Last week, Team TITAN convened in Copenhagen at the home of Democracy X (formerly DBT) to discuss the recent progress and future plans for the TITAN platform. This meeting focused on evaluating the outcomes of the last co-creation session and the initial Living Labs phase, planning the integration of Large Language Models (LLMs) into structured Socratic dialogues for the chatbot coach, exploring guardrails for user protection, as well as examining exploitation opportunities for the upcoming year.
Co-Creation and Living Labs Feedback
The team began by reviewing the feedback from the co-creation sessions and the first prototype used during the Living Labs phase. The feedback was mixed, with positive responses to the chatbot's concept and functionality, but concerns about its long-term value. This feedback is invaluable for the agile development process, emphasising the need for continuous iteration and improvement.
Current State of the TITAN AI Platform
To date, two releases of the TITAN platform have been made, both of which included core functionalities for the chatbot. The chatbot is designed as a virtual coach, utilising Socratic questioning techniques to encourage users to critically evaluate information and determine its factual accuracy. This approach is crucial for fostering critical thinking skills, a major life skill in today's information-rich society.
Planning for Future Releases
The next release of the TITAN platform will include several enhancements:
Improved Workflow and Value Proposition: Refinements to the workflow and a clearer value proposition to enhance user experience and engagement.
Additional Landing and Support Pages: New pages to support users and provide more context and guidance.
Propagation Impact Assessment (PIA): This new feature will estimate and predict the propagation of news articles or posts on social media, helping users understand the potential impact of spreading fake news.
Large Language Models (LLMs) Integration
The integration of LLMs into the chatbot solution is a key focus area. While LLMs currently struggle with aspects of logic and reasoning, TITAN aims to innovate by implementing structured dialogues using controllers. There is even consideration of embedding these controllers within the LLMs themselves, if feasible, to enhance their effectiveness.
Standards and Guardrails
In addition to the technical advancements, the team is also working on establishing standards and implementing guardrails for the AI. Ensuring trust and transparency in AI operations is essential for user confidence and the ethical deployment of the technology.
The meeting concluded with the presentation of an updated development roadmap and firm commitments for the next couple of releases. Team TITAN is poised for continued innovation and improvement, driven by user feedback and a commitment to enhancing critical thinking skills through technology.
Stay tuned for more updates as the TITAN platform evolves and continues to make strides in AI-driven educational tools.
Comments