As a writer and content marketer, I can’t say enough how useful AI summaries have been for research. I no longer have to click through and read each search result. I get the summarised version right away.

Case in point: I was updating my article on the top trending topics. I’m not that old, but I’m clearly not up-to-date with most trends. Previously, each update took time because I had to read multiple articles to get a sense of what was happening. Now, Perplexity solves this for me:

You can do this with ChatGPT, too, but what I like about Perplexity is you get access to the source of information. So, in the event you’re suspicious that the answer is a hallucination, you can fact-check it by clicking on any of the cited sources. That’s way better than ChatGPT confidently magicking something out of thin air.
In fact, if Perplexity can’t find anything or doesn’t understand your query, it won’t hesitate to tell you:

ChatGPT doesn’t do this nearly enough. Even if it doesn’t know something, it’ll still try to create an output, some of which are hallucinations.
That’s not to say Perplexity doesn’t hallucinate at all. It just does so less. For example, I asked why a road in Singapore had the same name as the wife of the Prime Minister of Singapore.

The answer was wrong—it’s not named after her; it was simply a coincidence. If someone did not click through to source #2, they would have been led astray.
My example: I was reading a book about a Singapore pioneer. The book mentioned Dr. Oxley, the namesake of a road in Singapore. I asked Perplexity, and it gave me a modern-day Dr. Oxley:

The same search on Google gave me the right person immediately:

Of course, this is because Google is collecting location data. It knew I was in Singapore and therefore more likely to search for Dr. Thomas Oxley rather than a physician in the US.
Despite the uproars over Google’s data collection, it’s clear that location data is important. With this data, Google knows where you are and can personalize their search results. After all, if you’re looking for the “best Italian restaurants,” you don’t want suggestions halfway across the globe. You want those that are in your area.
This is where, in my opinion, Google excels and where it distinguishes itself from AI. AI doesn’t collect location data or know where you are (unless you tell it.) So, extra steps are involved in getting the right results when you ask AI for local information.
Another example: I was looking at buying an ergonomic mouse. When I asked AI, it suggested places that were in the UK (Argos):

On the contrary, Google gave me a list of local places in Singapore to buy from:

Final thoughts
Personally, I’ve mostly replaced Google with Perplexity and use it 90% of the time. A clean interface, not having to click multiple links, and summarised answers fit almost all my use cases.
If I’m using Google, it’s because it was out of habit or because of work (I work in SEO, so as long as people still use Google, I’ll have to use it.)
The founder of Perplexity, Aravind Srinivas, tweets that Perplexity is what Google wanted to become 23 years ago. They just didn’t have the right technology.
Perhaps if we want to predict Google’s future, we should see where Perplexity moves next.