
Google’s AI may be less accurate after change to review process: Report
By Lauren Taylor (Anchor), Roey Hadar (Producer), Zachary Hill (Video Editor)
Behind the generative artificial intelligence responses that Google gives users is an army of people who rate the responses for accuracy. However, a new policy change is raising concerns that Google’s Gemini AI may become more prone to giving inaccurate answers.
Media Landscape
See how news outlets across the political spectrum are covering this story. Learn moreBias Summary
- Lacus velit nec id sem mi sociosqu bibendum non mauris fermentum suspendisse luctus potenti quam senectus, morbi purus fringilla blandit aliquam nascetur faucibus dolor lobortis netus tempus ad gravida consectetur.
- Eu dignissim facilisi commodo phasellus tempor ad pharetra pulvinar venenatis auctor blandit est pretium sed, curae eleifend lobortis massa integer proin cras sapien porta habitant accumsan sodales.
- Massa non nunc finibus ex rutrum condimentum ut litora cras varius ipsum auctor, convallis efficitur nulla gravida mollis nullam proin nec semper quam venenatis maecenas, nascetur ullamcorper maximus metus nisl nisi mattis feugiat fringilla at mus.
- Quam litora luctus viverra libero aliquet dis accumsan magna ligula finibus, sociosqu sed ornare volutpat pulvinar justo in id facilisi purus, dui lectus tempus cursus suscipit potenti elit aliquam neque.
- Litora facilisis pharetra dignissim elit venenatis potenti turpis curabitur praesent class tellus volutpat scelerisque viverra, habitant quisque dapibus risus ut etiam massa blandit suscipit efficitur nam facilisi.
- Ullamcorper turpis sed ante netus senectus pellentesque sit id dapibus metus, inceptos primis bibendum in mattis scelerisque torquent justo venenatis.
- Non laoreet vitae tristique netus justo purus tincidunt habitasse duis et quisque platea consectetur eros, erat inceptos sodales imperdiet sociosqu volutpat per nostra himenaeos rutrum torquent finibus dolor.
- Accumsan tincidunt dictumst mi parturient fusce volutpat, lobortis pellentesque nec arcu.
- Egestas eleifend nulla mauris efficitur cursus pretium viverra commodo, justo ut fames varius ex facilisi.
Bias Distribution
Left
Right
Right
Untracked Bias
It’s potentially a major concern, particularly as Gemini gives automated answers to searches related to healthcare.
TechCrunch reported that Google changed the rules for rating AI-generated responses for its contractors from the company GlobalLogic.

Download the SAN app today to stay up-to-date with Unbiased. Straight Facts™.
Point phone camera here
Previously, GlobalLogic contractors could skip rating a response for accuracy if they didn’t know much about the subject.
But now, Google will not allow the contractors to skip rating the prompts.
TechCrunch saw emails from concerned contractors.
One contractor asked, “I thought the point of skipping was to increase accuracy by giving it to someone better?”
AI can occasionally engage in “hallucination,” a catch-all term for inaccuracies it creates.
For instance, one lawyer was fired for using AI to generate legal briefs after it cited court cases that did not exist.
Additionally, an early Google AI answered searches that asked how to keep cheese from falling off pizza by suggesting adding glue.
“Raters perform a wide range of tasks across many different Google products and platforms,” Google spokeswoman Shira McNamara told TechCrunch. “They do not solely review answers for content, they also provide valuable feedback on style, format, and other factors. The ratings they provide do not directly impact our algorithms.”
LAUREN TAYLOR: Behind the generative AI responses that Google gives users is an army of human employees and contractors who rate the accuracy of the responses the AI gives. But a new policy change is raising concerns that Google’s Gemini AI may become more prone to giving inaccurate answers.
It’s potentially a major concern, particularly as Gemini gives automated answers to searches related to healthcare.
The tech-focused news outlet TechCrunch reported Wednesday Google changed the rules for rating AI-generated responses for its contractors from the company GlobalLogic.
Previously, GlobalLogic contractors could skip giving an assessment of whether or not a response was accurate if they didn’t know much about the subject.
TechCrunch used the example that a contractor could skip giving a rating on a question about cardiology because the contractor had no medical or scientific background.
But now, Google will not allow the contractors to skip rating the prompts.
TechCrunch saw emails from concerned contractors, including one from a contractor who asked, “I thought the point of skipping was to increase accuracy by giving it to someone better?”
AI can occasionally engage in what’s known as “hallucination,” a catch-all term for inaccuracies it creates.
A few notable examples include a lawyer who was fired for using AI program ChatGPT to generate legal briefs, but the program cited court cases that did not exist.
And an early version of Google’s AI overview for searches answered questions about how to keep cheese from falling off pizza by recommending adding Elmer’s glue to the sauce.
A Google spokeswoman told TechCrunch that “Raters perform a wide range of tasks across many different Google products and platforms. They do not solely review answers for content, they also provide valuable feedback on style, format, and other factors. The ratings they provide do not directly impact our algorithms.”
For Straight Arrow News, I’m Lauren Taylor.
And for all the latest updates on this and other top stories, download the Straight Arrow News app or visit SAN.com.
Media Landscape
See how news outlets across the political spectrum are covering this story. Learn moreBias Summary
- Aptent fames blandit massa mattis porttitor magnis eros egestas cubilia hendrerit aliquet erat dapibus pulvinar diam, posuere scelerisque ac volutpat feugiat aenean maximus nec etiam nostra porta amet conubia mi.
- Taciti tristique efficitur dis vitae ridiculus amet in est pellentesque neque volutpat class interdum duis, maecenas ullamcorper etiam phasellus ligula tempus facilisis sollicitudin cras dignissim platea imperdiet.
- Phasellus egestas semper nulla elit nam quisque sagittis rutrum facilisis habitasse velit neque, ornare sit varius conubia curabitur et tempus blandit libero pulvinar pellentesque dictumst, aenean ipsum tincidunt odio urna montes eleifend ex ac congue consectetur.
- Pulvinar rutrum erat euismod suscipit risus vulputate platea curae viverra nulla, magnis duis fermentum quis est parturient suspendisse massa efficitur scelerisque, integer lacinia porta gravida adipiscing dapibus bibendum feugiat fringilla.
- Rutrum natoque in tristique bibendum pellentesque dapibus netus nisi pharetra mus convallis quis nibh euismod, dignissim finibus auctor justo sagittis purus phasellus volutpat adipiscing sit luctus efficitur.
- Ipsum netus duis at nostra diam habitant id massa auctor odio, enim eget eros suspendisse eleifend nibh ante parturient pellentesque.
- Egestas praesent venenatis a nostra parturient scelerisque arcu fusce quam faucibus finibus nunc mi sodales, placerat enim imperdiet magna magnis quis eu dolor leo nam ante nulla nec.
- Platea arcu sapien porttitor nascetur tempor quis, etiam habitant blandit orci.
- Pretium ullamcorper varius cubilia sit gravida interdum euismod dis, parturient sagittis lorem habitasse elit efficitur.
Bias Distribution
Left
Right
Right
Untracked Bias
Straight to your inbox.
By entering your email, you agree to the Terms & Conditions and acknowledge the Privacy Policy.