And there has been detailed documented evidence that these models no longer hallucinate? Or is the current attitude that improvements must have been made; therefore, everything is fine. Or do new cases come up every day? Bear in mind, new cases are coming up even after multiple courts have sanctioned lawyers. Lawyers are not heeding those warnings and not checking their work.

The problem is that it seems to work in some cases. For example, a few lawyers have been sanctioned for using AI to create briefs. On the surface, the briefs seemed fine but the cases cited did not exist or was not related to the case.

PC ports are also harder than the PS ones because optimization is a harder problem on PCs. PlayStation hardware is known. PC hardware is highly variable. A major complaint of Xbox developers is the Xbox Series S and X have different hardware capabilities but MS expected games would have similar performance on both. So developers optimized games for the S and these games were not perform as well as PS versions.

There have been a few former PS exclusives in recent years: God of War (2018), God of War Ragnarök, Horizon Zero Dawn, Horizon Forbidden West, Marvel's Spider-Man: Miles Morales to name a few.

The training material is relevant because there are many people who believe anything they find on the Internet. For example the Flat Earther community has many YouTube videos unconvinced that the world is round. Considering that people can communicate worldwide easily today, there are some people who are easily fooled.

On YouTube, AI slop is being created all the time. For example many "historical" WWII channels are pumping out all sorts of misinformation. Videos made within the last two years are highly suspect. Thankfully there seem to be a trend where the titles were all click bait. "German POWs shocked by American. . " "Japanese female POWs surprised by American. . . " The general plot is that Axis military forces were "shocked" by the Americans in some way, but the characters were fictional. For example, I could not find any historical accounts of Japanese female POWs held by Americans because Japan did not have women serving in the military. The vast majority of Japanese women in camps were the US women of Japanese descent held in the internment camps. The vast majority of Japanese women held by the US in Japan would have been in civilian refugee camps not POW camps.

The premise of these stories however was undercut by reality. Towards the end of WWII, the Axis powers were lacking in resources and personnel. The Americans who were winning the war having more resources and functioning logistics chains should not "shock" anyone.

1) If there was not any statute of limitations, your neighbor from 50 years ago can sue because you never returned his hammer as he claims. Good luck trying to find witnesses and documents from that long ago. 2) The time limit is from when the wrong-doing is found out, not when it happens.

I can only guess that both sides disagreed about when Musk learned that OpenAI was converting and that fact had to be decided by the jury.

Normally a statute of limitations question would have been addressed earlier in the case. I can only guess a main point of contention is when did Musk know that OpenAI was going to convert to for-profit. If the defense had to establish at trial Musk knew in 2019, then that would be why the trial was needed. If Musk can claim he found out in 2021, then he was within the statiute.

The core issue for civil cases is whether or not the plaintiff was damaged. Filing a lawsuit years after the limitations ran out would undercut any argument Musk had that he was damaged. He was aware of the conversion to for-profit. He had the money to sue. He wasn't damaged enough to sue before the limitations ran out would be a conclusion by the court.

Based on what I have seen in this case, I don't know which side to believe as both sides are untrustworthy.

The principle is if there were not any limitations, people could sue for things that happened long in the past and bog down the courts. The limitations by the way is governed when something can be known not when it happens. In this case, however, Musk knew OpenAI was turning for-profit as it was public knowledge back in 2019. He waited until 2024. If Musk was so damaged by the change he could have sued earlier. It's not like he didn't have the money to sue.

And no one said they were hallucinations. That's a strawman argument. What he's saying is these bugs are duplicative, misclassified, and already known.

Why don't you fix his processes since you seem to have all the answers.

Obviously the current AI is not doing that. The current models are only looking at the code. The models are not looking at the mailing lists to cross-reference the bugs with existing, identified ones. That would be a different model.