Perplexity AI: A Better Way to Search for Information
Summary: AI is changing the way we work. Unlike traditional search, which delivers a list of links, Perplexity synthesizes a single clear answer from multiple web sources. It focuses on reliable answers, cited sources and user-friendly design.
AI is the fastest-adopted technology in history. So, life’s always changing, and who knows what it’ll look like in a few years. It seems like everyone’s talking about how we need to start using AI at work. But honestly, most people are just talking without really understanding or taking any real action. That’s why I’ve decided to start talking about AI tools and how they can benefit UX designers. In this article, I’ll be talking about Perplexity, which I’ve been using extensively since October.
Perplexity is a fast-growing AI startup — a conversational search engine — that answers queries by delivering reliable information as quickly as possible, utilizing sources from the web. The company has raised over $102 million in funding so far from prominent VC companies such as New Enterprise Associates and Bessemer Venture Partners, as well as companies including Nvidia and various angel investors like Elad Gil, Nat Friedman and Jeff Bezos.
When I was in primary school, I didn’t have a computer and internet at home. To do my homework, I’d often go to a neighbor’s house and use their Armenian Soviet Encyclopedias. Later, when I finally got a computer, Wikipedia became my best friend for research. Then I started using Google for searching and more recently I’ve been reading Reddit discussions. But now, I’m finding that Perplexity is the best place for quick and reliable answers.
It’s crazy how much things have changed. Back then, I never would’ve imagined I’d have so much information right at my fingertips. But now, it feels like we’re almost drowning in it, doesn’t it? From my own experience, I feel that we are already used to getting information quickly. Everyone wants everything instantly. While getting information quickly seems simple, simple problems are often the hardest to solve.
Simplicity is Key: The easier it is for a user to perform an action, the greater chance of habit formation. To make something truly simple, you have to delve really deep into the depths of complexity.
Why is Perplexity a great choice for UX designers? Even though it doesn’t attempting to compete directly with Google or other AI tools, let’s compare.
Perplexity AI vs. Google
During the 90s, Google beat Yahoo and dominated about 80% of the search market. In various books, I’ve come across the idea that “When I release the second version of this book, Google will lose its position.” However, it’s hard to imagine that scenario. To change ingrained behavior, the new version must roughly 2x outperform the old one. So, we still can’t imagine Google giving up its position in the market. However, tools like Perplexity show the potential for advancements in search.
Let’s consider a common question that a beginner UX designer might search for:
“What are the best practices for displaying error messages in a web application?”
Both AI and the search engine offer correct answers. While Google wins on brevity, Perplexity provides more user-friendly experience. Instead of making you do all the hard work of figuring out which information is important and putting it all together, it does that for you. It’s like it writes a short, helpful article just to answer your question. Perplexity notes important factors that users might not have initially considered. Additionally, unlike Google, where ads often dominate in SERPs, Perplexity eliminates this distraction and provides a more enjoyable experience. While there isn’t a clear winner, Perplexity stands out as my preferred choice due to its helpfulness to designers and clear links to additional information.
One reason why Google beat Yahoo was its minimalist design, which kept users focused on the main action without distractions. However, Google itself now faces this problem. Sometimes paid search results (ads) prevent us from seeing organic search results. Perhaps a company valued at $1.69 trillion may not even view a $520 million company as a competitor, but it’s always smart to stay alert because threats can arise unexpectedly from any direction.
Perplexity AI vs. Mistral AI, Gemini, & GPT-4
Jacob Nielsen’s recent research, “Search vs. AI” reveals:
- Productivity was 158% higher with ChatGPT compared to Google.
- ChatGPT answered questions in 5.79 minutes, while Google took 14.95 minutes.
- ChatGPT users scored 8.55 for accuracy, while Google users scored 8.77, on a scale of 0–10, which was statistically insignificant.
- Users strongly prefer AI over Google, rating AI 6.06 and Google 5.27, on a 1–7 satisfaction scale.
- The AI narrows the skill gap between users with low and highly educated backgrounds.
Note: The AI mentioned in Nielsen’s article refers to ChatGPT 3.5.
