I can still remember the first time I saw Stanley Kubrick’s masterpiece, “2001: A Space Odyssey” in the late 1960s. I was totally overwhelmed by Kubrick’s vision of the future and impressed with the effort at scientific accuracy brought to the project by his collaborator, Arthur C. Clark. This was unlike any science fiction cinema previously seen. There had been many great science fiction movies before this raising questions about the meaning of human life, our future in the universe, the nature of reality, the transcendence of human physical and mental limitations, and the presence of other forms of life or intelligence on or beyond the earth.
These began with the first science fiction film in 1902, “Le Voyage Dans La Lune” and continued with others such as “20,000 Leagues Under the Sea,” “Metropolis,” “The Day the Earth Stood Still,” “War of the Worlds,” “Forbidden Planet.” The film 2001 raised the bar substantially for science fiction and since then we have had intellectually challenging movies such as “Andromeda Strain," “Alien," "Blade Runner,” “Contact," “Avatar,” “Cloud Atlas."
No other movie has given such a challenging depiction of what our encounter with a superior alien intelligence might be like. From today’s vantage point, however, it may be that the movie’s most important philosophical musings were those concerning artificial intelligence. HAL was the sentient computer controlling every aspect of life on the Discovery 1 spacecraft until he seemed to malfunction under the weight of his duties.
Today we seem to be moving ever closer to the reality of powerful artificial intelligence. Watson, the IBM computer system, successfully challenged the best human competitors in the history of Jeopardy. Many of us are using Siri on our devices. While none of these comes close to the level of Hal, they do show the ways in which we will increasingly deal with “alien intelligence” of our own design.
Psychotherapy is not immune to these developments. The first demonstration of a computerized therapist was Eliza, a Rogerian “therapist” created in the 1960s by Joe Weizenbaum. Using straightforward algorithms, it was surprisingly compelling while being simple enough that a later version of it could be run on my 1985 Commodore 64 computer. This “therapist” could not pass the Turing Test because it was easily tripped up grammatically if questions were asked in a certain way. It is not, however, necessary to actually use powerful artificial intelligence to computerize some psychological treatments.
For example, a number of behavioral treatments can be implemented in computerized virtual reality. These programs are being used to help patients overcome the fear of heights, spiders, and even PTSD symptoms in veterans returning from war. The empirically supported psychotherapeutic approaches for insomnia can be delivered by therapists trained to use manualized interventions; it was only a matter of time before these methods would be computerized and shared on the Internet.
A number of systems already exist and some are commercially available. Last year the annual meeting of the national sleep societies held in Boston had an entire seminar that was devoted to these relatively new systems. I will briefly review several of these below. All use the empirically supported techniques that I have discussed in previous posts but deliver them with online questionnaires, lectures, and interactive learning experiences.
Professor Colin Espie at the University of Glasgow, Scotland and has developed a program that uses an animated virtual Dr. Espie as the therapist. Lee Ritterband, Ph.D., and his colleagues at the University of Virginia are using an online program for research, with interactive learning modules. Norah Vincent, Ph.D. at the University of Manitoba, Canada, has created a CBT program for insomnia and sleep. This system uses online recorded lectures and questionnaires. Gregg Jacobs, Ph.D., has set up another program and uses PDF format lesions and MP3 recordings to deliver information.
Some empirical evidence exists that these online treatments actually work. Dr. Vincent, using her online program, conducted an extremely interesting study, a randomized controlled trial with a five-week treatment intervention on 118 adults who had chronic insomnia. A waiting list control was used. Treatment was received at home and was composed of education, sleep hygiene, stimulus control, sleep restriction, relaxation training, cognitive therapy, and medication tapering. The attrition rate was 33 percent (treatment referrals, probably with more significant insomnia, were more likely to drop out than those recruited from the community) and online treatment resulted in significant improvement in sleep quality, insomnia severity, and daytime fatigue, as well as reduced pre-sleep arousal and dysfunctional thoughts about sleep
It seems that Dr. Espie’s concept of stepped care for insomnia probably applies to these systems as they lack the true interactive and creative aspects of intelligence, artificial or natural. That is, in a stepped care model as used in Europe, for simple cases of insomnia a self-help program using a workbook may be sufficient while deeply entrenched, severe, and complex cases may require intervention from a highly trained behavioral sleep medicine specialist. The current online systems are probably good for the middle-range cases, a large number of patients who can benefit from less than the most comprehensive treatment program but who need something more than a workbook to guide them. With growing artificial intelligence capabilities, future human therapists may find it increasingly difficult to stay ahead of these systems. If and when artificial intelligence exceeds human intelligence, I hope that we have worked out an alternative economy that will allow all of the humans displaced by robotics and artificial intelligence to have a meaningful role in society.
While it is unlikely that these systems will replace human therapists in the near future, it is already clear that they can be used to treat many patients successfully. The near-term promise is that they will increase the impact that important breakthroughs in cognitive behavioral therapy can have. At the rate improvements in natural language systems are occurring, like with IBM’s Watson, Apple’s Siri, and Google’s Google Now, it seems that HAL, for better or worse, may be coming. And that could mean sweet dreams or nightmares.