AI Chatbots and Assistants – How do the top 3 differ?

As a marketer, I know we’ve been hearing for a few years now that AI is going to take over our jobs. But I’ve also seen enough content to show a) that’s not true and b) utilization of something like ChatGPT can be advantageous to help make everyday tasks easier. (Think of how pivot tables can change your life with data in Excel) I’ve personally used Google Gemini (back when it was called “Google Bard”) and then dove headfirst into ChatGPT because hey…it’s ChatGPT, it’s the most well-known AI assistant out there. Today I discovered another platform, Perplexity, and it had me thinking…what’s the difference? And is there one better than the other? For that last question—the answer is technically yes, but it depends on what you want to use it for, so let’s start with the OG: ChatGPT.

ChatGPT

ChatGPT is an AI chatbot developed by OpenAI, an artificial intelligence research company, and was launched in November 2022. It uses natural language processing (NLP) to generate humanlike conversational responses. ChatGPT can answer questions and create content such as articles, emails, essays, social media posts, and code. As a form of generative AI, it produces text, images, or videos based on user prompts. Like customer service chatbots, it allows users to ask questions and request clarifications. The name stands for “Generative Pre-trained Transformer,” referring to how it processes and formulates responses. ChatGPT improves over time through reinforcement learning, using human feedback and reward models to refine its answers. [1]

ChatGPT is good for things like:

• Writing and Content Creation: ChatGPT can generate articles, blog posts, marketing copy, stories, and emails with humanlike fluency. [2]
• Answering Questions and Providing Explanations: It can respond to factual or conceptual questions across a wide range of subjects, offering explanations at different levels of complexity. [3,4]
• Brainstorming and Idea Generation: ChatGPT is a powerful tool for brainstorming names, strategies, creative ideas, or solutions to problems. This has been widely adopted in fields like marketing and design. [5]
• Learning and Tutoring Help: Educators and students use ChatGPT for tutoring in math, science, language, and more. It can break down complex ideas into understandable explanations. [6, 7]
• Coding and Technical Help: Developers use it to write, fix, and explain code. It supports multiple languages and frameworks and can also generate documentation. [8]
• Translation and Language Support: ChatGPT can perform language translation, help with grammar correction, and explain idioms or expressions. [9, 10]
• Productivity and Organization: It assists users in creating schedules, plans, summaries, outlines, and to-do lists, helping streamline both personal and professional tasks. [11]
• Customer Support and Communication: Companies and individuals use ChatGPT to draft professional messages and simulate customer service interactions. [12]

However, there are spots where ChatGPT falls short:

• Accessing Real-Time Information: ChatGPT’s training data cuts off at September 2021, preventing it from providing up-to-date information on current events, weather, or stock prices. It cannot access or retrieve information from the internet to update its knowledge base. ​[13]
• Generating Long-Form Structured Content: While capable of producing short to medium-length texts, ChatGPT struggles with maintaining coherence and structure in longer pieces. Its responses may become repetitive or lose focus over extended content. [13]
• Understanding Complex Contexts: ChatGPT may misinterpret nuanced prompts or fail to grasp intricate contexts, leading to responses that are off-mark or lack depth. ​[13]
• Providing Accurate Citations and References: The model cannot reliably generate accurate citations or verify the sources of its information, which is crucial for academic and professional writing. ​[14]
• Handling Ambiguity and Uncertainty: Unlike humans, ChatGPT doesn’t effectively acknowledge ambiguity and may provide definitive answers even when information is incomplete or uncertain, potentially leading to misinformation. ​[15]
• Processing Large Data Sets: ChatGPT has limitations in handling extensive data inputs, often resulting in timeouts or incomplete processing when dealing with large files or datasets. ​[16]
• Understanding and Contextual Limitations: ChatGPT may struggle with grasping the full context of a conversation or prompt, leading to responses that are out of context or missing the nuance of the inquiry. ​[17]
• Bias and Ethical Considerations: The model can inadvertently produce biased or inappropriate content, reflecting biases present in its training data. This raises concerns about its application in sensitive contexts. [18, 19]

