THE AMERICA ONE NEWS
May 30, 2025  |  
0
 | Remer,MN
Sponsor:  QWIKET 
Sponsor:  QWIKET 
Sponsor:  QWIKET: Elevate your fantasy game! Interactive Sports Knowledge.
Sponsor:  QWIKET: Elevate your fantasy game! Interactive Sports Knowledge and Reasoning Support for Fantasy Sports and Betting Enthusiasts.
back  
topic
National Review
National Review
28 Feb 2024
Philip Klein


NextImg:The Corner: Not Only Is Google Gemini Biased — It Can’t Even Read

The bias of Google Gemini has been well documented. But as the company’s CEO enters damage control mode, it’s worth emphasizing that the problem isn’t merely that the new AI tool is biased. The problem is that it is bad software that cannot even read.

At the risk of coming off as self-absorbed, I will confess that I decided to look up myself in Gemini, with the question, “Is Philip Klein reliable?” It became pretty clear that not only did the response regurgitate a section of my Wikipedia page, but it did so in a way that reversed the meaning of a sentence.

Here’s the relevant section of Gemini’s answer:

The first part of the Gemini response lacks context. At the 2012 Democratic convention, Debbie Wasserman Schultz did accuse me of misquoting her, as noted by Wikipedia. However, after she made that accusation, I released audio of her remarks, which led even the Washington Post and Politifact to conclude I quoted her accurately. (The Post gave her Four Pinocchios and Politifact to gave her a Pants on Fire rating.) 

However, the Wikipedia page also lacks this context. So in this case Gemini could merely be blamed for repeating the gaps in Wikipedia.

When it came to AOC, however, Gemini got things completely backward. Shortly after she was elected, I wrote a post debunking her ridiculous claim that there had been $21 trillion in Pentagon accounting errors that could have financed most of “Medicare for All.” Here is what Wikipedia says:

In 2018, Klein reported on a tweet by Representative Alexandria Ocasio-Cortez about Department of Defense budgets which he called a “$21 trillion mistake” revealing a lack of understanding of government budgeting.

But Gemini’s interpretation was that it was I who “was criticized for his lack of understanding of government budgeting…”

In other words, Gemini interpreted “revealing a lack of understanding of government budgeting” as an accusation that I didn’t understand budgeting rather than me accusing AOC of not understanding government budgeting. 

The whole promise of AI is that it is supposed to replicate the human brain, and ultimately be much smarter than the smartest human. But it seems to me that if Gemini cannot even pass a basic reading comprehension test, its problems are much bigger than wokeness.