AI models

Probably the best-known model among AIs. It was published at the end of 2023 and has been indispensable ever since. It is developed and managed in the USA by the company OpenAI. Some of the results are only moderately transparent. Although there are good explanations, even the developers of the AI do not always understand why the answers turn out the way they do. It is therefore important to check the sources and results. The answers are based on training data and partly on web access. The strengths of ChatGPT clearly lie in its creativity, versatility and programming support. The AI's answers are often humorous, colloquial and unexpected, which makes interaction with ChatGPT more exciting.

ChatGPT is free (with under 16 requests every 3 hours), but has a paid feature with advanced options and access.

You can find presentations on ChatGPT here:

Developed by Google, this AI is already perfectly integrated into all other Google tools. In the meantime, Gemini results are delivered with standard Google searches. The results are also only moderately transparent, as Google remains vague in this respect. In addition, the AI does not always process prompts completely. It is therefore important to check carefully whether the answers contain all the important points. Gemini's strengths lie in its many functions and detailed, analytical interactions. Like all Google services, Gemini is free of charge - we pay with our data, so to speak. However, there is also a paid option with extended functions.

Le Chat was developed by the French company Mistral AI and is currently the only European AI model on the market and therefore GDPR-compliant. The focus here is on high transparency of results through a source and document function. Le Chat currently has fewer additional functions (e.g. no audio function) and a smaller user base compared to the other models. Here too, there is a free and a paid version. The AI is EU-compliant, data protection-friendly and powerful.

You can find presentations on Le Chat here:

The AI developed by Meta is now integrated into all of the company's social platforms. These include Facebook, Instagram and WhatsApp. Thanks to the direct integration, the function can be used in the messenger services of the platforms and there is no need to switch back and forth between different applications. Meta AI is based on LLaMA, Meta's Large Language Model, which is partially open source, meaning it can be used and developed further free of charge. Meta emphasizes that it wants to promote the democratization of AIs. In the EU, Meta AI can currently only be used as a word processing assistant; other functions, such as image creation, are restricted due to a lack of data protection. The concerns here lie in the use of user data to train the AI. It can therefore be helpful to familiarize yourself with the data protection settings of the respective platform and adapt them accordingly.

Perplexity also comes from the USA. This model is actually an AI-based search engine and works with the results of other models, such as ChatGPT and another AI. The transparency of the results is high here, as the results can be checked against sources. Perplexity shows weaknesses in more detailed analytical queries. This model is particularly suitable for research purposes, such as product comparisons. You can submit 5 queries per day here, further queries are subject to a charge.

Grok was developed by xAI, Elon Musk's AI company. This AI model is integrated into the social media platform X and can provide up-to-date information in real time. Similar to ChatGPT, Grok is known for providing humorous and unexpected answers, which increases user-friendliness. With this AI model, as with many other models, there are concerns about data protection, especially with regard to the use of user data to improve the AI.

As with Meta AI, Grok's source code is open source and therefore accessible to developers and researchers who can study and further develop the AI.