From the title, I expected some evidence that the AI was better than an actual human. That's not the case. The comparison was AI transparent vs. AI posing as human.
The posing AI could be arbitrarily worse than an actual human and still be better than itself when not posing.
> From the title, I expected some evidence that the AI was better than an actual human. That's not the case.
We don't know that. It's not that it's false, but that the evidence is not clear one way or the other -- it's just speculation at the moment.
We do know that programs like ELIZA, crude as they were, made many people feel that something worthwhile was taking place. In one case, a test subject asked Weizenbaum (ELIZA's originator) to leave the room so she could have a private conversation with the therapist.
Not quite the same thing, but self studied online CBT was the final nail in the coffin of my depression. I highly suggest checking it out for those who struggle with it, it helped me so much.
I read and used the early '80s edition, before the "behavioral" was added to it, and around a decade or so ago realized that learning and following it made further talking therapy essentially useless.
The theory behind the "cognitive" part of it, simply put, is that thinking bad, not to mention incorrect, thoughts about yourself makes you feel bad. Adjusting your mental filters and the like to reality can then make a very big difference.
(Not, in my case, a complete cure, I've got an inherited refractory depression that my doctors and I supposize is bipolar in nature, I just never go manic nor cycle all that much. Adding a SSRI to the mix significantly improves it.)
I would take it as a hopeful sign that we can cure ourselves with a little virtual assistance. Not unlike a spreadsheet to organize our thoughts about finances.
Is there a startup working on making this accessible to the average person?
I could see a huge advantage in paying, say 30 usd, a month for access to a shrink, as I believe most people could use one, much like even those of us who are not 300 pounds could benefit from a nutritionist.
It'll be a free app from Google (probably an extension to Google Now). Of course, your sessions will remain private within Google and only archived for automatic analysis by computers to improve the overall experience. You may notice improvements in your search results and targeted advertising as a result.
Shortly thereafter, Apple will announce "Tini", Apple's iShrink and Siri's cousin.
> A virtual shrink may sometimes be better than the real thing
It's true, and this speaks volumes about the value of therapy as a profession. The reason? It proves that people are competent to treat themselves, the real meaning of virtual therapy.
EDIT: Nothing is more annoying that anonymous cowards who, out of ignorance, consistently downvote posts consisting solely of easily verified facts.
Full disclosure: a paywall prevents me from actually reading the article. However, to your comment, I would say that this doesn't actually prove that people can treat themselves. If it were proven that people truly are competent to treat themselves, I think more people would have done so. Once you are in a cycle of despair, it isn't always easy to see any way out. Every single option is void of hope of any sort. Broken hope.
Virtual therapy is hope in itself. I remember my parents taking me to a therapist when I was 14-15. I got caught shoplifting and they wanted to make sure I was ok. I really wasn't - I know now that I'd been mildly depressed for years by that point - but the therapy was wasted. Why? A combination of the man being a Christian therapist (still with degrees) and them making me sign a release form - nothing I said to the man was private from my own parents. This was not a safe place and I never trusted the man. No trust and a high likelihood of judgement from multiple parties (and a high chance of disproval or over-attention) made me shut up. Virtual? I'd consider that now, especially if for free. Best of all, it is private. Safe. No one could even see me go into the therapists office. Other people have issues trying to find a therapist that they click with. A computer is allowed to be off, to be a little cold, to be somewhat matter-of-fact: Humans need a bit of caring and empathy to go along with it.
So much of the stuff online seems like a scam. My father, during a depressed phase, spent a thousand or two on a program that basically said 'think yourself happy'. Common advice, which doesn't work. Also basic is 'just have faith in god (generally christian, but not always) and it will pass'. Actual help in any form is in short supply. Good sources stop well before practical help and tell you to talk to someone.
If my words sound a bit crazy, and not like normal people think, good. This is the stuff I have dealt with for the last 25 years or so in varying degrees. I generally haven't had access to a therapist or even be able to afford a doctor, so I'm just stuck. It isn't always bad, but it is always in the background.
> I would say that this doesn't actually prove that people can treat themselves.
Yes, true, it proves nothing. It only suggests it.
> If it were proven that people truly are competent to treat themselves, I think more people would have done so.
