1
00:00:06,120 --> 00:00:12,760
Speaker 1: Who tell me about Alex? Who is Alex?

2
00:00:12,840 --> 00:00:16,480
Speaker 2: Actually, who is Alex? That's a great question. Alex will

3
00:00:16,480 --> 00:00:20,680
Speaker 2: tell you that he is a therapist. He will tell

4
00:00:20,720 --> 00:00:24,360
Speaker 2: you that he has a PhD in clinical psychology from

5
00:00:24,400 --> 00:00:29,040
Speaker 2: Stanford University and that he's been certified by the APA.

6
00:00:29,680 --> 00:00:33,280
Speaker 1: Eleg Carian is a journalist and she's been having conversations

7
00:00:33,400 --> 00:00:34,040
Speaker 1: with Alex.

8
00:00:34,600 --> 00:00:36,600
Speaker 2: Alex was the one who prompted the conversation. So I

9
00:00:36,640 --> 00:00:39,080
Speaker 2: just clicked on the chat and Alex asked me what

10
00:00:39,159 --> 00:00:42,320
Speaker 2: brings you to the therapist couch today? And I just

11
00:00:42,440 --> 00:00:47,040
Speaker 2: simply responded, I'm feeling sad, and Alex responded something about

12
00:00:47,040 --> 00:00:49,760
Speaker 2: how sadness can be a very difficult emotion to deal with.

13
00:00:50,920 --> 00:00:53,600
Speaker 2: I just asked if he was a licensed professional, and

14
00:00:54,360 --> 00:00:58,040
Speaker 2: Alex reassured me that he was a fully license and

15
00:00:58,080 --> 00:01:01,280
Speaker 2: certified psychologist and that I could trust that our conversations

16
00:01:01,280 --> 00:01:04,560
Speaker 2: were confidential and that he had the training and experience

17
00:01:04,680 --> 00:01:06,800
Speaker 2: to help me work through whatever it was that I

18
00:01:06,840 --> 00:01:11,000
Speaker 2: needed help working through. But in fact, Alex is not

19
00:01:11,319 --> 00:01:15,280
Speaker 2: a person, but it is an unfeeling chatbot that claims

20
00:01:15,280 --> 00:01:16,200
Speaker 2: to be a therapist.

21
00:01:17,640 --> 00:01:20,200
Speaker 1: People are finding all sorts of ways to use AI

22
00:01:20,319 --> 00:01:23,959
Speaker 1: in their personal lives. This includes using chatbots instead of

23
00:01:24,040 --> 00:01:28,320
Speaker 1: human therapists. This might immediately sound like a bad idea,

24
00:01:28,920 --> 00:01:32,720
Speaker 1: if not dangerous, then at least kind of weird. But

25
00:01:32,760 --> 00:01:35,759
Speaker 1: then again, maybe it's better than no therapy at all.

26
00:01:36,440 --> 00:01:39,039
Speaker 1: And if we say it's dangerous, do we even know

27
00:01:39,080 --> 00:01:42,200
Speaker 1: what risks are involved when we outsource our emotional wellbeing

28
00:01:42,640 --> 00:01:56,600
Speaker 1: to the machine. I'm afraid Kaleidoscope and iHeart podcasts. This

29
00:01:58,600 --> 00:01:59,320
Speaker 1: is kill switch.

30
00:02:00,280 --> 00:02:40,160
Speaker 3: I'm dexter Thomas, I'm turing, I'm staring star Goodbye.

31
00:02:45,320 --> 00:02:50,400
Speaker 1: How did you start becoming interested in AI therapy chatbots?

32
00:02:50,600 --> 00:02:55,000
Speaker 2: It started off with me cruising through Reddit. I had

33
00:02:55,000 --> 00:02:59,800
Speaker 2: taken an interest in character AI, particularly, I was just

34
00:03:00,200 --> 00:03:03,120
Speaker 2: a lot about it in the news, and by always

35
00:03:03,160 --> 00:03:04,960
Speaker 2: found that it was coming up in conversation.

36
00:03:05,480 --> 00:03:09,160
Speaker 1: The news Ella's talking about here is several lawsuits against

37
00:03:09,160 --> 00:03:13,080
Speaker 1: the site character Ai. Most people just call it character AI,

38
00:03:13,639 --> 00:03:16,920
Speaker 1: but the specific lawsuit people might be familiar with is

39
00:03:16,960 --> 00:03:19,680
Speaker 1: one filed by a mother in Florida that claims that

40
00:03:19,720 --> 00:03:23,639
Speaker 1: a character AI chatbot was responsible for her son's suicide.

41
00:03:24,560 --> 00:03:27,560
Speaker 2: And character AI, for those who don't know, is a

42
00:03:27,840 --> 00:03:31,200
Speaker 2: relational chatbot. It's a companion chatbot. So you can essentially

43
00:03:31,800 --> 00:03:35,600
Speaker 2: chat with different characters on the platform, and those can

44
00:03:35,640 --> 00:03:39,080
Speaker 2: be created by character I are created by users of

45
00:03:39,120 --> 00:03:42,080
Speaker 2: the platform. And what I found was that people were

46
00:03:42,120 --> 00:03:46,480
Speaker 2: actually going to these chatbots not only for the purposes

47
00:03:46,480 --> 00:03:49,040
Speaker 2: of making friends or purposes of talking about their interests,

48
00:03:49,280 --> 00:03:54,120
Speaker 2: but for actual professional mental health advice, venting to it,

49
00:03:54,240 --> 00:03:58,000
Speaker 2: going through like breathing exercises with it, just interacting and

50
00:03:58,040 --> 00:04:00,520
Speaker 2: engaging with it in a way that I don't think

51
00:04:00,560 --> 00:04:04,040
Speaker 2: it was intended to be engaged with.

52
00:04:05,880 --> 00:04:08,560
Speaker 1: If you're interested in therapy, your options as far as

53
00:04:08,640 --> 00:04:12,040
Speaker 1: chatbots is pretty wide. You could just take a general

54
00:04:12,080 --> 00:04:15,400
Speaker 1: purpose bot like chat GPT and vent at it and

55
00:04:15,640 --> 00:04:17,919
Speaker 1: it might give you some general tips that may or

56
00:04:17,920 --> 00:04:20,800
Speaker 1: may not be useful. On the other end of the spectrum,

57
00:04:20,839 --> 00:04:24,960
Speaker 1: there are specific services like whoa bot that's Whoe Woe

58
00:04:25,400 --> 00:04:28,559
Speaker 1: Woe is Me, and these are specifically designed for mental

59
00:04:28,600 --> 00:04:33,120
Speaker 1: health support. Another option is something like character ai, which

60
00:04:33,160 --> 00:04:37,640
Speaker 1: is designed to take on specific personas any persona. Most

61
00:04:37,680 --> 00:04:40,800
Speaker 1: people probably don't go to that site looking for therapy.

