Artificial intelligence (AI) might soon be able to “read” dreams. With the right prompts, AI tools like OpenAI’s Sora or Luma AI’s Dream Machine can already create video sequences resembling dreams. However, these sequences can be unsettling and remind one of nightmares.
Despite the progress, AI is still far from being able to accurately reproduce human dreams. Yet, this is not impossible, as explained by Peter Bentley, a British AI researcher from University College London, in an interview with BBC Science Focus.
A key challenge is teaching AI the necessary skills. Training AI requires a lot of data, which can only be generated by reading and analyzing human dreams. The tool for this purpose already exists: functional magnetic resonance imaging (fMRI). Bentley highlights a promising research approach in this field.
In a Japanese study from 2023, researchers recorded the brain activity of sleeping participants and used machine learning to classify recognized objects like keys, people, or chairs. However, this analysis focused on the first two sleep phases, where visual images are experienced, but actual dreaming has not yet occurred. This was done so participants could be quickly woken up and describe what they saw.
To enable AI to recognize and reproduce human dreams, much more detailed data is needed. This would require many dreaming volunteers who can accurately recall their dreams upon waking.
Bentley envisions the development of portable and quiet fMRI devices to ensure good quality sleep during analysis. However, he cannot predict when this will be possible. He is optimistic that we are not far from this goal.
Studies in related fields give Bentley confidence that research is on the right track. These studies have examined the brain activities of people watching videos, listening to audio recordings, and reading texts. According to Bentley, large datasets of the corresponding images already exist and could be used for dream analysis.
The potential of AI in dream analysis is vast. It could provide insights into the human mind and help understand the nature of dreams. However, ethical considerations must be addressed, such as privacy concerns and the implications of accessing personal dreams.
As AI technology advances, the possibility of reading and interpreting dreams becomes more feasible. Researchers continue to explore the intersection of AI and neuroscience, aiming to unlock the mysteries of the human brain.
The journey to dream-reading AI is ongoing, with promising developments on the horizon. Scientists and technologists are working together to push the boundaries of what AI can achieve. The future holds exciting possibilities for understanding dreams through the lens of artificial intelligence.