Google Gemini

Google Gemini is a family of advanced generative AI models developed by Google DeepMind, designed to rival tools like OpenAI’s ChatGPT by offering powerful multimodal capabilities—processing text, code, images, audio, and video. Initially introduced as Bard in March 2023, the chatbot evolved into Gemini with the launch of Gemini 1 in December 2023, marking a major leap in reasoning and AI performance. By February 2024, the Gemini 1.5 series introduced long-context processing of over 1 million tokens, enhancing its utility for complex tasks. Now rebranded from Bard to Gemini, the platform powers Google’s AI across consumer and enterprise tools, including Workspace, Android, and Google Cloud. [20]

Here are Gemini’s strengths:

• Multimodal Understanding and Generation: Gemini can process and generate content across text, images, audio, and video, enabling applications like detailed image descriptions and video content analysis. [21]
• Advanced Reasoning and Analysis: The Gemini 2.5 model introduces “thinking” capabilities, allowing it to reason through tasks before responding, enhancing performance in complex problem-solving scenarios. [22]
• Coding and Software Development: Gemini assists in code generation and understanding, offering developers support in writing, debugging, and explaining code across various programming languages. ​[23, 24]
• Integration with Google Services: Gemini seamlessly integrates with Google Workspace applications like Docs, Gmail, and Sheets, enhancing productivity by providing AI-driven assistance within these tools.​[25]
• Language Translation and Support: With its multimodal capabilities, Gemini can interpret and respond to various content types, including text, video, audio, and code, facilitating tasks like language translation and grammar correction.[26]

And then, of course, the weaknesses:

• Inaccurate Information Retrieval: Gemini may provide incorrect or misleading information, especially when used as a primary search tool. Studies have shown that AI models like Gemini can present inaccurate data in a confusing manner, highlighting the need for human oversight. ​[27]
• Limited Coding Execution: While Gemini can understand programming concepts, it cannot execute code, which restricts its utility for tasks requiring code testing or execution. ​[28]
• Bias and Fairness Issues: Gemini may inherit biases from its training data, leading to outputs that could be unfair or unbalanced. Addressing these biases is crucial to ensure the ethical and fair use of its output.[29]
• Explainability and Transparency: While Gemini offers explanation capabilities, these might not always be clear or easily interpretable, making it challenging for users to understand the reasoning behind certain outputs. ​[29]
• Quota Limitations: Users may encounter restrictions on the number of requests they can make within a certain timeframe. For example, Gemini in BigQuery has daily quotas for code completion, code generation, and other tasks, which can hinder extensive use. ​[30]
• Potential for Hallucinations: Gemini might generate outputs that are plausible (sounding) but factually incorrect or nonsensical, a phenomenon known as “hallucination.” This can lead to the dissemination of false information if not carefully monitored. ​[31]
• Contextual Limitations: In some cases, Gemini may struggle with understanding complex contexts or nuanced queries, leading to responses that lack depth or miss the intended point. [32]

This last one I JUST learned about today and is probably one of the most intriguing platforms I’ve run across—based on my research tendencies (and steadily growing post reference lists…)

Perplexity AI

Perplexity AI is an AI-powered conversational search engine that delivers real-time, sourced answers to user queries using advanced language models like GPT-4. Launched in 2022, it aims to enhance information retrieval by combining the depth of large language models with web browsing capabilities. Unlike traditional search engines, it presents direct answers with inline citations, making it easy for users to verify sources. Available via web, mobile apps, and browser extensions, Perplexity offers both free and Pro versions—the latter includes access to premium models (like Claude and Gemini), multimodal features, and document analysis tools. [33]

Where Perplexity shines is:

