Most People Use ChatGPT as a Search Engine. Perplexity AI Is Better for That.

I’ve been tooling around with Perplexity Pro for a couple of weeks now, and I’m fairly impressed. More so than when I was using ChatGPT Plus. That comes down largely to how I’ve been using it.

Adobe ran a survey last month in which they polled 1,000 people and found that 77% of those who use ChatGPT use it as a search engine. Three in 10 people trust ChatGPT for search results more than other search engines. “Other search engines” is the bucket into which Google, Bing, and DuckDuckGo fall into.

The thing is, ChatGPT is a conversational generative AI. There’s a better generative AI that’s built from the bottom-up to be optimized for sniffing out search results, like a smarter search engine. That AI is Perplexity.

a different breed

I’ve got no interest in asking any generative AI to create writing passages for me. Aside from taking all the fun out of writing, they output some truly dreadful writing that, even aside from being full of made-up “facts,” is merely forgettable and bland at its absolute best.

After all, that’s what these AIs do. They intake, recycle, and regurgitate existing written content that’s out there floating around on the web, and most of the writing out there isn’t very good. Garbage in, garbage out.

But when it comes to search, yeah. That’s where I’m most curious about AI, however cautiously. Google has felt broken for a long time, with the ads and clutter of junk results rising toward the top of the search results.

Perplexity functions a bit differently than OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini, Microsoft’s Gemini, or xAI’s Grok. It has access to several models from these competitors, including ChatGPT, Claude, and Grok, along with Perplexity’s own Sonar model.

These are upgraded, introduced, and discontinued constantly. Perplexity’s Deepseek-based R1 1776 model was just discontinued within the last week, for example. Using Perplexity, or paying for the Pro or Max tiers, runs your queries through these selectable models.

Some are built more for search, others more for reasoning, but what they all have in common is that they’re centering around researching sources and presenting the information to you, rather than going one step further and interpreting it (too much) for you, which is what I found ChatGPT doing. ChatGPT simply makes too many assumptions and too many false conclusions for me.

Perplexity’s far less sycophantic tone, too, is music to my… eyes, I guess. ChatGPT would routinely make up some bullshit (hallucinate, in AI parlance), and when I would correct itself it’d fawn all over me like Mr. Smithers. If I corrected it again, even in a way I knew to be wrong, it’d fawn again. So on and so on. Ass-kissers suck, even when they’re disembodied robots.

Perplexity has a free tier, like ChatGPT, Claude, and most other generative AIs. Its $20-per-month Pro tier, matching ChatGPT’s $20-per-month Plus tier, is worth the extra cash, though. Try out Perplexity’s free version, using your three free daily, upgraded “Pro” searches to see if you like it before you commit.

I’ve been using Perplexity AI’s Comet browser for the last week and change, too, so look out for my impressions on that in the coming weeks.