Levente Szabados, co-founder of Neuron Solutions Ltd and scientific advisor, and Florian Ellsaesser, Professor for International Entrepreneurship at Frankfurt School of Finance&Management, held an interesting presentation about ChatGPT in business.
The presentation started by introducing the concept of intelligence and artificial intelligence. Levente highlighted that artificial intelligence refers to the ability of machines to perform tasks that require human intelligence, such as problem-solving, learning, reasoning, perception, and language understanding. He explained that language models like ChatGPT can play a crucial role in understanding and generating text, which is a fundamental aspect of human intelligence.
Levente then delved into the applications of large language models like ChatGPT and discussed how these models can be used for information extraction, sentiment classification, and communication. He provided examples of how ChatGPT can extract relevant information from documents or reviews, classify sentiments expressed in text, and generate summaries or responses for various purposes.
However, Levente acknowledged that there were challenges and considerations when it comes to the use of large language models in business. Three key areas were highlighted that need careful attention: grounding the generated text in accurate and reliable data, protecting sensitive information and ensuring data privacy, and integrating the models effectively into existing business systems and processes.
To tackle these challenges, he introduced the concept of grounding, which involves providing the models with accurate and reliable data to generate meaningful and valid responses. Levente emphasized the need for systematic solutions and demonstrated how language models can be connected to specific sources of information to improve their accuracy.
Regarding data protection, Levente addressed concerns about sharing sensitive information with external models, and mentioned open-source alternatives to proprietary models like ChatGPT, which allow companies to run models on their own infrastructure, thereby maintaining control over their data.
In terms of integration, Levente explained the importance of systematically integrating language models into existing business systems. He discussed the vision of software agents communicating with each other and accessing multiple parts of an organization to autonomously solve problems and achieve business goals.
The presentation concluded by highlighting the potential and challenges of using ChatGPT in a business context: emphasized the importance of addressing grounding, data protection, and integration to fully leverage the capabilities of large language models in the business world.
You can watch the full video in the following page:
https://www.frankfurt-school.de/en/home/newsroom/news/2023/Mai/ChatGPT-in-businesses