• Real-Time Information Retrieval: Perplexity AI performs real-time web searches, accessing the latest information from various sources, including websites, forums, social media, and news articles. This ensures users receive up-to-date and comprehensive answers to their queries. ​[34]
• Citation Tracking and Source Transparency: The platform provides answers with inline citations, allowing users to verify the origins of the information presented. This transparency enhances trust and enables users to assess the credibility of the responses. ​[35]
• Contextual Understanding and Natural Language Processing: Leveraging sophisticated algorithms, Perplexity AI comprehends and responds to complex questions, delivering concise and contextually relevant answers. Its ability to handle natural language queries makes it user-friendly and effective for diverse inquiries.[36]
• Integration with Various Data Sources: Perplexity AI integrates information from multiple data sources, providing comprehensive and diverse perspectives on topics. This integration ensures that users receive well-rounded and informed responses. ​[37]
• Document Analysis and File Uploads: In its Pro version, Perplexity AI allows users to upload various file types, including PDFs, CSVs, images, and text documents, for analysis. This feature is particularly beneficial for professionals and researchers who require insights from their documents. [38]
• Multimodal Capabilities: Perplexity AI supports multimodal inputs, enabling users to interact with the platform using text and images. This versatility enhances the user experience and broadens the scope of tasks the AI can assist with. [38]

And where that shine dulls:

• Accuracy Issues: Despite its sophisticated algorithms, Perplexity AI can generate inaccurate outputs or lack coherence, which is particularly concerning in fields requiring precision, such as legal or medical domains. [39]
• Handling Complex Scientific Queries: The platform may struggle with complex scientific topics beyond mathematics, leading to inconsistent or superficial responses. [40]
• Problem-Solving Limitations: Perplexity AI lacks robust problem-solving capabilities, which restricts its effectiveness in scenarios requiring deep analytical reasoning. ​[40]
• Subjective and Ambiguous Questions: The tool may not handle subjective or ambiguous questions well, as it relies on objective facts and data. [41]
• Rate Limits and Usage Tiers: Users may encounter restrictions on the number of API calls or searches within a specific timeframe, potentially hindering extensive research activities. ​[41]
• Potential for Hallucinations and Plagiarism: Perplexity AI may produce information that appears plausible but is inaccurate or fabricated, and there have been concerns regarding the potential for generating content that closely resembles existing sources without proper attribution. ​[41]
• Legal and Ethical Concerns: The platform has faced allegations of copyright infringement and improper content aggregation, raising questions about its adherence to ethical standards in information sourcing. [33]

So did I miss anything? Which is your favorite AI assistant?

References:

[1] Hetler, A. (2025, March 4). What is ChatGPT? WhatIs. https://www.techtarget.com/whatis/definition/ChatGPT

[2] Introducing ChatGPT. (2022, November 30). OpenAI. https://openai.com/index/chatgpt/

[3] Meo, S. A., Al-Masri, A. A., Alotaibi, M., Meo, M. Z. S., & Meo, M. O. S. (2023). ChatGPT Knowledge Evaluation in Basic and Clinical Medical Sciences: Multiple Choice Question Examination-Based Performance. Healthcare (Basel, Switzerland), 11(14), 2046. https://doi.org/10.3390/healthcare11142046

[4] Balhorn, L. S., Weber, J. M., Buijsman, S., Hildebrandt, J. R., Ziefle, M., & Schweidtmann, A. M. (2024). Empirical assessment of ChatGPT’s answering capabilities in natural science and engineering. Scientific Reports, 14(1). https://doi.org/10.1038/s41598-024-54936-7

[5]Lakhyani, S. (2024, November 14). 10 ChatGPT Prompts for Creative Brainstorming – Shushant Lakhyani – medium. Medium. https://medium.com/%40slakhyani20/10-chatgpt-prompts-for-creative-brainstorming-62ae571729a1

[6] Wardat, Y., Tashtoush, M. A., AlAli, R., & Jarrah, A. M. (2023). ChatGPT: A revolutionary tool for teaching and learning mathematics. Eurasia Journal of Mathematics Science and Technology Education, 19(7), em2286. https://doi.org/10.29333/ejmste/13272

