Where does the AI-powered assistant bring benefits?
ChatGPT 3.5 is an LLM (Large Language Model) developed and permanently trained by OpenAI. It uses the power of the transformer architecture, a technology that enables revolutionary machine learning.
ChatGPT was created to respond to a wide range of requests using natural language. The assistant can be used for a wide range of applications in everyday work:
Customer Support Chatbots
ChatGPT is able to provide fast and accurate responses to customer inquiries. Customer-specific knowledge of the products can be achieved by fine-tuning the model.
Text generation
Based on given requirements ChatGPT manages to generate texts. This is useful for social media captions, websites, etc., for example, and speeds up the generation of these texts. Our marketing team appreciates the inspiration provided by Artificial Intelligence.
Conversational AI
ChatGPT offers the possibility of human-like communication, i.e. responding to posed questions and providing adequate answers.
Information Retrieval
Information can be filtered from large amounts of data and relevant answers to questions and can be provided in natural language.
What are the risks of AI?
Like any new technology, AI has risks. Simply put, ChatGPT is a good storyteller, but sometimes he fibs a bit. Checking AI-generated content (fact checking) is thus becoming mandatory. The ChatGPT 3.5 model was trained with data from the Internet, but does not have access to actual data. So good at history, but simply ignorant to current news.
Online meets ChatGPT
At the Online have also been thinking about where we can incorporate ChatGPT into our daily business. In the course of our ShipIT-Day (developer day) a group got together and dealt with this question.
Generate product descriptions with AI
One of the challenges in eCommerce is writing targeted product descriptions. Balz Zürrer, CEO, and Michal Kulecki, software developer, addressed this issue. With the help of bubble.io they created a functional landing page.
By means of simple text input one defines certain parameters such as product name and specifications as well as the language tonality (marketing, technical, etc.). Descriptions in other languages can also be generated by entering e.g. “Language: German” in the Description field. A bidirectional API interface connects bubble.io (leading low-code application) to GPT 3.5 and thus sends the request for the text. GPT 3.5 sends the requested text modules in different execution variants back to bubble.io in the shortest possible time. The construct went from idea to practice in a matter of hours, making it a great example of how AI can support humans in their work.
Convince yourself of our GPT Product Description Writer.
