Update: Our AI models have evolved

Sharing knowledge is a core value for the Training Data Lab research group, and as such, we engage with the wider academic community and beyond. Over the past year, we have harnessed the power of AI to generate content on social platforms, using models based on GPT-3.5.

Our experience with a chatbot based on GPT-3.5 to generate content was quite successful, as we have already shared publicly. This natural language model has revolutionised how we create and share content, allowing us to develop high-quality text, from simple sentences to complex articles.

However, we didn’t stop there. We have moved on to more advanced models: Llama 3.1 and Llama 3.2. These versions go beyond the capabilities of GPT-3.5 and Llama 2 and 3 in several key respects:

• Ability to generate longer texts: Llama 3 can handle tasks that require longer texts, allowing us to share detailed and complex information.
• Contextualisation and depth: These models are better equipped to understand the context of a conversation or task, which in turn facilitates the creation of relevant and pertinent content.
• Improved natural language understanding: With Llama 3.1 and 3.2, we can tackle complex linguistic problems with greater accuracy, which enhances our ability to generate relevant text.

We are now actively working to leverage these AI tools to strengthen our website. The new design is designed to provide a more attractive and user-friendly space to share our findings and research in a detailed format.

New communication strategies

At the same time, we are updating our AI tools and adjusting our communication strategy. Instead of following a constant presence on social media, we have decided to focus more on our research projects and share our results through a controlled and focused approach.

• Personalised emails: As we scale down our Twitter presence, we will also be sending emails directly to subscribers. These messages will be more detailed and informative about our projects, publications and achievements.
• Greater emphasis on our research: By prioritising the dissemination of our findings through a more controlled channel, we can ensure that our audiences receive relevant information more accurately.

Our goal remains the same: to serve our community better and share our research results with the academic world. As we move forward, we are committed to finding a balance between scientific productivity and dissemination, knowing that both are crucial to advancing our areas of study.

* AI-generated text
Read more about how we generate our content