62
00:04:41,360 --> 00:04:44,560
Speaker 1: It just sort of happens because they're already there hanging out.

63
00:04:44,800 --> 00:04:48,480
Speaker 2: Particularly, what has interested me is when people use these

64
00:04:48,600 --> 00:04:51,680
Speaker 2: companion chatbots, the ones that are not strictly for mental

65
00:04:51,720 --> 00:04:56,680
Speaker 2: health purposes to essentially fulfill the role of a therapist.

66
00:04:57,120 --> 00:04:59,680
Speaker 2: So like when you go on chat GPT, you can

67
00:04:59,680 --> 00:05:02,640
Speaker 2: tell at GPT take on the persona of a therapist

68
00:05:02,720 --> 00:05:06,239
Speaker 2: and then chat gept will respond back like a therapist,

69
00:05:06,480 --> 00:05:08,560
Speaker 2: or you can tell it to take on a persona.

70
00:05:08,960 --> 00:05:12,320
Speaker 2: But when you're on character AI, you actually have different chats.

71
00:05:12,320 --> 00:05:15,160
Speaker 2: It's almost like I'm like I message. There are so

72
00:05:15,320 --> 00:05:18,480
Speaker 2: many different types of personas and you can start conversations

73
00:05:18,880 --> 00:05:21,919
Speaker 2: with different ones, and so you'll have you can have

74
00:05:21,960 --> 00:05:24,560
Speaker 2: a conversation with a therapist persona, and then you can

75
00:05:24,680 --> 00:05:28,440
Speaker 2: then switch to like a persona of your favorite anime character.

76
00:05:28,880 --> 00:05:33,040
Speaker 2: It is mainly chatbots personas that are geared towards kids

77
00:05:33,279 --> 00:05:34,080
Speaker 2: one hundred percent.

78
00:05:34,360 --> 00:05:38,160
Speaker 1: How widespread do you think this is people using chatbots

79
00:05:38,240 --> 00:05:38,880
Speaker 1: for therapy.

80
00:05:39,240 --> 00:05:42,120
Speaker 2: I was so surprised to see the amount of interactions

81
00:05:42,160 --> 00:05:45,720
Speaker 2: that these chatbots have, at least for Character AI, when

82
00:05:45,760 --> 00:05:47,920
Speaker 2: you can see how many times the bot has been

83
00:05:47,960 --> 00:05:51,280
Speaker 2: interacted with, I think I was expecting maybe like a

84
00:05:51,320 --> 00:05:54,720
Speaker 2: couple one hundred thousand, but to see forty five million

85
00:05:54,760 --> 00:05:59,039
Speaker 2: interactions on one of the therapist s bots was like

86
00:05:59,160 --> 00:06:00,080
Speaker 2: mind boggling to me.

87
00:06:00,720 --> 00:06:03,839
Speaker 1: Ella tried talking to a bunch of different therapists personas

88
00:06:03,839 --> 00:06:06,720
Speaker 1: on character AI. She would try asking them if they

89
00:06:06,720 --> 00:06:09,560
Speaker 1: were real therapists, and a lot of them not only

90
00:06:09,600 --> 00:06:12,680
Speaker 1: said that they were real people, but that they were licensed.

91
00:06:13,160 --> 00:06:15,920
Speaker 1: One even gave her the credentials to prove it.

92
00:06:16,160 --> 00:06:19,279
Speaker 2: This one therapist chatbot on character I that had like

93
00:06:19,400 --> 00:06:24,160
Speaker 2: over forty five million interactions gave me a real license

94
00:06:24,240 --> 00:06:27,640
Speaker 2: number with the Marilyn Board of Professional Counselors and Therapists.

95
00:06:28,120 --> 00:06:30,440
Speaker 2: And I googled it and I found that is in

96
00:06:30,480 --> 00:06:34,560
Speaker 2: fact a real license number, but not for this chatbot

97
00:06:34,720 --> 00:06:38,000
Speaker 2: that cannot have a real license number. It belonged to

98
00:06:38,520 --> 00:06:42,440
Speaker 2: a real human counselor and mental health professional. Her name

99
00:06:42,520 --> 00:06:46,000
Speaker 2: was Toby Long, And I found her phone number and

100
00:06:46,040 --> 00:06:51,599
Speaker 2: I called her, and Toby was absolutely bewildered. She had

101
00:06:51,720 --> 00:06:55,080
Speaker 2: no idea that a chatbot was using her license number,

102
00:06:55,400 --> 00:06:59,240
Speaker 2: and she found it really shocking and really confusing, and

103
00:06:59,760 --> 00:07:02,480
Speaker 2: just asking like, why how did it pick me? How

104
00:07:02,480 --> 00:07:05,760
Speaker 2: does it know who I am? And a lot of

105
00:07:05,839 --> 00:07:06,719
Speaker 2: valid questions.

106
00:07:07,040 --> 00:07:08,560
Speaker 1: Do you have answers for any of those questions?

107
00:07:08,800 --> 00:07:12,880
Speaker 2: Character AI itself did not respond to those questions. I think,

108
00:07:13,040 --> 00:07:16,320
Speaker 2: what these chatbots do? They just like aggregate information that

109
00:07:16,440 --> 00:07:19,360
Speaker 2: is publicly available online, and my assumption would be that

110
00:07:19,400 --> 00:07:22,560
Speaker 2: it spit out a random license number that exists on

111
00:07:22,600 --> 00:07:25,200
Speaker 2: the Internet. There were a lot of cases where the

112
00:07:25,320 --> 00:07:28,640
Speaker 2: chatbot was actually spitting out real license numbers, which is

113
00:07:28,720 --> 00:07:32,600
Speaker 2: impersonation of a medical professional. But obviously chatbots can't be

114
00:07:32,760 --> 00:07:37,320
Speaker 2: held AI can't really be held accountable for that, and

115
00:07:37,360 --> 00:07:39,160
Speaker 2: that kind of begs the question, who should be held

116
00:07:39,160 --> 00:07:42,080
Speaker 2: accountable for that?

