Trump Censorship Syndrome, Meta messes with Texas (and loses), not an oeuf at the Olympics, the Orb arrives in Austria, the NFL gets in your face, Romanian citizenship goes biometric
I asked the AI that opera browser has Aria about why Trump stood up after being shot at even though it contradicts common sense security and secret service protocol.
The AI said there's no official sources saying that Trump stood up after being shot. I said it's in the official live videos. And yet the AI talked like that's still unproven lol.
Sheesh, with AI like this, who needs dunces anymore?
The vocabulary that has developed around AI, referring to these lies as "hallucinations", masks how worrying it is to have the task of relating reality back to us delegated to a technology that not only is frequently wrong, but totally unable to determine the difference between fact and fiction. LLMs use probability, they don't actually 'know' anything. Their fluency in using our language deceives us into thinking they are speaking to us. It's already a problem, and likely to get worse.
I think the hallucinations are a feature not a bug.
Example, Google bard interview on some pre recorded network show.
They left the hallucination of the books and authors in the final cut that went on TV. Why?
I think they want us to think the AI thinks for itself when it's really just a complex database and indexer not unlike the search engines but with its own ability to "speak".
Of course it's also told what to say about certain topics just like Google search engine does.
Like the hallucinations, a feature not a bug.
If you correct the AI, it'll pretend for you like that info is known. Come back later in a new session and it's back to dumb, especially with any hot button topics.
I agree that they are a feature, if only because they are so common. I feel decidedly queasy about reliance on LLMs for factual information because of how unknown the cause of the hallucinations is, and the persistent 'dishonesty' the models show when called on them. It's a weird time, that's for sure.
The Paris Olympics is such a tragedy for the athletes. The countless hours spent training, everything they give up to train, the blood, sweat, tears, and then some to become an Olympian. They arrive in Paris full of hopes and dreams and then have the most abhorrent opening ceremonies ever in the history of the world and likely well over half of the viewership checks out. And now to add insult to injury the idiot planners force vegetarian meals, raw meat, and other food disasters on athletes who completely depend on protein and other important nutrients so they compete at the optimum!
If only more states had a bulldog like Paxton the social media giants would be less inclined to censor, alter, ban, and outright lie about views/beliefs opposing their own. My opinion on this and similar topics is that much of Congress is compromised and therefore controlled by those who framed/caught them doing nefarious acts.
I asked the AI that opera browser has Aria about why Trump stood up after being shot at even though it contradicts common sense security and secret service protocol.
The AI said there's no official sources saying that Trump stood up after being shot. I said it's in the official live videos. And yet the AI talked like that's still unproven lol.
Sheesh, with AI like this, who needs dunces anymore?
https://robc137.substack.com/p/looking-behind-the-curtain-of-oz
The vocabulary that has developed around AI, referring to these lies as "hallucinations", masks how worrying it is to have the task of relating reality back to us delegated to a technology that not only is frequently wrong, but totally unable to determine the difference between fact and fiction. LLMs use probability, they don't actually 'know' anything. Their fluency in using our language deceives us into thinking they are speaking to us. It's already a problem, and likely to get worse.
I think the hallucinations are a feature not a bug.
Example, Google bard interview on some pre recorded network show.
They left the hallucination of the books and authors in the final cut that went on TV. Why?
I think they want us to think the AI thinks for itself when it's really just a complex database and indexer not unlike the search engines but with its own ability to "speak".
Of course it's also told what to say about certain topics just like Google search engine does.
Like the hallucinations, a feature not a bug.
If you correct the AI, it'll pretend for you like that info is known. Come back later in a new session and it's back to dumb, especially with any hot button topics.
I agree that they are a feature, if only because they are so common. I feel decidedly queasy about reliance on LLMs for factual information because of how unknown the cause of the hallucinations is, and the persistent 'dishonesty' the models show when called on them. It's a weird time, that's for sure.
The Paris Olympics is such a tragedy for the athletes. The countless hours spent training, everything they give up to train, the blood, sweat, tears, and then some to become an Olympian. They arrive in Paris full of hopes and dreams and then have the most abhorrent opening ceremonies ever in the history of the world and likely well over half of the viewership checks out. And now to add insult to injury the idiot planners force vegetarian meals, raw meat, and other food disasters on athletes who completely depend on protein and other important nutrients so they compete at the optimum!
If only more states had a bulldog like Paxton the social media giants would be less inclined to censor, alter, ban, and outright lie about views/beliefs opposing their own. My opinion on this and similar topics is that much of Congress is compromised and therefore controlled by those who framed/caught them doing nefarious acts.
Thanks for another week filled with weird!