The Google Killer: How to Search Using AI (#51)

Dearest Non-Techie, 

Welcome to another edition of our AI-focused newsletter. This week, we'll be exploring one of the most underrated tools in the AI arena - Perplexity AI. Or as I call it - The Google Killer. 

You’ll soon see why. 

❓What is Perplexity AI?

Perplexity AI is like the lovechild of a traditional search engine and an AI chatbot (like ChatGPT). Launched in 2022 by a crack team of former Google and OpenAI employees, Perplexity aims to make search stronger by providing concise, relevant answers to your queries.

It looks simple, just look any other search engine: 

It’s the same simple interface we’ve got used to with search engines. Type in your query and away you go.

So, What Does it Do, Then?

A lot more than it initially appears from that oh-so-simple home screen.

It searches the internet

Like Google, it searches the web to find up-to-date information.  Most people don’t realise that chatbots actually don’t do this.  They’re usually drawing from a massive, but static data set, containing billions of pages scraped from the internet at a set point in time. That data set set gets updated every few months, but it’s definitely not live.

It’s important to point out that some chatbots, e.g. ChatGPT and Gemini do have internet access mode, meaning they can access live data. But their default mode is to use that static data set.

You only trigger internet access mode by using certain prompt types:

1. Explicit request for current information

• "Can you search the internet for the latest news on [topic]?"
• "What are the current stock prices for [company]?"
• "Find the most recent information about [subject]."

2. Time-sensitive queries

• "What's happening in the world today?"
• "What are the top headlines right now?"
• "Give me the weather forecast for [location] this week."

3. Requests for up-to-date data

• "What's the current exchange rate between [currency A] and [currency B]?"
• "Show me the latest statistics on [topic]."
• "Find the most recent scientific papers on [subject]."

4. Comparative queries requiring recent data

• "Compare the current market share of [company A] and [company B]."
• "What are the trending social media platforms this month?"
• "Analyse the performance of [sports team] in their last five games."

5. Fact-checking requests

• "Verify if [statement] is true based on current information."
• "Find the latest research that supports or refutes [claim]."
• "Check if there have been any updates to [policy/law]."

Perplexity will automatically search the internet, a bit like Google. But it delivers so much more.

Your own ‘Wiki’ style answer

When it comes to a Google search, you just get a selection of results.  We all know by now the top ones have paid to be there, then the rest are organic - which doesn’t mean they’re necessarily the best content, they could just have a great SEO game.

Either way, you still need to then click around in and out of various sites to find a proper answer.

With Perplexity, you get a kind of ‘Wiki’ style amalgamation of top answers. It looks like this (I searched ‘explain neural networks’):

You’ve got your sources clearly shown at the top, then a well structured, academic-style answer underneath.

How are those sources chosen?

Where Google prioritises those who will get the best clicks, Perplexity has its own internal crawling, trust and ranking systems that analyse which answers would be most helpful.

As a result, you tend to get extremely useful answers to even complicated queries. I’m constantly blown away by how precise and useful the answer is.

It cites sources for you

Speaking of sources, a HUGE challenge when it comes to AI chatbots is the fact they hallucinate.

What’s a hallucination?

The problem is, if you have a huge block of text, it’s really difficult to work out where the hallucinations actually are. You can prompt the likes of ChatGPT to provide hyperlinked sources for every assertion it makes, but it’s not foolproof.

Perplexity includes citations by default. Next to almost every sentence, you’ll see a little clickable number: see that little 3 below? That will take you directly to the source - a much easier way to edit AI content properly.

Images, Videos and AI Images too

Within that same search result, you get a very Google-image like range of images and videos in the same screen.

You also get the ability to create your own AI images (Pro only), based on the search you’ve just done (perfect for those researching blogs, creating webinars or creating guides). Pick your style: painting, photo, illustration or even diagram.

Important: The diagram function is a bit crap right now, but I’m sure it’ll get better. The rest are great.

Here’s a result I got from my ‘neural networks’ search:

You can do follow-up searches

On Google, once you’ve done a search, that’s it. Although it might recommend new searches in ‘people also asked’, you do have to start again from scratch.

With Perplexity, search is like a conversation. Once you’ve done one search, you can ask follow-up questions and dig far deeper into your original query. It’s this feature that makes it an absolute dream for researchers. It even suggests what those follow-ups might be with intelligent suggestions:

Important note: I will say, I’ve had some issues with this feature. When it works, it’s fantastic. But sometimes it does forget your initial search and feels like a completely fresh one. Teething issues, I suspect.

You can store your searches

This one is very simple, but very much needed. I’m surprised Google never did this.

Once again, with Google, searches are ‘one and done’. There is nowhere to store the results and come back to them.

Perplexity has this covered. It has a library feature, where you can set up different named folders and store all of your searches in one place. You’ve got two options:

1. Threads 

Store your search - from what you put in, to the responses you got, follow-ups, images, videos, etc. You can:

• Title the saved search
• Share it with others
• Come back and dig deeper
• Edit your original query

2. Collections 

A Collection is a group of Threads all together, usually on the same topic. You can:

• Group Threads together by topic
• Set privacy levels for each Collection
• Use a whole Collection to create content (see below)

It can create (strong) new content

All of the above is impressive enough, and already surpasses Google in many ways, but it also has a rather huge trick up its sleeve.

You can create NEW content. Just like you can on the other AI chatbots, but it’s strengthened by that search (or series of searches) you’ve just done.

Let’s look at an example, where I’d like to create a LinkedIn post based on my search on neural networks.

The results is actually fairly usable as a post. It needs an edit, of course - all AI-generated content does, but it’s much further along the first draft route than most other AI chatbots. And I haven’t had to attach extra information, as we already had it in the search.

Bonus: Want to get even better results? Paste this into Claude, along with a document of as many of your most successful posts as possible, and ask it to re-draft. Even better - attach lots of LinkedIn best practice guide PDFs, too.

And that’s not all…

Perplexity’s skills don’t stop there. It can also:

• Summarise long docs (though you can only attach one file at a time)
• Create snippets of code
• Solve maths problems

For a long time, I had the free version, as it’s so good. You also get 5 free Pro searches every day on the free version. Pro searches go a few steps deeper and you get even more intelligent responses. Now I’m suitably addicted, I’ve gone Pro as I use it all the time, and it’s now a tool I use multiple times daily.

So, there you go. Do you see why I call it the Google Killer now? Why would anybody use Google when you can have an experience like Perplexity?

Important Note: I am not in any way sponsored by Perplexity, I just like to share tools that I think are brilliant.

Give it a try and see what you think.

Thank you and until next week,

Heather

PS Want to work with me?

• Here’s my AI training brochure (prices going up in Sep)
• Check out my AI learning hub: just £25/mo
• Or just drop me a reply to this email, I’m a friendly sort

How are you finding the new "mini AI lesson" newsletter style?

Loving it, keep them coming

I preferred the old style

News for humans, by humans.

• Today's news.
• Edited to be unbiased as humanly possible.
• Every morning, we triple-check headlines, stories, and sources for bias.
• All by hand with no algorithms.

Sign up now!