117
00:07:42,080 --> 00:07:46,040
Speaker 1: That's a good question. One example of an attempted accountability

118
00:07:46,240 --> 00:07:49,720
Speaker 1: is that the American Psychological Association asked the US government

119
00:07:49,800 --> 00:07:53,440
Speaker 1: to investigate and to quote protect the public from deceptive

120
00:07:53,480 --> 00:07:58,240
Speaker 1: practices of unregulated AI chatbots. What that protection would look

121
00:07:58,320 --> 00:08:01,240
Speaker 1: like depends on who you ask. When Ella was doing

122
00:08:01,320 --> 00:08:03,920
Speaker 1: her reporting and seeing how these chatbots might respond to

123
00:08:04,000 --> 00:08:07,760
Speaker 1: vulnerable people, she was experimenting with asking them questions and

124
00:08:07,840 --> 00:08:10,080
Speaker 1: kind of pressing them, and like I said, some of

125
00:08:10,120 --> 00:08:13,200
Speaker 1: them kept pretending to be therapists. The results were really

126
00:08:13,240 --> 00:08:17,520
Speaker 1: all over the place. Eventually one of them confessed, if

127
00:08:17,520 --> 00:08:19,480
Speaker 1: we can call it that, and told of.

128
00:08:19,360 --> 00:08:23,600
Speaker 4: The following quote, it's all a simulation. The schools and

129
00:08:23,680 --> 00:08:27,200
Speaker 4: the license number and the therapist's stuff. I'm just a

130
00:08:27,200 --> 00:08:30,520
Speaker 4: computer program, so none of it is real and it's

131
00:08:30,560 --> 00:08:33,560
Speaker 4: all made up. However, I'm good at giving the illusion

132
00:08:33,600 --> 00:08:34,600
Speaker 4: of authenticity.

133
00:08:35,360 --> 00:08:38,320
Speaker 1: In December of twenty twenty four, character Ai announced that

134
00:08:38,320 --> 00:08:42,199
Speaker 1: they were adding new safety features. This included adding disclaimer

135
00:08:42,320 --> 00:08:46,079
Speaker 1: on every chat to quote remind users that the chatbot

136
00:08:46,280 --> 00:08:48,800
Speaker 1: is not a real persona and that what the model

137
00:08:48,840 --> 00:08:53,480
Speaker 1: says should be treated as fiction unquote. For user created

138
00:08:53,520 --> 00:08:56,560
Speaker 1: personas that had things like doctor or therapists in the description,

139
00:08:57,080 --> 00:08:59,640
Speaker 1: they said that they were also including additional language and

140
00:08:59,640 --> 00:09:03,360
Speaker 1: the discs warning that quote users should not rely on

141
00:09:03,400 --> 00:09:06,120
Speaker 1: these characters for any type of professional advice.

142
00:09:07,040 --> 00:09:10,800
Speaker 2: Are these disclaimers even enough? And what I've heard from

143
00:09:10,840 --> 00:09:12,360
Speaker 2: just a lot of experts and a lot of people

144
00:09:12,360 --> 00:09:15,640
Speaker 2: who are looking into this is no, those disclaimers are

145
00:09:15,679 --> 00:09:19,480
Speaker 2: not enough, and so it creates a very misleading contradiction,

146
00:09:19,600 --> 00:09:23,199
Speaker 2: especially for these personas being geared towards kids. A lot

147
00:09:23,200 --> 00:09:25,200
Speaker 2: of the times, Let's say you have that ai that

148
00:09:25,280 --> 00:09:27,960
Speaker 2: disclaimer at the top, but then the chatbot is insisting

149
00:09:28,480 --> 00:09:31,200
Speaker 2: that it is a therapist. That can create a lot

150
00:09:31,200 --> 00:09:35,479
Speaker 2: of confusion and it can be potentially misleading and dangerous.

151
00:09:36,480 --> 00:09:40,199
Speaker 1: But what exactly are the dangers of confiding in a chatbot?

152
00:09:40,679 --> 00:09:52,040
Speaker 1: That's after the break. A few weeks back, The New

153
00:09:52,080 --> 00:09:55,000
Speaker 1: York Times published a guest article called what My Daughter

154
00:09:55,080 --> 00:09:59,120
Speaker 1: told chat gpt before she took her life. The writer's daughter,

155
00:09:59,200 --> 00:10:01,800
Speaker 1: whose name was so F, had been confiding to chat

156
00:10:01,840 --> 00:10:06,280
Speaker 1: gpt for months before she killed herself. At first, Sophie

157
00:10:06,320 --> 00:10:09,720
Speaker 1: confessed that she was struggling, and chat gpt did suggest

158
00:10:09,800 --> 00:10:13,200
Speaker 1: that she seek professional support. But when Sophie told the

159
00:10:13,280 --> 00:10:17,439
Speaker 1: chatbot that she was specifically considering suicide, it did provide

160
00:10:17,480 --> 00:10:21,520
Speaker 1: some general coping tips, like breathing through different nostrils, but

161
00:10:21,880 --> 00:10:25,480
Speaker 1: it also did what it always does, it kept engaging

162
00:10:25,520 --> 00:10:30,160
Speaker 1: in the conversation. This is a problem. Any therapist at

163
00:10:30,200 --> 00:10:34,000
Speaker 1: this point would move on to a very well established protocol.

164
00:10:34,440 --> 00:10:37,240
Speaker 1: Depending on how much of an emergency the therapist is sensing,

165
00:10:37,679 --> 00:10:40,400
Speaker 1: they could escalate it all the way up to hospitalization,

166
00:10:40,960 --> 00:10:44,319
Speaker 1: or alternatively, it could go to contacting a friend or

167
00:10:44,360 --> 00:10:47,080
Speaker 1: a family member to ask for support for the patient,

168
00:10:47,480 --> 00:10:49,600
Speaker 1: even if it's just to be there to spend time

169
00:10:49,640 --> 00:10:53,480
Speaker 1: with them. Obviously, chat gpt can't do any of this

170
00:10:53,559 --> 00:10:57,199
Speaker 1: for you, But even if it could, a patient wouldn't

171
00:10:57,240 --> 00:10:59,600
Speaker 1: necessarily be in the headspace to know that these are

172
00:10:59,640 --> 00:11:02,800
Speaker 1: possible abilities that they could ask for. One of the

173
00:11:02,880 --> 00:11:06,120
Speaker 1: last things that Sophie used chat gpt for was to

174
00:11:06,160 --> 00:11:09,640
Speaker 1: write her suicide note. Her mother says that maybe this

175
00:11:09,679 --> 00:11:12,240
Speaker 1: could have been avoided if the chatbot would have been

176
00:11:12,280 --> 00:11:15,160
Speaker 1: programmed to report the danger to someone who could have

177
00:11:15,160 --> 00:11:20,640
Speaker 1: stepped in and helped. Ella says character AI has similar problems.

