If you have already used ChatGPT, go to 'Differences'.
Large Language Models (LLMs) are advanced language models that are artificial intelligences capable of analyzing, understanding, and generating human language. They are powerful tools used in the field of Natural Language Processing (NLP).
LLMs are based on deep neural networks, such as Transformers, enabling learning from large text datasets. The main task of LLMs is processing and generating text, which means they can perform various language-related tasks, such as:
- Machine Translation: LLMs can transform text from one language to another, attempting to preserve its meaning.
- Text Generation: they can create long texts, such as articles, stories, or poems, to varying degrees simulating works written by humans.
- Speech Recognition: they assist in speech recognition and processing it into text.
- Sentiment Analysis: they can recognize emotions expressed in texts and assign them appropriate labels.
- Answering questions: LLMs can answer questions based on available information.
- Text summarization: They can shorten long texts while preserving essential information to create a summary, overview, or list of key points.
Popular examples of LLMs include models such as GPT-3 (Generative Pre-trained Transformer 3) created by OpenAI, BERT (Bidirectional Encoder Representations from Transformers) developed by Google, and many others.
LLMs are of great importance for various applications. They are used, among others, in voice assistants, translations, content generation, textual data analysis, and in many other fields that leverage and can benefit from natural language processing.
Among the most well-known LLMs, ChatGPT from OpenAI and Bard from Google stand out. ChatGPT has been available for some time, while Bard was introduced in February 2023 and, though somewhat less popular, also finds its applications. To get to know Bard better, we decided to test it by putting it through several tasks and checking how its performance differs from the freely available version of ChatGPT.
Differences
Google LLM already emphasizes at the beginning that it is in the learning phase and asks us to keep that in mind. However, the first significant difference that stands out is the fact that Bard is connected to the internet and uses Google Search. This feature is also available in ChatGPT, but only in the paid version. Therefore, Bard has a clear advantage in this respect.

Another aspect that draws attention is Speech Recognition. Unlike ChatGPT, Bard allows dictating prompts. Bard handles this task quite well, unfortunately, for now, it can only be spoken to in one language ?

Bard also gives us the opportunity to listen to its responses – a nice gesture towards users and the idea of accessibility. In this case, it copes not only in English.
Bard certainly wins when it comes to working with images compared to ChatGPT – it offers the ability to add pictures to its prompts for: description, text suggestions based on the image, supplementing information, and also searching for similar images on the internet.

Plugins
One of the significant advantages of ChatGPT is the plugins, which Bard currently lacks. Plugins are relatively easy to install and significantly enhance the usability of ChatGPT. Below, we present a few noteworthy additions that may prove useful.
WOLFRAM
The Wolfram plugin makes ChatGPT more reliable and useful – it gains access to computations, accurate mathematics, curated and verified knowledge, real-time data, and visualizations. All of this is achieved through the use of Wolfram Language and Wolfram Alpha.
ChatWithPDF
This plugin is an ideal solution if we are short on time or need to work with a large amount of data in PDF format. For example, when a regular search in the file doesn't yield the expected results because the key word appears in the document 2137 times. ChatWithPDF allows you to interact with your PDF document in the form of a conversation.
LinkReader
If you use ChatGPT's free version 3.5 on a daily basis, there may be times when you miss access to internet knowledge. In such cases, it's worth using LinkReader, which allows you to enrich the prompt with knowledge from a link provided by you.
KeyMate AI Search
Although Bard outpaces ChatGPT with its integrated ability to use Google Search, ChatGPT has also been equipped with a plugin that opens up the same possibilities for it.
ShowMe
If your work involves not only textual data or you prefer visual analysis, this plugin can be a significant tool. Show me helps visualize textual data using graphs, diagrams, charts, or images, utilizing Google Images, Plotly, Google Maps, and Draw.io.
Regardless of the help of which LLMs we use, we should always remember that the way they operate and learn will always expose them to the possibility of hallucinations – often taking a very convincing form that is easy to be deceived by its truthfulness. This is because each LLM knows only as much as it has been taught. The more examples it has received on a given topic, the better it will isolate and describe that concept. However, it doesn't know how little it knows. When asked about something it doesn't know, it starts building connections based on 'distant associations,' confabulating.
Hence, among other things, the usefulness of plugins such as LinkRead or Wolfram, allowing, for example, to feed LLM with context-specific queries, verified knowledge.
Which one is better?
The answer, of course, is "it depends." Both popular LLMs have their limitations and solutions useful in many everyday and extraordinary situations. Therefore, it's worth trying them in the context of your needs and... using the one that suits your specific task better.