[7] Dolan, E. W. (2025, February 15). AI math tutor: ChatGPT can be as effective as human help, study suggests. PsyPost – Psychology News. https://www.psypost.org/ai-math-tutor-chatgpt-can-be-as-effective-as-human-help-study-suggests/

[8] Dohmke, T. (2024, May 21). GitHub Copilot X: The AI-powered developer experience – The GitHub Blog. The GitHub Blog. https://github.blog/news-insights/product-news/github-copilot-x-the-ai-powered-developer-experience/

[9] Cable, D. (2025, January 2). How to use ChatGPT for Translation + Prompt tips. https://blaze.today/blog/chatgpt-translation/

[10] 9 Ways to Use ChatGPT for Language Learning. (2023, May 30). Scribbr. https://www.scribbr.com/ai-tools/chatgpt-language-learning/

[11] Using ChatGPT to schedule my day, remember my tasks, and help me motivate to do them has literally changed my life. (2024). Reddit. https://www.reddit.com/r/adhdwomen/comments/1fwzi71/using_chatgpt_to_schedule_my_day_remember_my/

[12] Chui, M., Hazan, E., Roberts, R., Singla, A., Smaje, K., Sukharevsky, A., Yee, L., & Zemmel, R. (2023, June 14). The economic potential of generative AI: The next productivity frontier. McKinsey & Company. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-productivity-frontier#introduction

[13] The 8 main Limitations of ChatGPT. (2023, July 28). SOCi. https://www.soci.ai/knowledge-articles/chatgpt-limitations/

[14] Research Guides: ChatGPT: ChatGPT Limitations. (2023, January 19). Blackhawk College Library. https://bhc.libguides.com/c.php?g=1301302&p=9568968

[15] Admin. (2023, June 7). The limitations and challenges of ChatGPT. COLA. https://www.cola.org/the-limitations-and-challenges-of-chatgpt/

[16] Is there a limit to provide large amount of data to ChatGPT for analysis purposes? (2024). Reddit. https://www.reddit.com/r/ChatGPT/comments/1f52vo7/is_there_a_limit_to_provide_large_amount_of_data/

[17] Persistent Issues with GPT Model: Loss of Contextual Understanding and Continuity in Conversations. (2024, January 24). OpenAI Developer Community. https://community.openai.com/t/persistent-issues-with-gpt-model-loss-of-contextual-understanding-and-continuity-in-conversations/600365

[18] ChatGPT: A comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. (2023). Internet of Things and Cyber-Physical Systems, 3, 121–154. https://www.sciencedirect.com/science/article/pii/S266734522300024X

[19]Wang, C., Liu, S., Yang, H., Guo, J., Wu, Y., & Liu, J. (2023). Ethical considerations of using ChatGPT in health care. Journal of Medical Internet Research, 25, e48009. https://doi.org/10.2196/48009

[20] Wikipedia contributors. (2025b, April 1). Gemini (chatbot). Wikipedia. https://en.wikipedia.org/wiki/Gemini_%28chatbot%29

I know, I used Wikipedia…don’t tell my college professors…

[21] Baddepudi, A., & Kilpatrick, L. (2024, November 25). 7 examples of Gemini’s multimodal capabilities in action. https://developers.googleblog.com/en/7-examples-of-geminis-multimodal-capabilities-in-action/

[22] Lawler, R. (2025, March 25). Google says its new ‘reasoning’ Gemini AI models are the best ones yet. The Verge. https://www.theverge.com/news/635502/google-gemini-2-5-reasoning-ai-model

[23] Weatherbed, J. (2025, February 25). Google Gemini’s AI coding tool is now free for individual users. The Verge. https://www.theverge.com/news/618839/google-gemini-ai-code-assist-free-individuals-availability

[24] Gemini Code Assist overview. (2025, March 31). Google for Developers. https://developers.google.com/gemini-code-assist/docs/overview

[25] Pierce, D. (2025, January 15). Google is making AI in Gmail and Docs free — but raising the price of Workspace. The Verge. https://www.theverge.com/2025/1/15/24343794/google-workspace-ai-features-free

I didn’t really think I needed a citation to let you know that a Google product integrated with other Google products—because Google does this with all of their products– but there’s apparently a nuance to this, as stated in the article.