Based on these metrics, AI easily beat search. While I don’t currently have access to Perplexity’s paid version ($20/month, similar to premium Gemini and ChatGPT), it still surpasses the others AI tools. For comparison, I tested Mistral AI (the AI startup in France), Gemini Advanced and GPT-4. These tools clearly outperform Perplexity in many areas. However, that’s not the primary purpose behind Perplexity; let’s focus solely on its search function.
The key difference is that the paid Perplexity Copilot summarizes answers from a wider range of web sources, enhancing quality and insight. Additionally, you can generate images and attach files to get the summaries.
- Mistral AI: Its strength seems to be in clarity, explaining the main ideas in a friendly and clear format without being overly wordy. While it shows the basics well, it doesn’t go into as much detail as Perplexity or GPT when it comes to extra details. The information’s update date is unknown, but I guess it might be the oldest compared to the others.
- Gemini Advanced: Its strength is that its information is regularly updated. Gemini’s response is longer compared to others. There’s potential to condense it while maintaining the essential information. Although Gemini shows the basics, it doesn’t delve as deeply into topics as Perplexity. But Gemini offers an advantage here with its example (for example, only Mistral AI N8 has a text example).
- GPT-4: Seems like it could find a middle between clarity and providing in-depth explanations. It might offer deep insights and explanations that are more detailed than Mistral AI but not as exhaustive as Perplexity. However, it’s important to fact-check its responses for accuracy. Additionally, its information was last updated in April 2023.
Overall, Perplexity proves to be the most convenient option. Perplexity’s key strength is directly citing sources within answers. This allows users to verify information, explore topics further with reliable links and understand the origins of design patterns or research findings. With other AI tools, I often have to request links, which are sometimes inaccurate. Perplexity also excels at summarizing information from multiple sources, tailoring its responses to the specific question. This makes understanding complex research, best practices, or competitor analysis far easier than reading numerous articles. Perplexity stands out against that AI tools because it’s specifically designed for information seeking.
Perplexity Prioritizes User-Friendly Design (And No, Businessmen Don’t Always Need Earrings)
ChatGPT, Midjourney and similar apps are awesome engineering products, but let’s be real — they weren’t built with the average user in mind. They’re more focused on the engineering side than making things easy to use. Perplexity figures out a big part of the technology aspect but is also more usable than other AI tools. After raising a $73 million Series B round, it seems they’ll be paying more attention to usability. In a recent NNgroup podcast (minute 16:54), Perplexity’s Head of Design, Henry Modisett, talked about making the experience more personalized experience — which I believe is the future and write about it in my last article.
Henry in the podcast also said, “I think prompting is the worst software experience ever,” a sentiment I fully support. (minute 25:08)
Prompt engineering is a skill. Generative AI tools require users to write effective prompts to get the best results. Crafting the right prompt can be overwhelming, especially for inexperienced users. Simply typing a few keywords into ChatGPT won’t get significant results, but that’s not the case with Perplexity.
Last time I needed an image for my article, I tried many AI image generation tools (Midjourney, Leonardo AI, Microsoft Designer, Gemini, and DALL-E 3). Despite spending hours describing my ideal image, the results were all terrible. After spending hours of my life, I just Photoshopped a weird picture DALL-E made. It was puzzling why DALL-E decided to generate the businessman always with earrings. I faced the same problem this time, but decided to leave the flawed picture as it was.
Another noteworthy concept is how Perplexity aligns with people’s mental models (minute 25:36). The most successful products blend different interaction styles to create a user-friendly experience. UI plays a crucial role in bridging the gap between powerful technology and the average user. Perplexity uses a search-bar-like design that is familiar to everyone. Most people immediately understand how to use it, while ChatGPT can be intimidating. It starts with an empty space where you’re to chat.
Additionally, in Perplexity, the focus function allows users to fine tune their search narrowing down the sources, for more targeted and relevant results. On the other hand, in ChatGPT’s search, I rely on AIPRM plugin.
The future of UX design is intertwined with AI. If you want to stay ahead of the game, you’ve got to learn to work with these tools. Perplexity is a great starting point. It’s super useful and also a good example of how to design for AI. I hope Perplexity helps you in your work.
Just remember, no tool is perfect, so don’t rely solely on AI answers. It’s always a good idea to double-check the information they provide.
NOTE: I don’t get anything for recommending Perplexity AI. I just think it could be really helpful for readers.
