran, with it's global market impact. But it's not there in regards to human life or suffering. That was made _very_ clear during the electoral campaign. But now it's _very different_ because it's "them" doing it and we can say we were always against this.
We have to say "us good, them bad" because we're voting the same way next time, and we're not planning to vote for something bad, right?
> But it's not there in regards to human life or suffering. That was made _very_ clear during the electoral campaign.
I don’t know what to say but to call this retarded at best or evil at worst.
Trump ran on a platform that included purposefully harming people. He continues with it. Look at his response to Mueller’s death for instance.
Inb4 you mention something about a genocide in Guzah, and ignore how Trump is backing everything Netanyahu wants done in the region compared to the democrats attempts at holding him back.
> What would the alternative have been? Not reporting that they were purchased at all?
They could have purchased any other vehicle, from any other manufacturer. Any other heavy duty truck would be a better choice than the Cybertruck, which is impractical and poorly designed.
> The premise here seems to be that spacex purchased the trucks for the sole purpose of inflating the tesla reporting, which of course makes no sense from a business perspective
It's almost like the owner of SpaceX, which is doing quite well right now, made a decision designed to benefit the owner of Tesla, which is struggling to sell a bunch of 2-ton chrome shitboxes. Wonder what those two individuals have in common. I'm sure that anyone, even you, will be able to figure it out with a little googling.
Sure man. Nothing says "hacker ethos" like arresting a citizen for going 3 seconds over their allotted free speech time.
There are so many bootlickers here on HN these days. A sign of the times, I suppose.
But of course, what else can we expect? This is the natural consequence of putting ethics and morality after money. Money always wins. And once you start seeing the world like that, through the lens of "success makes right", you have to bend your view of reality to make it square. And then you wind up here, defending actions like this.
Again, that is not what happened. He went over time. He left the podium. He stopped and talked to the council. He was asked to leave. He refused. The police asked him to leave. At that point he made the decision to leave in handcuffs rather than in the normal fashion.
Asking him to leave sounds likely to be an abuse, but it's a far cry from "arresting a citizen for going 3 seconds over their allotted free speech time."
Where's the bootlicking? I'd the article chooses to lie about why he was arrested for no reason (and try to paint him as "a farmer" rather than a known activist), then people are going to focus on the untruths and wonder what else the article lies about.
There was no need for this article to lie. The actual events were already worth being upset about. It only serves to distract.
> BTW, the J6 protesters were all tracked and identified by their cell phone data.
Many of the insurrectionists were also caught on camera in congress after they broke down the doors and stormed the building. Some even took selfies in the offices of various senators and house reps.
It's all part of this administration's strategy to set the stage for next time. By pardoning violent criminals, they make it clear that they endorse political violence. Now, when he incites a mob to interrupt the elections next time he loses - in 2026 or 2028 - everyone in the next mob will know that their actions will be pardoned.
I've been building an HLS streaming engine with Elixir. It takes care of asset segmentation, transcoding, and streaming. It supports regular VOD playout, as well as live streaming by dynamically building HLS playlists from a variety of sources, including transcoded VOD assets and other HLS livestreams. It has a basic scheduling system and I'm integrating a lua engine to allow dynamic scheduling using user-provided scripts.
I'm hoping to continue extending it until it can act as a full internet TV delivery stack like Pluto or Roku TV. It still needs to be behind a CDN for efficient delivery but basically any CDN would work.
You know what's a bad idea from an engineering (that thinky thing we used to do as part of building software) perspective?
Building a dependency on an expensive remote API into your system.
This isn't just me bloviating, I've been down this road before. In my case I had a project using LLMs to automatically edit videos provided by Hollywood content owners. It seemed like a decent application, but LLMs are structurally unsuited for dealing with user data like this. The way that the prompt is evaluated means there is no separation between system and user input, so once you start dealing with a wide variety of topics you pretty quickly run into walls.
One example - ChatGPT refusing to summarize and pick a top segment from a news program because it contained references to a murder-suicide, and both murder and suicide are included in the many prohibited topics that are filtered in ChatGPT replies. This was through their API, not the regular user interface, so it is in theory as unrestricted as access gets. But because the LLM cannot be trusted to behave properly around the topic, they have to filter anything which touches it.
Structurally, I don't see a way this can be overcome - LLMs by design mix the entire prompt together, it's not like a parameterized SQL query where you can isolate the user and system data. That means that a long or bold enough user input is often enough to outweigh the system prompt, and that causes the LLM to veer into unpredictable territory.
reply