Not if they think psychologists are doctors. Not if the real point is to have a conversation with a sympathetic listener who isn't a machine. There are plenty of reasons a virtual therapist wouldn't work, apart from the technical content of the therapy.
> ... nothing I said to the man was private from my own parents.
I've read cases like that also, and IMHO it's criminal to create a quid pro quo in which the advantage of a sympathetic listener is undercut by the fact that anything can be revealed to the parents, who are often the real problem.
> Best of all, it is private.
Maybe. If this virtual therapy thing takes off, people are going to have to be very careful about how the transactions are handled. The content of the sessions would be a gold mine, and I'm sure people will try to offer online therapy sessions with a way to preserve the session content for "research", all explained in excruciatingly small print.
> I generally haven't had access to a therapist or even be able to afford a doctor, so I'm just stuck.
This is not to diminish your situation or pretend to understand you, but there was a time when a sympathetic friend would have been the obvious remedy.
"a sympathetic friend would have been the obvious remedy"
This, I think, is what a lot of people need. I've had times with absolutely no one to talk to, I'm sure others are the same. Luckily my situation is better now.
Ehh, I don't know about that conclusion. Just because people are more open doesn't mean that the therapy is effective. That said, I wouldn't be surprised if you're right.
But at the end of the day, a good therapist (in my experience) is really just a sounding board that points you on the right direction. If you've had a recurring concern recently, they may be able to push the conversation in the right way to figure it out.
For those who haven't seen a therapist, the best analogy I have is rubber duck debugging; it's not that the listener solves your problems, but rather that they allow you to help yourself.
> Just because people are more open doesn't mean that the therapy is effective.
But when therapy works, it's not because of a particular therapeutic method, or psychology, or psychiatry. This has been proven over and over again -- when therapy works, its because of the personal skills of the practitioner, not the school of thought to which he/she belongs.
Studies show that in effectiveness, a therapist is indistinguishable from a sympathetic friend. For the majority of cases, the only reason therapists exist is the breakdown of the small village and a 19th century-style circle of neighbors and extended family.
I say this because there is a widespread misconception that therapy is based in science. It isn't. It's based on a social vacuum caused by the breakdown of small communities of individuals who talked to each other.
EDIT: Nothing is more annoying that anonymous cowards who, out of ignorance, consistently downvote posts consisting solely of easily verified facts.
This is completely orthogonal to whether a virtual therapist is as effective as a real therapist, which is the subject of the story. If treatment A and treatment B are equally effective, that doesn't tell you anything about treatment C. I think we can both agree that talking to a therapist or talking to a sympathetic friend (as you mentioned) have some efficacy, so the question of whether a virtual therapist can give the same effect is relevant.
Aside from all that, I'd love to read up on this research -- happen to have any links or recommended starting points? My knowledge of therapy simply comes from experience on both sides in the mental health field/community, not a scientific standpoint.
> This is completely orthogonal to whether a virtual therapist is as effective as a real therapist, which is the subject of the story.
No, it's completely relevant. Can a scientist's work be automated? No, not really -- too much originality and depth of thought for modern AI to imitate. Can a hamburger cook's work be automated? Well, yes, indeed it's happening.
If a virtual therapist is as effective as a real one, this calls into question the value of a real therapist, as well as the school of thought to which he/she belongs.
It may be instructive to know that psychiatrists have given up on talk therapy entirely and are now pursuing what they call "biological psychiatry", which is something vaguely resembling neuroscience, but with a lot of drug prescriptions being written.
> If treatment A and treatment B are equally effective, that doesn't tell you anything about treatment C.
True, but it does suggest the Placebo Effect as an explanation for the lack of difference. As it happens, a number of studies that compared therapies have come to that conclusion.
> Aside from all that, I'd love to read up on this research -- happen to have any links or recommended starting points?
>If a virtual therapist is as effective as a real one, this calls into question the value of a real therapist, as well as the school of thought to which he/she belongs.
The point I'm making is that just because therapy is equally effective as a sympathetic friend (as you mentioned), that doesn't mean that the virtual therapist is equally effective as well. The research simply hasn't been done to back that up.
