Google DeepMind researchers have found that current AI models, specifically transformer models like OpenAI’s GPT-2, struggle to generate outputs beyond their training data. Despite large training datasets, these models fail to perform tasks outside their pre-training data domain. This finding challenges the notion of AI achieving artificial general intelligence (AGI), or human-level AI, as it suggests that AI’s effectiveness is limited to areas it has been thoroughly trained on.
Read more at Futurism…
