Revolutionizing the search landscape, Pearl’s human-aided AI engine challenges traditional query methods by incorporating expert human input to provide more accurate and reliable results.
The Human Touch in Search: Pearl‘s Unconventional Approach
When online search engines first emerged, they seemed like a revolutionary concept. However, over time, the landscape has changed, and search has become increasingly plagued by spam and ads. Even big players like Google are now relying on AI to salvage search, despite its checkered past.
A New Player in Town: Pearl
Enter Pearl, a new AI-powered search engine that’s taking a different approach. Unlike other services that rely solely on large language models (LLMs) for answers, Pearl incorporates human experts into the process. This hybrid model aims to provide more accurate and reliable results.
The CEO’s Vision
Andy Kurtzig, CEO of JustAnswer and the parent company of Pearl, believes that his product is an extension of a traditional search project he’s been working on for decades. He emphasizes that Pearl lowers the barrier to entry for answers from experts, making it more accessible than its subscription-based counterpart, JustAnswer.
The TrustScore: A Measure of Quality
Pearl’s LLM is built on top of popular foundational models, including ChatGPT, and is customized with JustAnswer’s extensive database. The AI answers are free, as well as the first layer of human fact-check, the TrustScore (a ranking from 1 to 5 about the quality of an AI answer). When users want more in-depth information, they can opt for a paid subscription service.
Section 230: A Shield Against Liability
Kurtzig claims that Pearl will have Section 230 protections, which shield traditional search engines from liability. He believes this is due to the incorporation of human experts into the process, making it less likely to provide misinformation. However, when questioned about this, both the AI and a human expert provided uncertain answers.
A Test Run
WIRED tested Pearl by asking various questions, including one about its own history. The AI response was serviceable but lacked depth, while the TrustScore indicated a mediocre answer. Connecting with a human expert took over 20 minutes, only to provide a remarkably similar answer to the AI.
The Verdict: A Mixed Bag
While Pearl’s hybrid model is an interesting concept, it falls short in practice. The quality of answers from both the AI and human experts was inconsistent, and the TrustScore did not accurately reflect the quality of the responses. Ultimately, users may find themselves relying on other online communities for guidance, rather than paying a subscription fee.
A Call to Action
If you’ve had a memorable experience with Pearl or any other AI search product, please share your story in the comments below. Your feedback will help shape the future of search and provide valuable insights into what works and what doesn’t.
