Generative AI has achieved remarkable success in recent years. At its core, however, it is an advanced form of Autocomplete, as it only predicts the next word. This panel will explore how language models move from simply processing language to competing with doctors, lawyers, and politicians. It will also address their inherent limitations in reasoning and in aligning with the values of human society.
The topics to be discussed:
- which analogies are most helpful to understand LLMs;
- how similar LLMs are to human language processing;
- how good LLMs are in common-sense reasoning;
- how well society is prepared to AI;
- the future of work: Will AI take our jobs?
The panel consists of the AI experts from the University of Leeds:
- Serge Sharoff, Professor of Language Technology (chair)
- Faruk Mardan, Director of Applied Translation Studies Programme
- Marc de Kamps, AI Programme Lead
- Richard de Blacquiere-Clarkson, AI Hub Lead
- Tony Cohn, Professor of Automated Reasoning
- Vania Dimitrova, Professor of Human-Centred AI
