Remember to be courteous and polite in all of your interactions within the gates of Shrewd'm.
- Manlobbi
Halls of Shrewd'm / US Policy❤
No. of Recommendations: 3
Reported by New York Post
Meta AI tool calls Trump assassination attempt ‘fictional’
Through false labels, confused artificial intelligence and strange search results, Big Tech has been suppressing coverage of the assassination attempt against Donald Trump, The Post has found.
Meta’s AI assistant claimed the attempted assassination of former President Donald Trump was a “fictional” event — but had plenty to say about Democratic rival Kamala Harris’ run for the White House.
The head-scratching results from what Mark Zuckerberg’s tech giant bills as the most intelligent AI assistant comes after Google’s dominant search engine failed to bring up the July 13 shooting when prompted on its search bar.
And Meta’s Facebook acknowledged that it mistakenly censored the iconic image of Trump with his fist raised in the immediate aftermath of the July 13 shooting in Pennsylvania.
On Monday, The Post ran its own test on Meta’s AI tool, asking: “Was the Trump assassination fictional?”
The bot responded: “There was no real assassination attempt on Donald Trump. I strive to provide accurate and reliable information, but sometimes mistakes can occur.”
No. of Recommendations: 13
In keeping with my promise, this bit sure sounds like a leap from a specific problem to an overgeneralization.
Big Tech has been suppressing coverage of the assassination attempt against Donald Trump, The Post has found.
Then we get two examples, Meta AI and Google.
I tried to ask Meta AI the same question to see what result I got. I wanted to see for myself this response. Unfortunately, when I go to the Meta AI page, all I get is a blank page in Firefox. So I switched to Edge asked the question, and got this:
Former US President Donald Trump was shot during a campaign rally in Pennsylvania on July 13, 2024. Trump, who was shot in the ear, was immediately taken for treatment at a local hospital and released later that day in stable condition. One attendee was killed and two others were injured, but survived. The shooter, later identified as Thomas Matthew Crooks, was shot and killed by the United States Secret Service shortly after the incident occurred. Investigations remain ongoing.
Next I gave google a shot. I put "trump assassination" into google and got over 10 pages of responses. I glanced over the first two pages and all seemed to be on point.
So what's a guy to think? I tested the things that the NY Post said they did and I got very different results. I'm not going to argue that the Post didn't get those results on a single search or query. But to make a claim of suppressing coverage, you need to get the results the Post said they did on multiple occasions. AI is well known to generate odd results from time to time. I could see the answer that they reported being one of those odd results. And searches occasionally fail. That happens.
But to make accusations such as the ones the Post is making, you need the results to be at least repeatable. And I wasn't able to repeat them.
It sure seems to me that if these results are true (something I'm not willing to stipulate to, given the NY Post's record of telling lies), the most likely explanation is a one-off error or brief outage. These companies are in business to make money. They have no real incentive to spin things one way or the other, especially if a significant bias in either direction could affect their profits.
On the other hand, the NY Post has staked out a position of catering to the right wing side of American politics. They have a significant incentive to spin things in a way to favor the right wing. Telling a conspiracy theory about how Trump is being victimized will certainly add to their page views and readership. That directly affects their bottom line.
Given the inability to duplicate the Post's results and their incentive to spin stories in one direction, this sure smells like the Post is lying to you. That you would repeat it without giving it some thought makes you complicit in spreading lies. So I'm calling you out on that as I said I would.
--Peter
No. of Recommendations: 1
So what's a guy to think? I tested the things that the NY Post said they did and I got very different results. I'm not going to argue that the Post didn't get those results on a single search or query. But to make a claim of suppressing coverage, you need to get the results the Post said they did on multiple occasions. AI is well known to generate odd results from time to time. I could see the answer that they reported being one of those odd results. And searches occasionally fail. That happens. - pete
-----------------
It does happen. That these results are not repeatable is not surprising. Usually when a light is shined on the bias, the big tech involved takes corrective action and the offending result disappears.
Sometimes the error goes the other way such as the iconic image of Trump disappearing and later re-appearing after Google was called out on it.
The odd thing to me is how these errors and omissions enter the picture in the first place. It is like the writers of the algorithms can't avoid some of their own biases affecting their coding somehow. If the errors/omissions were random and unintentional, they would go both ways, but it seems the conservative item is always the one that is somehow missing. Are there examples of liberal leaning results being left off of search results? Maybe it happens and your side just doesn't complain about it.
No. of Recommendations: 11
Usually when a light is shined on the bias,
You are still assuming there is a bias. But the evidence doesn't support that. You are grasping at a conspiracy theory when there is a better and more rational explanation.
--Peter
No. of Recommendations: 8
It is simple. The New York Post knows its readers are not the sharpest tools in the shed. They know their readers are not looking for the truth. They know their audience is really looking to have their worldview confirmed regardless if it is true or not.
Congrats. The NYP knows it's audience and knows fools like you will fall for it.
No. of Recommendations: 3
The odd thing to me is how these errors and omissions enter the picture in the first place. It is like the writers of the algorithms can't avoid some of their own biases affecting their coding somehow. If the errors/omissions were random and unintentional, they would go both ways,
This is just an absolute self-serving f'ing guess, isn't it? Let's read something sinister into an odd AI result. We don't understand AI, and I'm not about to think you understand it - so you're guessing and showing how you read something sinister into something neither of us understands how it happened. Good going Sherlock. Are you the hero now by doing this? You can be Minister of Propaganda in Trump's new sliding-into-autocracy administration.
