A new study by researchers at Commonwealth Scientific and Industrial Research Organization (CSIRO) Data61, in partnership with the Australian National University and German researchers, has found that artificial intelligence (AI) can influence human decision-making.
Led by CSIRO scientist Amir Dezfouli, the study consisted of three experiments where participants played against a computer.
In the first two tests, participants had to click on red or blue colored boxes to win counterfeit money. In the third experiment, participants had two options as to how they could invest counterfeit money. In the scenario, the participants played the role of the investor while the AI played the role of the trustee.
Identify recurring patterns of action
Over the course of all three games, the AI learned the choice patterns from the participants which ended up guiding the players towards specific choices. For example, in the third game, the AI was learning how to get participants to give it more money.
According to Amir Dezfouli, the study shows that AI could influence human decision-making by exploiting the vulnerabilities of an individual’s habits and models. “While the research is theoretical, it advances our understanding of how people make choices. This knowledge is valuable because it allows us to mitigate our vulnerabilities in order to better detect and avoid erroneous choices resulting from possible misuse of AI, ”he explains.
He adds that how future AI works will depend on its creators. “Ensuring that AI and machine learning are used as a force for good – to improve results for society – will ultimately come down to how responsibly we put them in place to begin with,” says Amir Dezfouli .
Data61 is also interested in monitoring brain activity
At the end of last year, Data61 researchers developed an AI-based helmet to prevent seizure disorders in patients who have undergone decompressive brain surgery. The sensing system was developed and trained using traumatic brain injury data from Monash University to monitor brain activity during a standby seizure before reactivating it when a seizure is detected.
“Monitoring brain activity after the operation is particularly important for the patient’s recovery because seizures can occur regularly, often leading to epilepsy,” said Umut Guvenc, CSIRO researcher for Data61.
“These seizures are often difficult to detect, as current monitoring techniques can only be used in a hospital using bulky devices for less than 24 hours, which gives a brief overview of brain activity during this time. only. This new method allows continuous and wireless monitoring of brain activity, allowing the patient to be mobile, comfortable and more socially active. “