[26] Coursera Staff. (2024, July 24). Google Gemini AI: Meaning, Capabilities, and Use Cases. Coursera. https://www.coursera.org/articles/google-gemini-ai

[27] Caddy, B. (2025, March 28). I’ve got bad news for you if you use ChatGPT, Perplexity, or Gemini as your main search tool – AI web search isn’t… TechRadar. https://www.techradar.com/computing/artificial-intelligence/ive-got-bad-news-if-you-use-chatgpt-or-any-other-ai-as-your-main-search-tool

[28] Gemini Lags Behind: No Code Execution/Visual Output? (2025). Reddit. https://www.reddit.com/r/Bard/comments/1hkc9iz/gemini_lags_behind_no_code_executionvisual_output/

[29] Tyagi, A. (2025, January 10). What is Google Gemini? Features, Usage and Limitations. Analytics Vidhya. https://www.analyticsvidhya.com/blog/2023/12/what-is-google-gemini-features-usage-and-limitations/

[30] Does Gemini Advanced have limits on how many messages you can send it? (2024). Reddit. https://www.reddit.com/r/GoogleBard/comments/1ap8ks3/does_gemini_advanced_have_limits_on_how_many/

[31] Gemini for Google Cloud and responsible AI. (2025, March 25). Google Cloud. https://cloud.google.com/gemini/docs/discover/responsible-ai

[32] Akter, S. N., Yu, Z., Muhamed, A., Ou, T., Bäuerle, A., Cabrera, Á. A., Dholakia, K., Xiong, C., & Neubig, G. (2023, December 18). An in-depth look at Gemini’s language abilities. arXiv.org. https://arxiv.org/abs/2312.11444

[33] Wikipedia contributors. (2025b, March 31). Perplexity AI. Wikipedia. https://en.wikipedia.org/wiki/Perplexity_AI

I know…another Wikipedia reference…

[34] M, P. (2024, October 23). Perplexity AI review: features, benefits, and alternatives. ClickUp. https://clickup.com/blog/perplexity-ai-review/

[35] Naik, P. (2025, March 9). What does Perplexity AI have that other tools don’t? Medium. https://medium.com/readers-club/what-does-perplexity-ai-have-that-other-tools-dont-0b931bf8479b

[36] Perplexity AI: What it is and how to use it. (2024, June 7). https://www.getguru.com/reference/what-is-perplexity-ai-and-how-to-use-it

[37] GLBgpt. (2024, May 13). Exploring the advanced capabilities of perplexity AI in modern search practices. https://www.glbgpt.com/blog/exploring-the-advanced-capabilities-of-perplexity-ai-in-modern-search-practices/

[38] Garn, D. (2025, January 31). How to use Perplexity AI: Tutorial, pros and cons. Search Enterprise AI. https://www.techtarget.com/searchenterpriseai/tutorial/How-to-use-Perplexity-AI-Tutorial-pros-and-cons

[39] Quill, A. D. (2024, November 22). Comparative GPT Review: Perplexity and Kompas AI | Medium. Medium. https://medium.com/%40DigitalQuill.ai/gpt-generative-ai-llm-review-comparative-analysis-between-perplexity-ai-kompas-ai-fe1623b43071

[40] Perplexity AI: Strengths & Limitations [Discussion] [Research]. (2023). Reddit. https://www.reddit.com/r/MachineLearning/comments/13jrwe0/perplexity_ai_strengths_limitations_discussion/

[41] LibraryGuides: A Guide to Artificial Intelligence: Perplexity. (n.d.). https://culibraries.creighton.edu/c.php?g=1334271&p=10213131