Hardly any news feed appeared in the last few days without updates on ChatGPT – for experts an easy read, for the most readers hard to understand. With this post on the BIENE IT blog, we want to give you a brief overview of the hype issue and summarize the most important information about the OpenAI chatbot. And here we already come to the first special term: AI is an abbreviation for “Artificial Intelligence”. ChatGPT enjoys a high level of attention since its launch in November last year: after only two months, the application counted more than 100 million users worldwide. Among other things, the program is described as a significant milestone in the development of Artificial Intelligence. What makes the Generative Pre-trained Transformer, short ChatGPT so special?

Digital chatbots are not new per se: computer-based programs that are able to interact with real persons exist for quite some time now. But conventional chatbots are mostly topic-limited and often wrong with their answers: the ChatGPT mainly impresses with human-like interaction. Communication with the new AI chatbot is based on the principle of machine learning (ML in short), not on predefined answers as usual. The AI tool is built on a special model of Artificial Intelligence, called transformer. This allows the AI program to understand and operate on a large amount of data and then apply it accurately. The application remembers the conversation progress and keeps learning. This makes the dialogs less rough and more realistic: Artificial Intelligence that is intended to make the most natural impression possible. The self-definition of the OpenAI chatbot is: “My purpose is to engage in natural language conversations with users and provide them with accurate and relevant information on a wide range of topics. I am constantly learning and evolving as I receive more input and data from users, which helps me to improve my responses over time.” The AI program was trained on a large amount of texts data
from various sources, including books, articles, news, and websites. It was developed by OpenAI: first generations of GPT chatbots were released back in 2018. As for the technology, it is an in-house product of the AI pioneer. The Artificial Intelligence program is currently not connected to the Internet, so its knowledge is limited to the test data, which extends to 2021. The further development ChatGPT-4 provides for some changes, including real-time connection to the Internet as well as refined speech processing and conversation control.

ChatGPT: Use, perspectives and limits
The intelligent AI chatbot can be applied to a wide range of tasks, including text generation, translation and voice control. The main advantage of ChatGPT is time saving: thanks to its ability to perform human-like conversations, it can support customer service interaction, for example. Machine learning makes it possible for the AI tool to acquire more and more relevant knowledge and remember the previous conversations. Text generation is also of interest to marketing and sales departments, among others. The interactive AI program can furthermore improve the user experience in virtual and augmented reality applications.

The future of AI chatbots is promising. With continuous development and increasing amounts of data, follow-up versions of ChatGPT can take on more complex and challenging tasks and be used in e-commerce and online banking, among other fields. The use of AI chatbots in education and healthcare is also a possibility. But it is important to keep in mind potential risks and ethical concerns. As it is typical for any other Artificial Intelligence, ChatGPT can make mistakes or deliver problematic results. A filter is used to prevent the output of false information or violent and discriminatory content, but it doesn’t work properly at all times. There is also room for improvement in terms of understanding context: the OpenAI chatbot can have difficulty in categorizing complex contexts correctly. The AI tool cannot always respond accurately to emotional topics that require empathy. Also of concern, according to experts, are privacy and security issues: Artificial Intelligence relates to large amounts of data, some of which may contain sensitive and private information. These require special attention and explicit legal regulations.

Scripts and cyber security: what about ChatGPT in IT?
The AI chatbot can also be used the IT sector: Its script skills include common programming languages, such as Python, Javascript, C++, Java, PHP, Go, Swift, Typescript and PowerShell. Unlike traditional AI models that can write codes, ChatGPT is able to revise the codes it creates. If you find bugs, you can describe them to the AI program and ask it to make appropriate corrections. As a reminder, thanks to machine learning, the AI tool builds its capabilities with each request. There is no need to write a script from zero. So, ChatGPT can save a lot of time and effort, but it also poses risks. Hackers can take advantage of the AI program to generate malicious code. Further cyber security threats arise from the ability of the chatbot to produce error-free text. This makes it easier to create phishing emails: one of the criteria for unmasking them so far consisted of a big number of errors in spelling and grammar. A direct request to set up a phishing mail would be rejected by the AI program thanks to the built-in filter, but it composes a support mail asking for private data or offering malicious software for download without any problems. As mentioned above, ChatGPT cannot contextualize well yet.

Other IT risks include identity theft (the AI tool can simulate real people or organizations and request confidential information) and social engineering attacks (the ChatGPT can conduct human-like conversations and, in the worst case, persuade the conversation partners to take certain actions, such as downloading malware). Deception of customer support to illegally capture sensitive information is also possible. AI chatbots can help criminals improve the quality of their attacks: cyber security awareness is becoming increasingly important in this context.

How to use: try the ChatGPT for free
After informing you about the features, risks and opportunities of Artificial Intelligence, we are pleased to explain how you can use the AI Chatbot for free. You can access the ChatGPT via this link: https://chat.openai.com/. In order to use the application, you must register. During registration, you provide your email and phone number, and confirm your request with a verification link and an SMS code. In addition to free use, OpenAI also offers a paid version of ChatGPT. It stands out with more detailed answers, integration possibilities with other applications and priority support, just to name a few. If the website is not available due to many requests, just visit it again later. We from BIENE IT wish you a lot of fun with the innovative AI program!