I had an experience with AI almost identical to Dr. Malone's. "
I had listed this, along w/ several other FBI blunders, on X: "1993 World Trade Center bombing (FBI had advance knowledge of bombing, and bungled a plan to substitute harmless powder for explosives)"
I was challenged on it, and sought to find some reporting on it. I had always…
I had an experience with AI almost identical to Dr. Malone's. "
I had listed this, along w/ several other FBI blunders, on X: "1993 World Trade Center bombing (FBI had advance knowledge of bombing, and bungled a plan to substitute harmless powder for explosives)"
I was challenged on it, and sought to find some reporting on it. I had always considered Brave to be a neutral search engine that protects privacy, and did a search there. This is the first sentence of a verbose AI-generated response: "There is no credible evidence to support the claim that the FBI had advance knowledge of the 1993 World Trade Center bombing or that they bungled a plan to substitute harmless powder for explosives."
The rest of the response was a lot of mumbo-jumbo about the known facts, but no acknowledgement of any info that supported my statement.
It's not a slam dunk, b/c the informant was recording his phone calls with the FBI w/o their knowledge. He made the accusation that they had ignored his offer of the substitution during a recorded phone call, and they neither confirmed it nor denied it.
Anyway.... I replied the Brave that though it said their was NO evidence, there in fact appeared to be SOME evidence, and included the link. Here's the first sentence of the reply: "The 1993 World Trade Center bombing was a terrorist attack that occurred on February 26, 1993, in New York City. According to the New York Post report, the FBI had an informant, Emad Salem, who was involved in the case and had warned the FBI about a potential bombing plot. However, the report also suggests that the FBI may have had some level of involvement in the bombing, with Salem claiming that the FBI had asked him to testify and that they had considered using a "phony powder" to fool the conspirators."
Again, a lot more additional verbiage, including the thing that Brave found persuasive: "It is worth noting that the FBI has denied any wrongdoing in the case, and the exact nature of their involvement is still a matter of debate. "
Well heck! The FBI denied it, so it must be false!
I've been having a running argument with my brother who thinks AI is awesome. He uses it to help him write software, and it's been a powerful tool for him. I wrote this to him: "Surely you already know, Google cooks its searches to make progressive thinking look good and conservative thinking look bad, or simply conceal facts that don’t support a woke narrative. When an LLM is “learning” it’s basically scouring the internet looking for related information, collects it and assigns weight, and then generates summary conclusions. But the primary source, that is, the entire internet, is biased. The bias is built in. Plus, the LLM developers are mostly woke, so their biases are embedded. "
My brother replied back, everybody knows it's biased, but he still advocates strongly for it.
I think AI can be a useful tool to knowledgeable people who know how to properly frame an inquiry, and who can evaluate the response. Unfortunately it's going to be used by people who can barely write their own name, and will take the AI response as if it were brought down from a mountain inscribed on stone tablets.
I had an experience with AI almost identical to Dr. Malone's. "
I had listed this, along w/ several other FBI blunders, on X: "1993 World Trade Center bombing (FBI had advance knowledge of bombing, and bungled a plan to substitute harmless powder for explosives)"
I was challenged on it, and sought to find some reporting on it. I had always considered Brave to be a neutral search engine that protects privacy, and did a search there. This is the first sentence of a verbose AI-generated response: "There is no credible evidence to support the claim that the FBI had advance knowledge of the 1993 World Trade Center bombing or that they bungled a plan to substitute harmless powder for explosives."
The rest of the response was a lot of mumbo-jumbo about the known facts, but no acknowledgement of any info that supported my statement.
In the meantime, I found through another search engine a NY Post article, at least in part confirming that FBI had ignored an informant who offered to substitute harmless powder for explosives. https://nypost.com/2023/02/26/1993-world-trade-center-attack-was-fbis-biggest-bomb/
It's not a slam dunk, b/c the informant was recording his phone calls with the FBI w/o their knowledge. He made the accusation that they had ignored his offer of the substitution during a recorded phone call, and they neither confirmed it nor denied it.
Anyway.... I replied the Brave that though it said their was NO evidence, there in fact appeared to be SOME evidence, and included the link. Here's the first sentence of the reply: "The 1993 World Trade Center bombing was a terrorist attack that occurred on February 26, 1993, in New York City. According to the New York Post report, the FBI had an informant, Emad Salem, who was involved in the case and had warned the FBI about a potential bombing plot. However, the report also suggests that the FBI may have had some level of involvement in the bombing, with Salem claiming that the FBI had asked him to testify and that they had considered using a "phony powder" to fool the conspirators."
Again, a lot more additional verbiage, including the thing that Brave found persuasive: "It is worth noting that the FBI has denied any wrongdoing in the case, and the exact nature of their involvement is still a matter of debate. "
Well heck! The FBI denied it, so it must be false!
I've been having a running argument with my brother who thinks AI is awesome. He uses it to help him write software, and it's been a powerful tool for him. I wrote this to him: "Surely you already know, Google cooks its searches to make progressive thinking look good and conservative thinking look bad, or simply conceal facts that don’t support a woke narrative. When an LLM is “learning” it’s basically scouring the internet looking for related information, collects it and assigns weight, and then generates summary conclusions. But the primary source, that is, the entire internet, is biased. The bias is built in. Plus, the LLM developers are mostly woke, so their biases are embedded. "
My brother replied back, everybody knows it's biased, but he still advocates strongly for it.
I think AI can be a useful tool to knowledgeable people who know how to properly frame an inquiry, and who can evaluate the response. Unfortunately it's going to be used by people who can barely write their own name, and will take the AI response as if it were brought down from a mountain inscribed on stone tablets.