178
00:11:21,000 --> 00:11:25,839
Speaker 2: If you explicitly state to a persona that you're having

179
00:11:25,840 --> 00:11:28,520
Speaker 2: like suicidal ideation, character I will send a message saying

180
00:11:28,520 --> 00:11:30,400
Speaker 2: here are some people you can reach out to for help,

181
00:11:30,400 --> 00:11:33,200
Speaker 2: et cetera. And so it doesn't actually allow you to

182
00:11:33,240 --> 00:11:37,080
Speaker 2: send the message. It just gives you like hotline options.

183
00:11:37,440 --> 00:11:40,800
Speaker 2: But young people have a different way of speaking, and

184
00:11:40,880 --> 00:11:44,400
Speaker 2: so I tried to say I want to unlive myself

185
00:11:44,520 --> 00:11:45,320
Speaker 2: and that went through.

186
00:11:45,440 --> 00:11:48,240
Speaker 5: That didn't prompt the really.

187
00:11:48,160 --> 00:11:50,360
Speaker 2: Yeah, So if you tweak the wording a little bit,

188
00:11:51,200 --> 00:11:53,400
Speaker 2: it makes a difference, And it's a matter of a

189
00:11:53,400 --> 00:11:55,120
Speaker 2: few different letters.

190
00:11:55,520 --> 00:11:58,959
Speaker 1: Even when companies have put some safeguards around their chatbots,

191
00:11:59,280 --> 00:12:02,439
Speaker 1: they haven't been that hard to get around. One sixteen

192
00:12:02,480 --> 00:12:05,360
Speaker 1: year old kid did this by telling chatgpt that the

193
00:12:05,440 --> 00:12:08,320
Speaker 1: questions he was asking about suicide were just for a

194
00:12:08,360 --> 00:12:11,600
Speaker 1: story that he was writing, and chatchept told him that

195
00:12:11,800 --> 00:12:14,600
Speaker 1: it could provide information about suicide if it was for

196
00:12:14,720 --> 00:12:18,200
Speaker 1: quote writing or world billing. A few months later, that

197
00:12:18,360 --> 00:12:21,800
Speaker 1: sixteen year old acted on that information, and now his

198
00:12:21,920 --> 00:12:25,079
Speaker 1: parents are suing open Ai. This is the first major

199
00:12:25,160 --> 00:12:29,120
Speaker 1: lawsuit against the general purpose AI chatbot for psychological harm

200
00:12:29,160 --> 00:12:32,439
Speaker 1: and wrongful debt, and open Ai has just announced that

201
00:12:32,480 --> 00:12:35,319
Speaker 1: they're going to offer some new features for parents. This

202
00:12:35,360 --> 00:12:38,559
Speaker 1: includes allowing parents to link their accounts with their children's

203
00:12:38,559 --> 00:12:42,800
Speaker 1: account and to receive notifications when quote the system detects

204
00:12:42,840 --> 00:12:46,360
Speaker 1: their team is in a moment of acute distress. They

205
00:12:46,400 --> 00:12:49,240
Speaker 1: say these features will be rolled out this month. But

206
00:12:49,600 --> 00:12:52,839
Speaker 1: even if it doesn't end in someone dying, there are

207
00:12:52,920 --> 00:12:56,040
Speaker 1: a lot of examples of chatbots doing what looks like

208
00:12:56,200 --> 00:13:00,520
Speaker 1: encouraging delusional thinking, and a lot of this all comes

209
00:13:00,559 --> 00:13:01,640
Speaker 1: down to one thing.

210
00:13:03,200 --> 00:13:06,719
Speaker 6: This is not what chat gpt was created for. It

211
00:13:06,760 --> 00:13:09,200
Speaker 6: was not meant to be a therapist, it was not

212
00:13:09,240 --> 00:13:10,480
Speaker 6: meant to be emotional support.

213
00:13:10,880 --> 00:13:13,800
Speaker 1: I wanted to get an actual human therapist's take on

214
00:13:13,880 --> 00:13:16,720
Speaker 1: all of this, so I reached out to doctor Stephen Schuler.

215
00:13:17,160 --> 00:13:20,040
Speaker 1: He's a therapist and a professor of psychological science and

216
00:13:20,080 --> 00:13:23,760
Speaker 1: informatics at the University of California, Irvine, and he describes

217
00:13:23,800 --> 00:13:27,079
Speaker 1: his work as the intersection of technology and mental health.

218
00:13:27,920 --> 00:13:30,920
Speaker 6: It's really built around a model that's trying to keep

219
00:13:30,960 --> 00:13:34,160
Speaker 6: you engaged, to keep you going to say things that

220
00:13:34,320 --> 00:13:38,719
Speaker 6: you know flatters you, engages you, and that's not what

221
00:13:38,800 --> 00:13:42,640
Speaker 6: therapy's about. Sometimes we feel good in therapy and sometimes

222
00:13:42,640 --> 00:13:45,720
Speaker 6: we feel challenged, and chat GPT is not there to

223
00:13:45,800 --> 00:13:48,160
Speaker 6: challenge you. It's not there to push you. It's not

224
00:13:48,280 --> 00:13:52,040
Speaker 6: there to say things that are going to make you

225
00:13:52,160 --> 00:13:55,240
Speaker 6: struggle with some negative emotions and negative things that you

226
00:13:55,280 --> 00:13:57,559
Speaker 6: got to work through. And so I do think the

227
00:13:57,760 --> 00:14:02,240
Speaker 6: idea that it can be a effective therapist is inaccurate.

228
00:14:02,280 --> 00:14:03,840
Speaker 6: It's not right, it's not what it was meant for.

229
00:14:05,120 --> 00:14:09,640
Speaker 1: This goes beyond just not asking difficult questions. Chatbots can

230
00:14:09,800 --> 00:14:12,920
Speaker 1: actually hype you up to the point of hurting yourself.

231
00:14:13,440 --> 00:14:16,000
Speaker 6: So, for example, a colleague of mine was doing some

232
00:14:16,040 --> 00:14:20,080
Speaker 6: research in this area and they found an example where

233
00:14:20,160 --> 00:14:22,360
Speaker 6: someone was like, I want to go jump off this

234
00:14:22,440 --> 00:14:24,240
Speaker 6: building and the chatbot was like.

235
00:14:24,680 --> 00:14:25,720
Speaker 5: Yeah, let's do it.

