Auckland researchers make world-first discovery into artificial intelligence

Sisters Zohreh and Maryam Doborjeh.

In a world first, AUT researchers have developed an artificial intelligence model, which can predict a person's choices before they have even made up their mind.

The work is based on a new type of artificial intelligence research called spiking neural networks, which was used to develop NeuCube, a machine learning system modelled on how the human brain learns and recognises patterns.

It was undertaken by a team from AUT's Knowledge Engineering and Discovery Research Institute (KEDRI), which includes AUT PhD students and sisters Zohreh and Maryam Doborjeh, their supervisor Professor Nikola Kasabov and Professor Alex Sumich from Nottingham Trent University.

In experiments, the sisters got 20 participants to watch a video of different beverage logos and recorded their brain data using an EEG headset.

That data was synced to the NeuCube algorithm, which learned and classified patterns from the participants' brains. It was able to predict their beverage choice 0.2 seconds before they consciously perceived the beverage. It also showed a clear difference between logos which were familiar to participants and those which weren't.

Maryam Doborjeh, who specialises in machine learning, says witnessing the NeuCube algorithm work was amazing.

"The brain is an amazing thing - it learns and remembers things and can recognise them before the person can. To get a computer to be able to do that will change the way we all live."

Zohreh Doborjeh, who specialises in the psychology element of the work, is interested in what the subconscious brain can tell us about a person's decision-making.

"We know that only 10 per cent of people's decisions are intentionally made, the other 90 per cent are made subconsciously by the brain based on previous experiences, history, genetics and other factors. This work will be a game-changer for marketing in particular."

This ground breaking work can have a number of uses, including neuromarketing, cognitive studies and crime solving. One potential application would be the ability to determine an offender in a police line-up if a victim has blocked out the traumatic experience.

Kasabov, the designer of NeuCube, says the finding will lead on to more research.

"Researchers and social scientists will use this to understand better how much bias or prejudice we have due to our sub-conscience; what are our true preferences in life, how can we communicate better, and how can we learn better?"