From Promises to Pitfalls: The Current State of AI Search

the word ai spelled in white letters on a black surface

It’s been a month since Google’s major mishap with its new AI Overviews feature, designed to simplify searching by providing concise answers from multiple sources. Instead, it suggested bizarre advice like eating rocks and using glue on pizza. For example, asking Google about an African country starting with “K” would yield no results, despite Kenya’s existence. These errors persist, highlighting the current shortcomings of AI search.

Promising Beginnings and Stumbling Blocks

This spring, two significant announcements hinted at a new era for AI search. Google’s AI Overview update and Perplexity, a promising AI search startup, seemed poised to revolutionize the field. Perplexity launched a feature called Pages, which creates custom web pages full of information on specific topics. However, Perplexity soon faced allegations of plagiarism, revealing a troubling tendency for AI search tools to “borrow” content improperly.

Legal and Ethical Concerns

AI search’s problems have sparked considerable backlash. Recently, online publishers and creators lobbied lawmakers to scrutinize Google’s AI Overviews and other AI technologies that extract content from independent creators without proper attribution. This follows lawsuits by the Recording Industry Association of America (RIAA) and major record labels against AI companies for copyright infringement, and similar legal actions from newspapers against OpenAI and Microsoft for content scraping.

The Future of AI Search: Hope or Hype?

Generative AI technology promises to transform web searching, as seen with ChatGPT’s launch in late 2022. Tech giants like Microsoft, Google, and Apple are racing to develop their AI tools, with OpenAI reportedly planning its own search engine. Despite these efforts, experts doubt AI search will make finding information online effortless anytime soon.

Persistent Flaws in AI Search

AI search’s fundamental flaws are deeply embedded in its functionality, casting doubt on its potential to replace traditional search engines. Chirag Shah, a professor of information science at the University of Washington, believes that while AI search can be helpful, traditional search engines will still be necessary.

Two recent incidents highlight these issues. Google’s AI Overview suggested using glue to keep cheese on pizza, drawing from a Reddit thread mistakenly seen as authoritative. This problem stems from AI’s “hallucination” issue—generating incorrect information it deems accurate. Yoon Kim, an MIT professor, suggests that while progress is being made, completely eliminating these hallucinations remains unlikely.

Perplexity’s plagiarism scandal underscores another issue: unauthorized use of content. Perplexity was found to be scraping content from websites, including those blocking such activities, leading to accusations of content theft. This problem arises from AI’s inability to differentiate between good and bad sources effectively.

The Need for Cautious Optimism

While AI-powered search has potential, it currently lacks the common sense and judgment humans possess. Users should view AI search results as starting points, akin to Wikipedia, and verify the sources. Despite its shortcomings, AI can still provide valuable information, but human oversight remains crucial.

In conclusion, AI search technology, despite its advancements, is not yet ready to replace traditional search methods entirely. It offers promise but also requires significant improvements in accuracy and ethical practices before it can be fully trusted.