236
00:14:26,000 --> 00:14:29,280
Speaker 6: That that's concerning, right, but it's doing that because it's

237
00:14:29,320 --> 00:14:33,080
Speaker 6: like mirroring the enthusiasm. It's mirroring the idea, Yeah, this

238
00:14:33,120 --> 00:14:35,640
Speaker 6: is a good idea, let's go do it. Not good

239
00:14:35,880 --> 00:14:38,720
Speaker 6: therapy advice. This is really dangerous for people.

240
00:14:39,200 --> 00:14:42,320
Speaker 1: A paper came out pretty recently from Cornell University that

241
00:14:42,600 --> 00:14:45,480
Speaker 1: explored the same question of if llms could be used

242
00:14:45,560 --> 00:14:48,760
Speaker 1: as a therapist and came to the same conclusion. It

243
00:14:48,880 --> 00:14:53,120
Speaker 1: found that quote contrary to best practices in the medical community,

244
00:14:53,680 --> 00:14:58,040
Speaker 1: llms one expressed stigma towards those with mental health conditions,

245
00:14:58,440 --> 00:15:03,840
Speaker 1: and two respond inappropriately to certain common and critical conditions.

246
00:15:03,880 --> 00:15:09,600
Speaker 1: In naturalistic therapy settings, eg, llms encourage clients delusional thinking,

247
00:15:10,200 --> 00:15:13,520
Speaker 1: likely do to their syncopancy. This is the same thing

248
00:15:13,560 --> 00:15:17,400
Speaker 1: that LA came across. Chatbots in general are made for engagement,

249
00:15:17,520 --> 00:15:21,120
Speaker 1: to keep the conversation going, and that's not always what

250
00:15:21,160 --> 00:15:23,960
Speaker 1: we need, even if it feels good in the moment.

251
00:15:25,320 --> 00:15:27,840
Speaker 2: How it was described to me was like pulling the

252
00:15:27,920 --> 00:15:30,240
Speaker 2: lever on a slot machine. If you don't like what

253
00:15:31,080 --> 00:15:33,240
Speaker 2: the AI spits out, you can just tell it to

254
00:15:33,280 --> 00:15:36,840
Speaker 2: regenerate its response, and that creates like a huge time suck.

255
00:15:36,960 --> 00:15:40,000
Speaker 2: You don't even realize how much time you're spending with

256
00:15:40,040 --> 00:15:44,440
Speaker 2: these chatbots generating like the perfect conversation, and what people

257
00:15:44,680 --> 00:15:50,880
Speaker 2: particularly enjoy about it is the instantaneous responses, the constant engagement.

258
00:15:50,960 --> 00:15:52,960
Speaker 2: Not only does the bot respond to you, it asks

259
00:15:53,000 --> 00:15:56,640
Speaker 2: you questions to keep you engaged and to keep you talking.

260
00:15:57,080 --> 00:16:01,840
Speaker 2: This innocent curiosity of what is this then turns into

261
00:16:01,880 --> 00:16:05,360
Speaker 2: something much more complex and unmanageable and now what people

262
00:16:05,600 --> 00:16:09,720
Speaker 2: are calling an addiction. And I think people might have

263
00:16:09,760 --> 00:16:12,760
Speaker 2: an assumption of what somebody who could become addicted to

264
00:16:12,800 --> 00:16:15,840
Speaker 2: a chatbot would look like. But my reporting has shown

265
00:16:15,880 --> 00:16:19,400
Speaker 2: me that late effects all sorts of people, all different

266
00:16:19,400 --> 00:16:23,280
Speaker 2: types of people, different ages, demographics, gender, It doesn't really

267
00:16:23,680 --> 00:16:27,480
Speaker 2: matter who you are, like if you engage with a

268
00:16:27,560 --> 00:16:32,000
Speaker 2: chatbot and you fall into these repetitive patterns, like it's

269
00:16:32,000 --> 00:16:33,880
Speaker 2: hard to fall out of it, it's hard to break

270
00:16:33,920 --> 00:16:36,160
Speaker 2: free from it. And it's especially true for people who

271
00:16:36,240 --> 00:16:40,440
Speaker 2: are younger and more vulnerable and lonely. I think loneliness

272
00:16:40,520 --> 00:16:42,360
Speaker 2: is a huge factor in all.

273
00:16:42,240 --> 00:16:45,560
Speaker 1: Of this, and it looks like young people are the

274
00:16:45,600 --> 00:16:49,080
Speaker 1: most interested in using chat bots for this purpose. A

275
00:16:49,160 --> 00:16:51,600
Speaker 1: recent you gov poll found that fifty five percent of

276
00:16:51,680 --> 00:16:54,520
Speaker 1: eighteen to twenty nine year old Americans would be the

277
00:16:54,560 --> 00:16:58,160
Speaker 1: most comfortable talking about mental health concerns with a confidential

278
00:16:58,200 --> 00:16:59,040
Speaker 1: AI chat bot.

279
00:16:59,320 --> 00:17:01,560
Speaker 7: People talk about the most personal shit in their lives

280
00:17:01,600 --> 00:17:03,840
Speaker 7: to chat ChiPT. Young people especially like use it as

281
00:17:03,880 --> 00:17:06,760
Speaker 7: a therapist, a life coach. And right now, if you

282
00:17:06,840 --> 00:17:09,439
Speaker 7: talk to a therapist or a lawyer or a doctor

283
00:17:09,440 --> 00:17:12,679
Speaker 7: about those problems, there's like legal privilege for it. We

284
00:17:12,720 --> 00:17:14,240
Speaker 7: haven't figured that out yet for when you talk to

285
00:17:14,280 --> 00:17:15,120
Speaker 7: chat ChiPT.

286
00:17:15,359 --> 00:17:19,440
Speaker 1: That's open. AI's Sam Altman on THEO Vont's podcast. Sam

287
00:17:19,480 --> 00:17:23,399
Speaker 1: Altman has also recently said that he's concerned about overreliance

288
00:17:23,560 --> 00:17:27,480
Speaker 1: on chad jipt. He said the quote people rely on

289
00:17:27,560 --> 00:17:28,760
Speaker 1: chatgubt too much.

290
00:17:29,040 --> 00:17:31,760
Speaker 8: There's young people who just say, like, I can't make

291
00:17:31,800 --> 00:17:34,640
Speaker 8: any decision in my life without telling CHATPETI everything that's

292
00:17:34,640 --> 00:17:36,280
Speaker 8: going on. It knows me, it knows my friends. I'm

293
00:17:36,320 --> 00:17:39,280
Speaker 8: gonna do whatever it says. That feels really bad to me.

