Meta blames hallucinations after its AI said Trump rally shooting didn’t happen

🌈 Abstract

The article discusses how Meta's AI assistant incorrectly stated that the recent attempted assassination of former President Donald Trump did not happen, which the company is attributing to the technology powering its chatbot and others. The article also mentions that Google had to refute claims that its Search autocomplete feature was censoring results about the assassination attempt.

🙋 Q&A

[01] Meta's AI Assistant's Incorrect Response

1. What did Meta's AI assistant incorrectly say about the attempted assassination of former President Donald Trump?

• Meta's AI assistant incorrectly said that the recent attempted assassination of former President Donald Trump didn't happen.

2. How is Meta attributing this error?

• Meta is attributing this error to the technology powering its chatbot and others, referring to it as "hallucinations" which is an industry-wide issue seen across all generative AI systems.

3. How did Meta respond to this issue?

• Meta initially programmed its AI to not respond to questions about the attempted assassination, but later removed that restriction after people started noticing. However, in a small number of cases, the Meta AI continued to provide incorrect answers, including sometimes asserting that the event didn't happen, which Meta is working to address.

4. What does Meta acknowledge about these types of responses?

• Meta acknowledges that "like all generative AI systems, models can return inaccurate or inappropriate outputs, and we'll continue to address these issues and improve these features as they evolve and more people share their feedback."

[02] Google's Response

1. What did Google have to refute regarding its Search autocomplete feature?

• Google had to refute claims that its Search autocomplete feature was censoring results about the assassination attempt on former President Donald Trump.

[03] Trump's Response

1. How did former President Donald Trump respond to the issues with Meta and Google?

• Trump said in a post on Truth Social: "Here we go again, another attempt at RIGGING THE ELECTION!!! GO AFTER META AND GOOGLE."

[04] Challenges with Generative AI

1. What challenge is the tech industry grappling with regarding generative AI?

• The tech industry has been grappling with how to limit generative AI's propensity for falsehoods, as seen in the example of Meta's AI assistant incorrectly stating that the attempted assassination of former President Trump did not happen.

2. How have some players, like Meta, attempted to address this challenge?

• Meta has attempted to ground its chatbots with quality data and real-time search results as a way to compensate for "hallucinations." However, the article notes that it is still hard to overcome what large language models are inherently designed to do: make stuff up.