Edinburgh University teaches computers to binge on TV to solve mysteries

Gary Sinise, right, and Ted Danson star in CSI: NY. Picture: Cliff Lipson/ CBS Broadcasting
Gary Sinise, right, and Ted Danson star in CSI: NY. Picture: Cliff Lipson/ CBS Broadcasting
Have your say

Computers are being trained to spot the killer in fictional TV show CSI by bingeing on episodes of the popular forensic crime drama.

Artificially intelligent (AI) machines have been trained to channel their inner Gil Grissom by studying the hit show CSI: Crime Scene Investigation and have learned to identify the perpetrator in each case – but not quite as well as humans, yet.

The study at Edinburgh University was aimed at enabling machines to solve a problem – in this case identifying “whodunnit” – by assimilating information from images, audio, transcribed dialogue and scene descriptions.

The formula of the top TV series is almost always the same but figuring out who the perpetrator is not always obvious.

The AI watched 39 episodes of CSI: Las Vegas, involving 59 cases and correctly identified the perpetrator during the final part of an episode 60 per cent of the time. The study found people who watched the same show were able to identify the killer 85 per cent of the 

Researchers say such devices could tackle real-world tasks that need complex reasoning.

They studied if and how artificially intelligent machines can solve problems that are challenging for humans.

Dr Lea Frermann, one of the authors of the study, said: “Pinpointing the perpetrator in a TV show is a very difficult task for computers, but our model performed encouragingly well.

“We hope our findings will aid the development of machines that can take on board – and make sense of – large streams of information in real time.”

The computer model processed data in various forms – spoken, visual or textual – as the plot of each episode developed.

Scientists from the University of Edinburgh mapped footage, script and background sounds from five seasons of the show into a machine-readable format.

The data was fed into a computer model that learned to process the plot as each episode unfolded, continually revising the criminal’s identity.

And the computer’s successful identification of the fictional criminal, towards the end of an episode, was not far from human couch detectives, with a positive result 60 per cent of the time.

Scientists designed their computer model to solve arbitrary problems based on acquiring data.

Such devices could play a role in developing efficient 
algorithms for real-world tasks that require complex reasoning, researchers say.

Scientists taught machines to approach solving the crimes in the same way that people would – by considering which characters might be responsible from their behaviour as the plot unfolds.

The study, published in Transactions of the Association for Computational Linguistics, was funded by the European Research Council and H2020.

The American series Crime Scene Investigation, or CSI, began in 2000 and ran for 15 seasons.

Various spin-off shows followed the work of fictional forensic scientists teams across America and the technology they employed to solve crime teams.