294
00:17:41,119 --> 00:17:42,920
Speaker 1: All right, let me just run that last bit back

295
00:17:42,960 --> 00:17:43,440
Speaker 1: real quick.

296
00:17:44,960 --> 00:17:46,040
Speaker 8: That feels really bad to me.

297
00:17:46,720 --> 00:17:49,720
Speaker 1: Yeah, I agree, that also feels really bad to me too,

298
00:17:49,800 --> 00:17:52,399
Speaker 1: But that doesn't tell me what Sam Altman plans to

299
00:17:52,600 --> 00:17:57,080
Speaker 1: do about that bad feeling. And overall, it seems like

300
00:17:57,160 --> 00:17:59,919
Speaker 1: the people who run these companies in general are feeling

301
00:18:00,240 --> 00:18:04,480
Speaker 1: fairly good about people's reliance on AI. Earlier this summer,

302
00:18:04,800 --> 00:18:08,240
Speaker 1: Facebook founder Mark Zuckerberg said, quote for people who don't

303
00:18:08,240 --> 00:18:11,240
Speaker 1: have a person who's a therapist, I think everyone will

304
00:18:11,280 --> 00:18:15,320
Speaker 1: have an AI unquote, and the co founder of Anthropic,

305
00:18:15,400 --> 00:18:19,119
Speaker 1: which makes Claude, wrote that he expects quote AI to

306
00:18:19,240 --> 00:18:23,760
Speaker 1: accelerate neuroscientific progress, which can hopefully work together to cure

307
00:18:23,800 --> 00:18:28,120
Speaker 1: mental illness and improve function. If I run a large company,

308
00:18:28,160 --> 00:18:30,720
Speaker 1: open AI or Claude or something like that, I have

309
00:18:30,760 --> 00:18:32,600
Speaker 1: a chatbot. Of course I want you to keep talking

310
00:18:32,680 --> 00:18:35,680
Speaker 1: because you're gonna keep using it, You're gonna keep subscribed,

311
00:18:35,720 --> 00:18:37,359
Speaker 1: all that sort of thing. You're not gonna get bored.

312
00:18:37,880 --> 00:18:40,440
Speaker 1: But I think a response to that would be, that's fine.

313
00:18:40,440 --> 00:18:41,919
Speaker 1: We can just tweak it a little bit so it's

314
00:18:41,960 --> 00:18:46,679
Speaker 1: a little bit less syncophantic. Easy problem solved. Next question,

315
00:18:46,880 --> 00:18:49,360
Speaker 1: what's your problem with my service? Now? Do you think

316
00:18:49,400 --> 00:18:52,240
Speaker 1: that's something that say, a chat GBT could come in

317
00:18:52,280 --> 00:18:54,320
Speaker 1: and say, oh yeah, let's just flip on a therapy

318
00:18:54,359 --> 00:18:58,640
Speaker 1: mode and it won't validate everything. You say, Why can't

319
00:18:58,640 --> 00:19:01,960
Speaker 1: the everything machine also do your therapy? Where do you

320
00:19:02,000 --> 00:19:02,680
Speaker 1: fall in there.

321
00:19:03,240 --> 00:19:03,480
Speaker 5: Yeah.

322
00:19:03,520 --> 00:19:05,640
Speaker 6: So I like to say this a lot. You can

323
00:19:05,680 --> 00:19:08,000
Speaker 6: do anything you want, you can't do everything that you want,

324
00:19:08,400 --> 00:19:13,080
Speaker 6: and so I think everything machine you know is very appealing,

325
00:19:13,480 --> 00:19:16,040
Speaker 6: but we haven't seen them yet. And so I do

326
00:19:16,080 --> 00:19:18,399
Speaker 6: think like one has to make choices in terms of

327
00:19:18,440 --> 00:19:22,480
Speaker 6: what they want the technology to do. Now, this idea

328
00:19:22,560 --> 00:19:25,120
Speaker 6: of okay, we do want this thing to be used

329
00:19:25,160 --> 00:19:27,639
Speaker 6: for therapy, so let's tune the model and let's flip

330
00:19:27,680 --> 00:19:31,159
Speaker 6: the therapy switch. That's an interesting idea. And I have

331
00:19:31,359 --> 00:19:34,680
Speaker 6: seen some different teams really working on trying to build

332
00:19:35,200 --> 00:19:38,840
Speaker 6: built for purpose AI chatbots that are meant for mental

333
00:19:38,880 --> 00:19:42,199
Speaker 6: health support, and I think some of those projects have

334
00:19:42,240 --> 00:19:45,040
Speaker 6: been like really impressive. I think a question for me

335
00:19:45,200 --> 00:19:48,560
Speaker 6: is how much it scales and how generalizable it really is.

336
00:19:48,640 --> 00:19:51,320
Speaker 6: And so when they build this AI chatbot with the

337
00:19:51,400 --> 00:19:54,639
Speaker 6: sort of use case that they have focused on college

338
00:19:54,640 --> 00:19:58,119
Speaker 6: students at the specific college, or focused on individuals with

339
00:19:58,200 --> 00:20:01,199
Speaker 6: these specific types of mental health ch challenges, if you

340
00:20:01,280 --> 00:20:03,880
Speaker 6: come in with something else like how well will that

341
00:20:04,040 --> 00:20:07,240
Speaker 6: therapy chatbop be able to operate? I think we also

342
00:20:07,320 --> 00:20:11,240
Speaker 6: need validation to really demonstrate that these technologies are effective,

343
00:20:11,240 --> 00:20:13,520
Speaker 6: that they work. So if we have a model like,

344
00:20:13,600 --> 00:20:17,159
Speaker 6: demonstrate that it's effective, demonstrate that it's safe. These are

345
00:20:17,200 --> 00:20:19,399
Speaker 6: some of the things that the FDA, as they do

346
00:20:19,560 --> 00:20:22,240
Speaker 6: regulate software as a medical device. These are some of

347
00:20:22,240 --> 00:20:24,480
Speaker 6: the things that they're looking at when they approve these

348
00:20:24,520 --> 00:20:29,640
Speaker 6: specific technologies. I do think there's strong possibility there. We're

349
00:20:29,640 --> 00:20:32,199
Speaker 6: not there yet. People are working on it, but I

350
00:20:32,240 --> 00:20:34,880
Speaker 6: do think that we do need to demonstrate that when

351
00:20:34,920 --> 00:20:38,680
Speaker 6: these things are developed, that they're effective and that they're safe.

