Perplexity has emerged as the newest player in the AI sector to introduce an extensive research tool, with its latest feature revealed on Friday.
In December, Google launched a comparable feature for its Gemini AI platform. Shortly after, OpenAI debuted its own research agent earlier this month. Notably, all three organizations have named this feature the same: Deep Research.
The primary aim is to deliver more comprehensive answers with verified citations for professional applications, as opposed to the responses from a standard consumer chatbot. Perplexity detailed in a blog entry announcing their Deep Research feature, stating it “shines in a variety of expert-related tasks—from finance and marketing to product analysis.”
Currently, Perplexity Deep Research is accessible via the web, with plans to roll it out on their Mac, iOS, and Android applications soon. Users can activate the feature by selecting “Deep Research” from a dropdown menu upon submitting their query in Perplexity, which will generate a detailed report that can be exported as a PDF or shared as a Perplexity Page.
To craft this report, Perplexity explained that Deep Research “iteratively explores, examines documents, and reasons about what steps to take next, refining its research strategy as it gathers insights on the topic,” which is allegedly “akin to how a human researcher would investigate a new subject.”
The organization has also showcased its capabilities on Humanity’s Last Exam, a benchmark test comprising expert-level questions across various academic disciplines. Perplexity indicated that its Deep Research tool achieved a score of 21.1% on this assessment, surpassing many other models such as Gemini Thinking (6.2%), Grok-2 (3.8%), and OpenAI’s GPT-4o (3.3%)—although it fell short of OpenAI’s Deep Research, which scored 26.6%.
Currently, to access OpenAI’s Deep Research, users must subscribe to a $200-per-month Pro plan (with intentions to broaden availability to other subscription levels), while Perplexity’s Deep Research is offered at no cost—non-paying users are allowed a limited number of queries per day, while subscribers can enjoy unlimited access.
Moreover, Perplexity’s Deep Research appears to operate more swiftly, completing the majority of tasks in under three minutes, whereas OpenAI’s Deep Research might take between 5 to 30 minutes.
When prompted to compare various deep research tools, Perplexity provided an overview detailing the differing technologies, pricing structures, and performances across a range of applications and subjects, along with references to relevant articles about each feature. The company summarized the distinctions as follows:
- Perplexity AI stands out for its speed and user-friendliness for casual researchers
- OpenAI excels in analytical rigor for enterprise-oriented applications
- Google integrates most effectively with current productivity environments
It’s premature to gauge the impact these tools will have on both everyday and specialized research as they gain traction. However, The Economist recently pointed out certain deficiencies in OpenAI’s Deep Research that may also pertain to these offerings: including constraints on its “creativity” in analyzing data, a reliance on sources that are “readily accessible,” and a broader concern that “delegating all your research to a super-intelligent assistant” might “limit the chances of developing your most innovative ideas.”
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
