We deal with issues all the time that require multiple rounds of fact checking. That’s because there are so many bogus comments, stories, reports and studies that have surfaced that it makes us skeptical. There is no substitute for doing your own homework.
The latest “arbiter” of truth is AI. From our perspective, AI is a useful tool, but it is no more dispositive than any other resource. This is particularly true when it is generating interpretative responses, or making normative judgments about persons or events.
Here is one example where AI did not fare well. We sought to evaluate a comment recently made by Miranda Devine, a columnist for the New York Post.
In her April 2 column, Devine wrote the following:
“When a black cop, Michael Byrd, shot dead an unarmed white woman, Iraq war veteran Ashli Babbitt, on January 6, 2021 during the Capitol riot, Democrats conspired to hide his identity, pay him secret bonuses, and ensure he suffered no consequences.”
Asking Google AI to evaluate the veracity of Devine’s statement generated this response: “The claim is largely false or misleading, mixing verified facts with unsupported conspiracy allegations.” The response also notes “there is no evidence of a partisan conspiracy to hide his identity or provide secret bonuses.”
The AI response offers these points as evidence for its claims:
“Shooting Fact: U.S. Capitol Police Lt. Michael Byrd shot and killed Air Force veteran Ashli Babbitt on January 6, 2021, as she attempted to climb through a broken window near the Speaker’s Lobby while lawmakers were being evacuated.
“No Consequences/Clearance: The U.S. Department of Justice (April 2021) and the Capitol Police Office of Professional Responsibility (August 2021) concluded the shooting was lawful, within department policy, and did not violate criminal law. Byrd was not disciplined and was later promoted to captain.
“Hidden Identity & Threats: Byrd’s identity was not ‘hidden’ by Democrats, but rather withheld for months by Capitol Police due to ‘numerous credible and specific threats’ against him and his family after his name was leaked on right-wing websites.
“Secret Bonuses: While Byrd has been praised by his department, claims of ‘secret bonuses’ to reward the shooting are largely unsupported or misinterpret standard department retention or security actions.
“Pardon Claim: Claims that Joe Biden granted a ‘secret pardon’ to Byrd are false; Reuters reports that Byrd’s name does not appear on any official pardon lists.”
We investigated Google’s AI statement and found it wanting.
“Shooting Fact:”
This is an accurate summary.
“No Consequences/Clearance:”
As AI notes, Byrd was investigated by Capitol Police and the Justice Department for the shooting. Both investigations cleared him of wrongdoing. But as Judicial Watch president Tom Fitton points out, in most officer involved shootings the cop is usually investigated by a grand jury. In this instance, Byrd faced no such scrutiny. Rather, he was investigated by his own agency and the Justice Department. This may be defensible, but it is in stark contrast to how Black Lives Matter offenses are handled. Grand juries are typically demanded to ensure there is no cover-up. Apparently, AI cannot make this connection.
“Hidden Identity & Threats:”
AI tries to make a technical distinction between the Capitol Police and Congress. Yes, technically the Capitol Police withheld Byrd’s identity. However, as the attorney for the Babbitt family, Terrell Roberts, claimed, “The U.S. Congress wants to protect this man. He’s got friends in high places, and they want to protect him.” AI did not seem to believe this information was worthy of consideration.
“Secret Bonuses:”
It was later disclosed that Byrd received a $37,000 retention bonus, and that he received help accessing funds for officers wounded in the line of duty. But this was not well reported, and indeed it was kept under wraps for some time. The chairman of the Capitol Police Union complained that the bonuses were above what other officers received. This prompted the union boss to call on Capitol Police to either give everyone else the same bonus that Byrd received or force him to repay the excess amount.
“Pardon Claim:”
This is a red herring. Devine never said he received a pardon. Thus, no correction was needed. AI could just as well have told us that Byrd was suspended for leaving his loaded firearm in a public restroom in the Capitol, or that he was disciplined for shooting at a moving car while he was off-duty. In other words, AI not only makes a false comment, it makes no mention of relevant facts.
AI is not a light switch that either works or it doesn’t. It is a resource that requires discernment, and a willingness to probe further.
In the showdown between Miranda Devine and AI, she wins going away.