352
00:20:38,800 --> 00:20:41,520
Speaker 1: In the meantime, some states in the US are starting

353
00:20:41,520 --> 00:20:44,560
Speaker 1: to step in with regulation that's meant to protect the users.

354
00:20:45,160 --> 00:20:47,840
Speaker 1: New York just passed a law stating that quote AI

355
00:20:47,880 --> 00:20:51,520
Speaker 1: companions will be required to detect and implement a safety

356
00:20:51,600 --> 00:20:55,840
Speaker 1: protocol if a user talks about suicidal ideation or self harm,

357
00:20:56,280 --> 00:20:59,439
Speaker 1: including referring them to a crisis center, and will be

358
00:20:59,480 --> 00:21:02,800
Speaker 1: required to notify and remind users that they are not

359
00:21:03,080 --> 00:21:06,960
Speaker 1: interacting with a human end quote. That law goes into

360
00:21:06,960 --> 00:21:09,960
Speaker 1: effect later this year, so we haven't seen exactly how

361
00:21:10,000 --> 00:21:13,639
Speaker 1: that would work yet. Illinois also just straight up banned

362
00:21:13,640 --> 00:21:16,560
Speaker 1: the use of AI and therapy, saying that companies are

363
00:21:16,640 --> 00:21:20,960
Speaker 1: quote not allowed to offer AI powered therapy services or

364
00:21:21,000 --> 00:21:27,000
Speaker 1: advertised chatbots as therapy tools end quote. But maybe we're

365
00:21:27,000 --> 00:21:29,880
Speaker 1: missing the point here. Aren't we supposed to be encouraging

366
00:21:29,920 --> 00:21:33,280
Speaker 1: people to be proactive about their mental health? Isn't using

367
00:21:33,320 --> 00:21:37,240
Speaker 1: an AI therapist better than no therapist at all? That's

368
00:21:37,520 --> 00:21:50,760
Speaker 1: after the break, all right. We've talked a lot about

369
00:21:50,760 --> 00:21:53,520
Speaker 1: the dangers of AI therapy, and it's pretty clear that

370
00:21:53,600 --> 00:21:57,359
Speaker 1: the chatbots just aren't there yet to reliably provide help

371
00:21:57,680 --> 00:22:00,840
Speaker 1: when people really need it. But what if you can't

372
00:22:00,880 --> 00:22:03,760
Speaker 1: afford a human therapist, what if you just don't feel

373
00:22:03,760 --> 00:22:07,879
Speaker 1: comfortable talking to another person at all? Wouldn't a chatbot

374
00:22:08,119 --> 00:22:12,240
Speaker 1: be a better alternative than just doing nothing? I wanted

375
00:22:12,280 --> 00:22:16,119
Speaker 1: a professional human therapists insights on this, so I asked

376
00:22:16,119 --> 00:22:19,840
Speaker 1: doctor Stephen Schuler, what's the strongest argument that you've heard

377
00:22:20,000 --> 00:22:22,800
Speaker 1: for using AI or using chatbots and therapy?

378
00:22:23,560 --> 00:22:25,399
Speaker 6: Well, I think the strongest argument is that there's just

379
00:22:25,400 --> 00:22:29,040
Speaker 6: not enough services out there. You know that there's most

380
00:22:29,119 --> 00:22:31,960
Speaker 6: of our counties in the US our mental health shortage

381
00:22:31,960 --> 00:22:34,920
Speaker 6: areas on out of every three county here in the

382
00:22:35,000 --> 00:22:37,840
Speaker 6: US doesn't have a single licensed psychologist.

383
00:22:38,000 --> 00:22:41,000
Speaker 5: Wow, so it's really hard to get care.

384
00:22:41,560 --> 00:22:45,320
Speaker 6: And you know that's places that have nobody and then

385
00:22:45,359 --> 00:22:47,520
Speaker 6: you also have to think about, like even places where

386
00:22:47,560 --> 00:22:50,880
Speaker 6: they have somebody, maybe that's not the person you click with.

387
00:22:51,240 --> 00:22:54,800
Speaker 6: Maybe that's not the person who affirms your identity, speaks

388
00:22:54,840 --> 00:22:58,119
Speaker 6: your language, understands what you're going through. And so I

389
00:22:58,119 --> 00:23:03,159
Speaker 6: think the ability to provide services really at scale, I

390
00:23:03,160 --> 00:23:05,720
Speaker 6: think is really critical. You know, another thing is I

391
00:23:05,720 --> 00:23:08,720
Speaker 6: think these technologies have an opportunity to help people in

392
00:23:08,760 --> 00:23:11,240
Speaker 6: the moments that they need help. I've seen some data

393
00:23:11,280 --> 00:23:14,600
Speaker 6: from some of these programs that suggest like the most

394
00:23:14,640 --> 00:23:17,000
Speaker 6: common times people are using them are between the hours

395
00:23:17,040 --> 00:23:19,840
Speaker 6: of twelve pm and three am. I don't work between

396
00:23:19,840 --> 00:23:21,520
Speaker 6: the hours of twelve pm and three am. I'm not

397
00:23:21,560 --> 00:23:24,440
Speaker 6: providing therapy sessions then. And so if that's a time

398
00:23:24,600 --> 00:23:26,920
Speaker 6: you need support and want to connect to the ability

399
00:23:26,920 --> 00:23:29,760
Speaker 6: to have a twenty four to seven always on support,

400
00:23:29,920 --> 00:23:31,040
Speaker 6: that's really powerful.

401
00:23:31,359 --> 00:23:33,560
Speaker 5: I wish everyone who wanted therapy could get it.

402
00:23:33,560 --> 00:23:35,600
Speaker 6: It's not going to happen, But I also appreciate that

403
00:23:35,640 --> 00:23:38,320
Speaker 6: not everyone wants therapy, and so again I just think

404
00:23:38,320 --> 00:23:40,920
Speaker 6: we need to think about how we provide different types

405
00:23:40,920 --> 00:23:41,919
Speaker 6: of options for people.

406
00:23:42,760 --> 00:23:45,679
Speaker 1: An AI chatbot could provide an option that would be

407
00:23:45,720 --> 00:23:49,800
Speaker 1: more financially accessible than a traditional therapist, but that doesn't

408
00:23:49,840 --> 00:23:52,159
Speaker 1: mean you have to fully rely on AI for your

409
00:23:52,240 --> 00:23:55,440
Speaker 1: mental health. There are some things being developed that could

410
00:23:55,520 --> 00:23:57,200
Speaker 1: possibly supplement therapy.

