AI Chatbot Chaos: Errors, Bias, and the Rise of Google’s “Bad Bot”

4

Google’s AI Oversights: A Rocky Start

LA News Center brings you an exclusive investigation into Google’s AI Overview feature, which has stirred controversy due to its inaccurate and nonsensical responses.

A Glimpse into AI Overview

AI Overview, a recent addition to Google Search, aims to provide users with quick, convenient answers to their queries. However, the tool’s performance has raised serious questions about its reliability.

Based on information gathered from across the web, AI Overview presents a summary of results at the top of the search page. While this approach can be helpful, as seen in the example of cleaning leather boots, it has also led to a slew of errors and misstatements.

A Series of Misfires

Social media platforms have been flooded with screenshots showcasing AI Overview’s alarming responses:

  • Incorrectly stating that Barack Obama was the only Muslim president of the United States.
  • Suggesting the addition of nontoxic glue to pizza sauce to prevent cheese from sticking.
  • Quoting UC Berkeley geologists (nonexistent) to advocate for daily consumption of rocks.
  • Providing misleading health advice, such as recommending staring at the sun for up to 30 minutes.
  • Fabricating a list of fruits ending with “um.”

Accuracy and Attribution Issues

AI Overview’s reliance on information synthesis from various websites has created problems with accuracy and attribution. For instance, the tool attributed incorrect information to medical professionals and scientists.

Eye-Opening Inquiries

When asked about antitrust laws, AI Overview accurately reported the ongoing lawsuit against Google by the U.S. Justice Department and 11 states. However, its responses to other queries, such as meal planning, were unsatisfactory.

A Google Spokesperson’s Response

In a statement to LA News Center, a Google spokesperson acknowledged that while the majority of AI Overviews provide valuable information, some uncommon and doctored examples have been identified. The company emphasizes that AI Overview underwent rigorous testing and is taking swift action when content violates its policies.

Historical Context: Gemini’s Image-Generation Fiasco

Google’s AI challenges are not isolated to AI Overview. In February 2023, the company launched Gemini, an image-generation tool that raised concerns over historical inaccuracies.

User inputs resulted in images that misrepresented historical figures, outfits, and events. Google briefly paused the tool to address the issues.

Ethical Concerns and Criticism

The Gemini episode ignited a debate within the AI industry, with some labeling the tool too sensitive while others criticized Google for insufficient investment in AI ethics. This follows controversies surrounding Google’s handling of its AI ethics group in 2020 and 2021.

Industry Response

Sundar Pichai, CEO of Alphabet (Google’s parent company), faced criticism from employees for the rushed rollout of Bard, which came amid the popularity of ChatGPT.

Conclusion

While AI has the potential to revolutionize various industries, Google’s ongoing struggles with AI Overview and Gemini highlight the challenges of developing accurate and reliable AI systems. Transparency, accountability, and careful implementation are paramount to ensuring the benefits of AI outweigh the potential pitfalls.

Data sourced from: cnbc.com