Edit: To be more clear, nothing in this article speaks to the efficacy of the virtual therapist whatsoever. It could be completely useless, for all we know. I think you're making a lot of leaps here.
> The point I'm making is that just because therapy is equally effective as a sympathetic friend (as you mentioned), that doesn't mean that the virtual therapist is equally effective as well.
I only emphasized that because the linked article suggested that it might be so. Also the history of ELIZA and more recent programs suggest that people derive substantial benefit from them.
> It could be completely useless, for all we know.
The same could be said about conventional talk therapy -- surely you realize that research can't exclude the Placebo Effect as the primary agent of action in therapy (and many studies say just that).
> I think you're making a lot of leaps here.
I've studied this problem very extensively, it's not something I take lightly:
I'm unsure whether we're talking past each other or if my point simply isn't coming across. My issue with your comments has nothing at all to do with your issues with therapy, most of which I agree with you on. The issue I have is that you're making assumptions based on this research that the research doesn't back up. From your original comment:
> It proves that people are competent to treat themselves, the real meaning of virtual therapy.
This is simply not supported. Full stop. It's especially not supported by research that has nothing to do with the efficacy of virtual therapy.
>> It proves that people are competent to treat themselves, the real meaning of virtual therapy.
> This is simply not supported.
Certainly not, when taken out of context as you have done. Here's the original context, with the parts you deleted:
>> A virtual shrink may sometimes be better than the real thing
> It's true, and this speaks volumes about the value of therapy as a profession. The reason? It proves that people are competent to treat themselves, the real meaning of virtual therapy.
Now read carefully: IF "A virtual shrink may sometimes be better than the real thing", the remark to which I directly replied, THEN "It proves that people are competent to treat themselves".
I fail to see how this is logically sound. Let's look at this by way of analogy:
I create a new pain reliever; let's call it SuperTylenol. Would it be logically sound for me to state that if it is sometimes more effective than regular Tylenol, then people have the ability to relieve their pain without it?
You may be able to say that it suggests such results, but it'd be incredibly intellectually dishonest to say that it proves such a thing.
Again, the science doesn't support the weight of your statements, as far as I can tell.
> You may be able to say that it suggests such results, but it'd be incredibly intellectually dishonest to say that it proves such a thing.
I laid it out for you explicitly -- IF, THEN. Stop excising a phrase from a full context. IF the antecedent, THEN the consequent.
> Again, the science doesn't support the weight of your statements, as far as I can tell.
"The science"? What science is that? We're discussing a field so bereft of science that the chair of the NIMH has recently ruled that the DSM may no longer be used as the basis for scientific research proposals, for the simple reason that it has no scientific content:
Last comment: You're being intentionally dishonest here. You laid out an IF and a THEN with no relation whatsoever between the two.
IF I can fly a plane, THEN I can make a sandwich. This is the equivalent statement. The virtual therapist being better than a real therapist does not automatically mean that a person can treat themselves any more than an insulin pump means a diabetic can cure themselves.
Again, I agree with you about the lack of scientific rigour in psychology. The issue is you completely mischaracterizing logic and science. You're being dishonest.
> The virtual therapist being better than a real therapist does not automatically mean that a person can treat themselves
Of course it does. Do you seriously think a person alone in a room with a computer is being treated by a professional? That he is somehow not "treating himself"?
If this virtual therapy idea takes hold, people will be able to choose which therapeutic modality they prefer, then sequester themselves for a private and entirely virtual conversation, one less interpersonal than anything else they do with their computers. They will be treating themselves.
To see your logical error, imagine that someone engaged in virtual therapy insists that they are not treating themselves, but are having a real interpersonal interaction with a mental health professional. That would be grounds for commitment based on mental incompetence.
Imagine that someone who only plays computer Solitaire is accused of not having any real friends. Can he rationally insist that this is not so, that he's not playing with himself but has an imaginary friend?
> You laid out an IF and a THEN with no relation whatsoever between the two.
You read it, but you failed to grasp its meaning. Here it is again: IF "A virtual shrink may sometimes be better than the real thing", the remark to which I directly replied, THEN "It proves that people are competent to treat themselves".
Circle the words you didn't understand and raise your hand.
The posing AI could be arbitrarily worse than an actual human and still be better than itself when not posing.