Originally published on Medium last year, this article has been updated to reflect the latest advancements in AI technology. With recent updates to OpenAI’s ChatGPT, Google’s Gemini, Inflection’s Pi, and Anthropic’s Claude 2, it’s time to reassess these popular large language models (LLMs) and see how they stack up against each other.
What is an LLM?
For those unfamiliar, an LLM is a type of artificial intelligence (AI) model trained on vast amounts of text data. These models generate human-like text by predicting which word is most likely to follow the previous one. While the technical details are complex, the gist is that LLMs can create coherent and contextually relevant responses.
User-Centric Review
Before diving into the comparison, it’s important to note that this review is from a user perspective, not a technical analysis. I’ll be comparing the free versions of these chatbots, so keep in mind that paid versions may offer additional functionalities. All opinions are based on my personal experiences with these AI tools.
OpenAI’s ChatGPT (With Microsoft Copilot)
ChatGPT remains my go-to AI tool, perhaps because it was my first experience with generative AI. However, I’ve noticed a slight decline in its performance recently. It seems to require more detailed prompts to produce the desired results, which can be frustrating.
Despite this, ChatGPT still excels at coding assistance, especially in Python, and offers a wide range of capabilities, from solving math problems to writing essays and summarizing large texts. It also supports various data formats, including tables and JSON. Each chat session is automatically saved and named, making it easy to revisit past conversations.
One of ChatGPT’s strengths is its relatively low rate of AI hallucinations, where the AI presents false information as facts. While other chatbots are catching up, ChatGPT generally delivers accurate responses, provided the prompts are well-designed.
The main drawback of the free version is its data limitation, only covering up to January 2022. This means it can’t analyze current events, which can be a limitation for certain tasks. Additionally, it doesn’t support input or output of images or other file types, unlike the paid version, ChatGPT 4.
A great alternative to the paid version is Microsoft Copilot, available in preview for Microsoft 365 users and through the Edge browser. Copilot runs on ChatGPT 3.5 but is connected to the internet and can generate images using OpenAI’s DALL-E 2. It also integrates seamlessly with Microsoft apps, making it a convenient tool for those already using the Microsoft ecosystem.
Google’s Gemini
Gemini, previously known as Bard, competes closely with ChatGPT in terms of capabilities, such as coding assistance, math problem-solving, and data analysis. However, Gemini’s key advantage is its internet connectivity, allowing it to provide real-time information.
While Gemini has improved since its Bard days, particularly in reducing AI hallucinations, it still has room for improvement. Its built-in Google search verification feature is helpful but not always reliable, especially when Google can’t find exact matches for Gemini’s answers.
Gemini can handle more information than Bard, but it struggles with complex analysis when given large datasets. Overall, Gemini is a solid choice for analyzing current events, but I wouldn’t rely on it exclusively. If you work with Google apps, Gemini’s integration makes it a strong contender.
Inflection’s Pi AI
Pi is the most conversational AI among the four, often adding emojis and casual language to its responses. It’s also the most entertaining, making it a great option for light-hearted conversations or discussing life and career topics.
However, Pi falls short in areas like coding assistance and data summarization. Its character limit for prompts has increased, but it still can’t handle large datasets effectively. Conversations are typically long and linear, although you can create separate threads for different topics.
Pi shines in its conversational abilities, offering empathetic and engaging responses. It’s not the most technically proficient chatbot, but it’s the one you’ll want to talk to when you need a virtual friend. Just remember that Pi is an AI and not a substitute for professional advice.
Anthropic’s Claude AI
Claude initially left me unimpressed, but its latest version, Claude 2, has significantly improved. It now offers better coding assistance, data summarization, and even the ability to summarize PDF files (with some limitations). In my experience, Claude 2 provides more accurate summaries of large datasets compared to ChatGPT 3.5 and Gemini.
However, Claude is still not connected to the internet, and its coding format can be cumbersome to use. The free version also limits the number of responses per session, so you’ll need to manage your queries carefully. Despite these limitations, Claude 2 has made significant strides and is worth considering for certain tasks.
Conclusion: Which AI Reigns Supreme?
Each of these chatbots has its strengths and weaknesses, making it difficult to declare a definitive “best” option. The choice ultimately depends on your specific needs:
- For coding and data tasks: ChatGPT (especially with Microsoft Copilot) remains a strong choice.
- For real-time information: Google’s Gemini is your go-to AI.
- For conversational and emotional support: Inflection’s Pi excels.
- For accurate data summaries: Anthropic’s Claude 2 is a rising star.
Let me know your thoughts in the comments, and don’t forget to follow us for more.
