Machine learning

The development of machine learning has opened up new avenues for achieving therapeutic AI, even as much from science fiction remains beyond our grasp.

In response to an unmet need

Are there any reasons to believe AI therapy could be as effective as human therapy? This is the case based on early evidence.

Using a virtual human to share negative emotions reduced negative emotions and provided emotional relief, according to a study published in Computers in Human Behavior.

An AI therapist could have a similar or magnified impact if even virtual humans without therapy-specific behaviors can offer tangible benefits.

An AI Road Map for Therapeutics

The question arises, however, as to whether therapeutic AI is technically feasible. The type of therapy one is seeking to replicate may significantly impact this.

However, on the whole, there is one therapeutic modality that AI can repeat, that of Cognitive Behavior Therapy (CBT).

A hallmark of CBT is role-playing to prepare for potentially problematic interactions and desensitization to help calm the mind and body.

The study of the human dopamine system has revealed numerous ways that simple visual or auditory cues can alter our behavior.

We are indeed far more programmable than we may like to believe, and one of the jobs of a therapist is to reprogramme us so we engage in fewer self-harming or maladaptive behaviors.

If a person is experiencing high anxiety levels due to an environment, the AI can adjust the content to lower their anxiety levels.

In this way, the client can gradually become desensitized to the triggering stimuli and learn new behaviors that are not triggered by trauma.

These adaptive and dynamic content mechanisms are still in their infancy, but wisdom suggests looking beyond mere entertainment to use these tools to ease society's trauma burden.

Applying machine learning to many forms of therapy, including talk therapy, may still be a long way off (though chatbots may soon be an answer to this problem).