How Effective is Perplexity Pro, and How Can You Get Two Months of Free Access?

Make sure to note the expiration date of the promotion and don’t let this opportunity slip away.

Looking Back

Perplexity isn’t a new concept; I’ve already introduced it in a video back in February this year, which you can watch here.

In that video, I discussed whether Perplexity could be a strong competitor to ChatGPT. My conclusion was that Perplexity is better suited as a complement to ChatGPT, enhancing each other’s capabilities when used together.

At that time, ChatGPT faced some limitations, like its inability to access the internet. Perplexity, on the other hand, can connect to the internet to retrieve real information, which is crucial. While it might not be as creative or agile as ChatGPT, Perplexity excels in aggregating professional information and offering factual data support, giving it an edge in this respect.

Having developed for over half a year, Perplexity has now earned a solid reputation among its users. Just yesterday, I came across a comment from a user, L,.G., who mentioned switching his Chrome search engine to Perplexity, and he thinks it can replace Google in most scenarios.

Given Google’s dominant status in the search engine market, this is indeed high praise for Perplexity.

Scientific Research

Perplexity’s interface is quite straightforward, primarily featuring a series of past conversations and a simple dialogue box.

However, there’s more to this dialogue box than initially apparent.

Take a look — by default, Perplexity searches data from various sources (as set in the Focus option), covering all types. However, you can choose a specific category. What I find most appealing about Perplexity is its academic search functionality.

After selecting the “academic” option in the dialogue box, I inquired Perplexity about the latest research on GPT-4.

The results encompassed a range of topics, including early experiments in artificial general intelligence, medical applications, molecular research applications, and explorations in developing target models. Moreover, there was a collection of related literature as support. Upon delving deeper, I discovered over 20 related scholarly articles.

The majority of these 20+ articles were sourced from reputable academic platforms such as NIH, Semantic Scholar, Arxiv, and others, encompassing languages including Chinese and English. I randomly selected one to ensure that the references were legitimate and not just made-up citations.

Then, I asked a more holistic question:

What challenges and issues has GPT-4 introduced to scientific research?

Perplexity’s summary was spot-on. The first issue it identified was the challenge of data interpretation and visualization. Despite its substantial assistance, the results aren’t always reliable. In my experience, this is a significant problem, involving aspects like data extraction, type conversion, and accurately understanding our requirements.

Following that is the issue with SQL queries, something I hadn’t previously noticed. Mathematical reasoning also poses a problem, particularly with slightly complex arithmetic operations, where large language models often err.

In terms of legal analysis, it hasn’t yet matched the expertise of a professional tax attorney. However, I believe this is positive, as surpassing this level might lead to job market instability. Reliability is another concern. Although GPT-4 has achieved human-level performance in many benchmark tests, it has its limitations, such as hallucinations and misinterpreting facts. This underscores the need for services like Perplexity, which bolster the reliability of responses through accurate news and robust argument sourcing.

Education and Learning

You can utilize Perplexity for knowledge explanation too. For instance, the topic of Q Star was trending recently, and I used Perplexity to get an overview of this subject.

Perplexity provided an introduction to Q Star along with multiple data sources. Notably, when I inquired about how to begin learning Q Star, it didn’t just suggest courses immediately but first inquired about my proficiency in artificial intelligence and machine learning, offering beginner, intermediate, and advanced level options.

This demonstrates its intelligent approach in tailoring recommendations based on the user’s existing level of foundational knowledge.

After selecting the beginner level, it advised starting with fundamental topics such as machine learning, reinforcement learning, Q Learning, and A* algorithms. This personalized educational guidance, tailored to my actual skill level, is incredibly beneficial.

I’ve also recommended Perplexity to my students to assist them in learning about areas outside their own specialties. It enables beginners to avoid feeling overwhelmed and aimlessly searching, thereby making their learning process much more efficient.

