Imagine a world where finding information isn’t a frustrating hunt but a simple conversation. You ask a question and get a clear, helpful answer instead of endless web pages. These smart engines go beyond just searching. They deliver direct, tailored responses to user inquiries, utilising a constantly growing database of information and interpreting context with ever-improving precision and accuracy. The traditional search is more about the keywords on a page and the backlinks that will lend credibility to that page. Unlike the traditional search engines focused on keywords and backlinks to ascertain a page’s credibility, these platforms offer something far more intuitive and user-centred.
This article will delve into the fascinating world of answer engines, uncovering the sophisticated technology that powers these intelligent tools. We’ll explore how they process information, understand user queries, and deliver precise answers, transforming how we access and utilize knowledge.
N0, we;re not just talking about genAIchatbots
The common assumption is that an “answer engine” must be a generative AI chatbot—an AI system that allows users to ask questions and generate responses. However, answer engines do not necessarily have to rely on artificial intelligence. Instead, they could provide answers by retrieving information from a large database or performing reliable calculations rather than using the capabilities of large language models (LLM).
In other words, an “answer engine” does not have to be a conversational AI system. The Wolfram Alpha platform is a popular example of an answer engine without generative AI. First released in 2009, Wolfram Alpha can answer various questions on everything traditional search engines struggle to address, from math and science to history.
Wolfram Alpha’s accuracy was so trusted that its information was directly integrated into early versions of Siri. Even ChatGPT sometimes defers to Wolfram Alpha’s “knowledge engine” when seeking the right answers. While Wolfram Alpha is particularly adept at handling mathematical queries, its capabilities extend far beyond just doing calculations. The platform remains widely used and is especially valued within the scientific community.
The new age AI-powered answer engine
The “answer engine” concept is also gaining traction among LLM-powered conversational chatbots. For instance, Perplexity AI, a search startup based in San Francisco, unveiled its answer engine in December 2022. Unlike traditional search engines, Perplexity AI provides a chatbot-like interface for users to ask questions and receive succinct answers supported by a carefully curated set of citations.
The key difference between Perplexity AI and ChatGPT is that Perplexity AI primarily focuses on providing information, unlike ChatGPT’s stronger conversational abilities. For example, if you ask about making Aglio e Olio pasta, ChatGPT will provide a brief description and the recipe. At the same time, Perplexity AI would first give you links to online recipes and then the steps, citing the sources. This saves the user time and effort compared to searching on Google. This information-centric approach makes Perplexity AI a more suitable tool for academic researchers and those working in research-heavy professions like data science and marketing, where quickly finding reliable sources and information is crucial. Perplexity AI leverages LLMs like GPT-4, Claude, Mistral Large, and its own proprietary models for natural language processing to power its search capabilities.
Perplexity AI has clients from a variety of industries, such as payment processing (Stripe), video conferencing (Zoom), investment management (Bridgewater), data warehousing (Snowflake), sports (Cleveland Cavaliers), advertising (Universal McCann), wellness (Thrive Global), data analytics (Databricks), mobile payments (Paytm), voice synthesis (ElevenLabs), technology (HP, Vercel, and Replit). Currently, the basic version of Perplexity AI is free to users. However, many of its more advanced features can only be accessed by signing up for the Perplexity Pro subscription, which costs US$20 (£15.34) per month or US$200 (£153.44)per year.
The drawbacksof AI-based answer engines
Answer engines excel at providing swift, concise information, transforming the search landscape. However, they also have their own set of drawbacks.
AI Hallucinations
Chatbots’ tendency to generate responses on the spot can lead to a significant issue known as AI hallucinations. Since these systems use LLMs, they can dynamically compose plausible-sounding information. However, this information may be factually incorrect or incomplete.
The problem arises because chatbots’ responses often appear authoritative, causing users to mistakenly accept them as truthful without realizing the system’s underlying limitations. This is a major drawback compared to traditional information retrieval systems, which pull from curated, verified data sources.
Verification challenges
The verification challenge is a significant hurdle for answer engines. Even when these systems provide citations or references to support their responses, this approach shifts the burden onto users to consult those external sources and verify the accuracy and reliability of the information themselves. Many users may not have the time, motivation, or ability to undertake this additional verification process, potentially leading them to accept the information as truthful, even if it is inaccurate or incomplete. Addressing the fundamental challenge of ensuring the trustworthiness and factual integrity of the responses generated by these systems remains a crucial area for improvement.
Distilled
In an age characterized by information overload, time has become an increasingly valuable resource. Answer engines offer a compelling solution by providing direct and efficient access to knowledge. However, answer engines must consistently provide accurate and comprehensive information to compete with established search giants like Google. Overcoming challenges related to data quality and factual correctness is crucial. If these obstacles can be addressed, answer engines have the potential to revolutionize how we interact with information, heralding a new era of search and discovery.
