I replaced Google Search with Perplexity for a week: Here’s why I liked and what I didn’t

Perplexity is an AI-powered search engine that aims to provide accurate and relevant results by citing sources and examples in its summaries. It’s a powerful tool that demonstrates how we could reimagine search engines using AI, and leave Google’s spam-filled results behind for good.

Perplexity’s up-to-date results, which are paired with summaries and citations, make it seem like the perfect replacement for Google. I use Google daily for work, so I tried replacing it with Perplexity for a week to see if it could become a permanent addition to my life.

Citations, relevant questions, and summaries make researching a breeze

Over the last week, I completely replaced Google Search with Perplexity. Whether I was searching for tips for my next hobby project, or conducting research for my next article, I plugged my query into Perplexity’s app or browser instead.

Perplexity shined in my personal life when I didn’t know exactly what I was searching for. For example, after being impressed by a new beer, I wanted to check out the brewery. However, I had forgotten the name of the brewery, and all I knew was that it was based in Nottingham, and the beer I had had a relatively high alcohol content.

Plugging these vague facts into Perplexity rewarded me with the brewery I was looking for in the first result. Google managed to list it second but crammed my screen with information I didn’t need at the time. This wasn’t a unique situation, Perplexity managed to provide me with bloat-free answers to my queries that were usually at least as accurate as Google’s.

One of the most notable Perplexity features is its citation feature. After each segment, Perplexity adds a citation link to its source. This is useful for personal use (the citation for the brewery took me straight to their website), but crucial for professional work. I tend not to use generative AI tools in my research due to its tendency for hallucination, but Perplexities’ citations let me quickly verify its statements.

So far, so good. Perplexity had impressed me within a day with its relevant and bloat-free responses. It’s related queries helped me dive deeper into my topic, and the citations helped allay my worries about using an AI-powered search engine. But it soon turned sour as the common issues with generative AI reared their head after just a day or two of use.

The Bad: I had to fact-check Perplexity and its source

Twice as much research for the same result

During my research for one article, I wanted to double-check I hadn’t missed any key facts about the upcoming Google I/O developer conference. I entered “Gemini at Google I/O 2025” and a quick scan of the results confirmed what I had already written. But I obviously can’t rely on just an AI summary, so I spend the next 15 minutes checking its sources.

The first statement from Perplexity cited two websites, Support Plan and Tech Radar. While I confirmed that it had fairly represented Tech Radar’s article, I was suspicious of the lack of authorship and publication date. These suspicions were confirmed when I checked the Tech Radar article. Support Plan had ripped its article directly from Tech Radar, which was listed as the second source in Perplexity’s summary. Websites ripping off articles from bigger publications without crediting the original author is nothing new and easy to spot for humans, but Perplexity had read it as a legitimate source and prioritized it over the original.

This problem cropped up a few more times during my week. Perplexity was citing sources that were plagiarizing articles, often from websites the search engine had previously cited. This immediately led me to question all my results. Perplexity was not providing me with a balanced summary as it was scraping data from the same text multiple times.

Paired with this problem came good old-fashioned AI misinformation. The same query returned a bullet point claiming that Google was planning on introducing AI overviews alongside its search results. This struck me as odd, as I knew this had already happened (and been subsequently retracted). Sure enough, the cited article was from May 2024 but Perplexity claimed the feature was still in the works.

All of the above meant I could no longer trust Perplexity for accurate summaries. I had to thoroughly fact-check each statement, which ended up taking me more time than googling the query and writing up a summary myself. These summaries gave me an overview that was useful for little more than a jumping-off point.

Searches in my personal time were also filled with inaccuracies. I spotted an absurd amount of mistakes in this search for upcoming Warhammer models, as the results were scraped from poorly written and inaccurate fan sites. Why didn’t Perplexity use information from the official website? No idea. Interestingly, Google’s AI Overview provided vaguer, but more accurate results. However, it still didn’t manage to scrape the original source.

The Ugly: Lawsuits and copyright infringements

Perplexity doesn’t respect its sources

By the end of my week with Perplexity, I was ready to give it up. While it was a fantastic research assistant when it worked, it suffered from the same issues as every other AI tool. Unless the problems around LLM’s are solved soon, I can’t rely on it.

Even if a miracle happens, and Perplexity stops citing plagiarized work and irrelevant articles, I still won’t be using it. Investigations have reported that Perplexity is ignoring the Robot Exclusions Protocol (which allows websites to stop web crawlers) by using third-party web crawlers. Perplexity is also subject to lawsuits claiming copyright and trademark infringement.

Will I keep using Perplexity? No

Perplexity offers nifty features that I think are a genuine improvement on the search engine formula. Its library of past searches is vastly more useful than Google’s search history, and the relevant questions feature helped me think of new questions. I unfortunately wasn’t able to test its reverse image search tool (which required me to sign up for a $20/month Pro subscription).

The problem is that I can’t rely on its summaries, and its plagiarizing of content raises serious ethical concerns. If Perplexity is anything to go by, AI isn’t even close to reinventing the search engine.