You can even use it to assist in finding datasets. For example, at the end of this semester, I gave my students a major assignment to locate unique datasets and perform related analyses. Many students, while attempting to find datasets akin to searching for a needle in a haystack, often failed and felt extremely helpless and discouraged.

Actually, Perplexity can effortlessly handle such tasks. As an instance, I tasked Perplexity with locating a sample SQLite dataset.

It swiftly offered four options, and I randomly selected one to examine.

The data on this search results page supports Oracle and SQLite formats and even shows the size of the data (4MB). It’s akin to effortlessly finding what you’ve been seeking after a long search.

Laboratory

Perplexity also provides a Labs feature. While I haven’t completely grasped how it ties in with its primary functionalities, I’ve found it to be very useful.

In the lower right corner of its interface, there are multiple models available for selection, many of which are open-source.

For instance, models like CodeLlama, Llava, and the increasingly popular Mixtral 8x7B, as well as Mistral Medium, which is reputed to perform on par with ChatGPT 3.5, are available. I experimented with Mixtral 8x7B, a Mixture of Experts (MoE) model. I tasked it with creating a Snake game. It’s worth mentioning that I’ve previously asked several open-source models to develop a Snake game, but the outcomes were often unsatisfactory, with many undefined calls leading to errors upon execution. However, the programming prowess of Mixtral 8x7B here is noteworthy.

Observe that I didn’t accelerate the video in any way; the speed of this code output is astonishing, isn’t it? It’s like a swift “whoosh”. This exemplifies the rapid inference speed enabled by the powerful hardware driving Perplexity Labs.

Though the quick code generation is impressive, the crucial part is whether the game is playable. I copied the code into Visual Studio Code and attempted to run it. There were no errors.

I played it for quite some time, and then, in my diary that day, I wrote down a Chinese saying — — “Indulging in play loses one’s ambition.”

Configuration Settings

Before you start using Perplexity, there are several key settings to consider.

In the settings located in the lower left corner, you’ll find the “AI data usage” option, which determines whether the model can use your conversations for training purposes. I typically disable it to safeguard my privacy.

Furthermore, the system by default utilizes Perplexity’s own large language model, which, to be honest, offers mediocre Q&A performance. However, you can opt for GPT-4, which markedly enhances the outcomes.

Ensure that you activate the Copilot option within the dialogue box during conversations.

Otherwise, even if you’ve chosen GPT-4, you won’t fully leverage the capabilities of GPT-4 without enabling this option.

Drawing from previous experience, it’s not surprising that using GPT-4 incurs a cost. However, don’t worry, as we’re about to tackle this issue.

Special Offer

Initially, using Perplexity Pro came with a monthly fee of 20 USD, identical to the pricing of ChatGPT Plus. But currently, you have the opportunity to use Perplexity Pro for free for two consecutive months. This promotional offer was announced by Perplexity on December 23.

Here’s my referral link . If you sign up through this link, both of us will get a ten-dollar credit reward. Make sure to use the promo code “HOLIDAYS23” provided by Perplexity at checkout, enter it in the ‘promo code’ field, and click the ‘Apply’ button. This will make your first two months’ subscription free. The validity of this promo code lasts about a week (as per Perplexity’s official guidelines), but I recommend acting quickly to take advantage of this offer.

Conclusion

In this article, I’ve introduced you to Perplexity, which can be utilized as an intelligent search engine. It facilitates searches based on diverse materials, including news, datasets, and more, accessing various channels, not limited to platforms like YouTube, Reddit, and academic databases. Presently, there’s a promotional offer that allows free use of the Pro account for two consecutive months. Additionally, you can benefit from the extra discount through my exclusive referral link.

Enjoy your experience with Perplexity AI!

If you find this article useful, please hit the Applaud button.

If you think this article might be helpful to your friends, please share it with them.

Feel free to follow my column to receive timely updates.

To watch video content, please subscribe to my Youtube channel.

My Twitter: @wshuyi