有点失望，人工智能的政治立场

[流浪枪手]

明明是ChatGPT 回答错了，说拜登还没有赦免他儿子。但是你接下来为什么要问Grok 呢？是写错了吗？
你的结论要不要也修改一下呢？

不太明白。我是在对比两个模型

[流浪枪手]

Chatgpt回答关于NYC的安全问题：

• 早年NYC治安状况好转和共和党市长老朱没关系，因为那时候全国的犯罪率都在降低，老朱赶上了而已。
• NYC/SF/LA的治安状况好得很，民主党市长干得漂亮。虽然你觉得更不安全了，但事实上凶杀等恶性案件的案发率并不高，只是抢劫这样的轻罪案件多了，让社区觉得不安全而已。
• 不逮捕/起诉抢劫之类的轻罪犯很靠谱，因为这些人犯罪有深刻的社会因素，整个社会都负有责任，不能简单地归罪于他们。

只要自己不尴尬，尴尬的就是别人

[诺华老药工]

问了 GROK 一下FBI 在胡安明案中的表现：

Why Did the FBI "Fabricate" Evidence Using Google Translate?

The term "fabricate" here refers not to deliberate invention from nothing, but to knowingly building and pursuing a case on flawed, unverified, and misleading evidence—a form of prosecutorial and investigative misconduct criticized by civil rights groups, Congress, and the judge. Court records and testimony show the FBI persisted despite red flags, prioritizing "statistical accomplishments" over accuracy. Key reasons include:

Institutional Pressure from the China Initiative
The program created quotas for cases against perceived Chinese threats, pressuring agents to open probes on minimal suspicion. Former FBI agents and critics, like Andrew German, noted it fostered "develop[ing] cases to fit the framing" rather than evidence-based work.

An anonymous tip + a hasty Google Translate output was "evidence enough" to launch surveillance, as Sadiku testified. When espionage leads dried up, the FBI shifted to fraud charges based on the same weak foundation, admitting they "cooked up a plan" to access UTK records.
FBI Director Christopher Wray's rhetoric framed China as the "greatest long-term threat," amplifying bias and rushing investigations.

Reliance on Inaccurate Tools and Lack of Expertise
Google Translate is not forensic-grade for legal evidence, especially nuanced academic Chinese. Its "rough" output misrepresented the BJUT documents, but the FBI didn't consult linguists or verify—despite Hu's denials. This echoed broader China Initiative flaws, where mistranslations (e.g., "spiritual hurts" as religious ties) undermined cases.
Sadiku's "summary translations" were presented as fact to UTK and NASA, misleading them into believing Hu hid ties—leading to his firing and arrest.

Racial Profiling and Confirmation Bias
Over 90% of China Initiative targets were of Chinese descent, fostering assumptions of guilt based on ethnicity rather than evidence. Rep. Judy Chu (D-CA) called it "racial profiling" that led to "flimsy cases," with the FBI fabricating to justify probes.
The FBI attempted to coerce Hu into spying for the U.S. on China; when he refused, they escalated. Nearly 100 House members urged AG Merrick Garland to investigate this "wrongful targeting."

Cover-Up and Persistence Despite Flaws
Post-tip, the FBI concealed the investigation from Hu, obtained records illegally, and misled UTK/NASA. Even after admitting falsehoods in court, the DOJ pushed for retrial until the judge intervened.
A juror from the first trial, Wendy Chandler, criticized the FBI for wasting resources on a baseless case, saying they owed Hu an apology.

[流浪枪手]

问了 GROK 一下FBI 在胡安明案中的表现：

Why Did the FBI "Fabricate" Evidence Using Google Translate?

The term "fabricate" here refers not to deliberate invention from nothing, but to knowingly building and pursuing a case on flawed, unverified, and misleading evidence—a form of prosecutorial and investigative misconduct criticized by civil rights groups, Congress, and the judge. Court records and testimony show the FBI persisted despite red flags, prioritizing "statistical accomplishments" over accuracy. Key reasons include:

Institutional Pressure from the China Initiative
The program created quotas for cases against perceived Chinese threats, pressuring agents to open probes on minimal suspicion. Former FBI agents and critics, like Andrew German, noted it fostered "develop[ing] cases to fit the framing" rather than evidence-based work.

An anonymous tip + a hasty Google Translate output was "evidence enough" to launch surveillance, as Sadiku testified. When espionage leads dried up, the FBI shifted to fraud charges based on the same weak foundation, admitting they "cooked up a plan" to access UTK records.
FBI Director Christopher Wray's rhetoric framed China as the "greatest long-term threat," amplifying bias and rushing investigations.

Reliance on Inaccurate Tools and Lack of Expertise
Google Translate is not forensic-grade for legal evidence, especially nuanced academic Chinese. Its "rough" output misrepresented the BJUT documents, but the FBI didn't consult linguists or verify—despite Hu's denials. This echoed broader China Initiative flaws, where mistranslations (e.g., "spiritual hurts" as religious ties) undermined cases.
Sadiku's "summary translations" were presented as fact to UTK and NASA, misleading them into believing Hu hid ties—leading to his firing and arrest.

Racial Profiling and Confirmation Bias
Over 90% of China Initiative targets were of Chinese descent, fostering assumptions of guilt based on ethnicity rather than evidence. Rep. Judy Chu (D-CA) called it "racial profiling" that led to "flimsy cases," with the FBI fabricating to justify probes.
The FBI attempted to coerce Hu into spying for the U.S. on China; when he refused, they escalated. Nearly 100 House members urged AG Merrick Garland to investigate this "wrongful targeting."

Cover-Up and Persistence Despite Flaws
Post-tip, the FBI concealed the investigation from Hu, obtained records illegally, and misled UTK/NASA. Even after admitting falsehoods in court, the DOJ pushed for retrial until the judge intervened.
A juror from the first trial, Wendy Chandler, criticized the FBI for wasting resources on a baseless case, saying they owed Hu an apology.

再问一下ChatGPT, 肯定大不一样