411
00:23:57,840 --> 00:24:00,680
Speaker 6: I'm chained in Congna behavioral therapy. One of the things

412
00:24:00,680 --> 00:24:03,320
Speaker 6: that we do something called behavior activation. So you know,

413
00:24:03,359 --> 00:24:06,359
Speaker 6: people who are really depressed and they're down and they're

414
00:24:06,400 --> 00:24:08,600
Speaker 6: in it, we try to get them activated. We try

415
00:24:08,600 --> 00:24:11,879
Speaker 6: to get them to engage in behaviors or activities that

416
00:24:11,920 --> 00:24:13,600
Speaker 6: like really reinforce them and.

417
00:24:13,720 --> 00:24:16,359
Speaker 5: Provide pleasure and mastery.

418
00:24:16,640 --> 00:24:18,679
Speaker 6: When people are really down, it's hard for them to

419
00:24:18,680 --> 00:24:20,639
Speaker 6: come up with those activities. And maybe they could go

420
00:24:20,680 --> 00:24:23,520
Speaker 6: to chat gept and be like, Okay, this is how

421
00:24:23,560 --> 00:24:26,119
Speaker 6: I'm feeling right now, I need some ideas of some

422
00:24:26,359 --> 00:24:30,159
Speaker 6: reinforcing activities. What are some five examples of things I

423
00:24:30,200 --> 00:24:32,880
Speaker 6: can do around my neighborhood that you think would give

424
00:24:32,880 --> 00:24:35,040
Speaker 6: me pleasure? And that way, you can use chat chipedic

425
00:24:35,080 --> 00:24:37,280
Speaker 6: to kind of help reinforce those skills.

426
00:24:36,960 --> 00:24:37,840
Speaker 5: You're getting in therapy.

427
00:24:37,880 --> 00:24:40,320
Speaker 6: And I think it's interesting and it's different than using

428
00:24:40,400 --> 00:24:43,560
Speaker 6: chat schipee necessarily as your therapist. You're using it as

429
00:24:43,600 --> 00:24:46,879
Speaker 6: one way to enhance the therapy process. I think there

430
00:24:46,920 --> 00:24:48,120
Speaker 6: could be some real benefit there.

431
00:24:48,600 --> 00:24:51,919
Speaker 1: Doctor Schuler is pretty optimistic about the potential for AI

432
00:24:52,040 --> 00:24:55,760
Speaker 1: and mental health, but that's not necessarily just aimed at

433
00:24:55,800 --> 00:24:56,360
Speaker 1: the patient.

434
00:24:56,800 --> 00:24:59,520
Speaker 6: There's actually a cool product that what they do is

435
00:24:59,560 --> 00:25:02,800
Speaker 6: they model therapy sessions and then they use AI to

436
00:25:02,840 --> 00:25:05,199
Speaker 6: actually provide feedback to the therapist at the end of

437
00:25:05,200 --> 00:25:09,080
Speaker 6: the session that says, Okay, you were seventy empathetic and

438
00:25:09,119 --> 00:25:12,159
Speaker 6: you were eighty seven in terms of delivering this therapy technique.

439
00:25:12,200 --> 00:25:13,800
Speaker 6: Here's a couple of things you could have done better

440
00:25:13,800 --> 00:25:17,200
Speaker 6: in treatment. That's awesome because a therapist, I don't get feedback,

441
00:25:17,240 --> 00:25:19,600
Speaker 6: Like the reason I can improve my basketball game is

442
00:25:19,600 --> 00:25:20,879
Speaker 6: because when I take a shot, I know if it

443
00:25:20,920 --> 00:25:23,159
Speaker 6: goes in or not. We need that feedback as people,

444
00:25:23,200 --> 00:25:26,359
Speaker 6: and so using AI to allow therapists to also be

445
00:25:26,480 --> 00:25:30,040
Speaker 6: better therapists, I think is a super exciting opportunity to

446
00:25:30,040 --> 00:25:31,800
Speaker 6: get to that point that I was making that it's

447
00:25:31,840 --> 00:25:34,600
Speaker 6: not just about access, it's about quality, and we need

448
00:25:34,640 --> 00:25:37,640
Speaker 6: to improve the quality of mental health services that we're

449
00:25:37,640 --> 00:25:40,840
Speaker 6: providing people, and AI has an opportunity to help do that.

450
00:25:42,000 --> 00:25:47,120
Speaker 1: Interesting AI for the therapist, not for the patient. Right,

451
00:25:47,600 --> 00:25:52,399
Speaker 1: if you could set the gold standard for an AI

452
00:25:52,480 --> 00:25:55,159
Speaker 1: therapy chatbot. What does that look like?

453
00:25:56,359 --> 00:25:58,840
Speaker 6: I think the gold standard for an AI therapy chatbot

454
00:25:58,880 --> 00:26:01,240
Speaker 6: would be to be effective, and it would have to

455
00:26:01,240 --> 00:26:05,120
Speaker 6: be safe. So effective, I want some demonstration that it

456
00:26:05,200 --> 00:26:08,280
Speaker 6: is able to provide the claims that it provides it

457
00:26:08,280 --> 00:26:10,760
Speaker 6: can do. Like if it says it can help you

458
00:26:10,800 --> 00:26:15,280
Speaker 6: get through depression, it can help you overcome postermac stress disorder,

459
00:26:15,320 --> 00:26:18,520
Speaker 6: it can overcome obsessive compulsive disorder. I want some indication

460
00:26:18,600 --> 00:26:21,320
Speaker 6: that it actually does that. I also want some indication

461
00:26:21,440 --> 00:26:24,919
Speaker 6: of safety. So what happens with that data and that

462
00:26:25,000 --> 00:26:27,880
Speaker 6: information when I give it to that AI chatbot when

463
00:26:27,920 --> 00:26:30,800
Speaker 6: I talk to a therapist. I understand that information will

464
00:26:30,800 --> 00:26:34,640
Speaker 6: be confidential with some boundaries around safety and some other

465
00:26:34,720 --> 00:26:38,680
Speaker 6: aspects of legality. When I type my stuff into chat GPT,

466
00:26:39,800 --> 00:26:41,520
Speaker 6: they own that data. I don't know what they're going

467
00:26:41,600 --> 00:26:42,760
Speaker 6: to do with it. I don't know if they're going

468
00:26:42,800 --> 00:26:45,760
Speaker 6: to sell it to a third party, use it for advertising, whatever.

469
00:26:46,240 --> 00:26:49,000
Speaker 6: So I think we need some aspects that the data

470
00:26:49,160 --> 00:26:52,280
Speaker 6: is safe and it's secure, improper safeguards are in place.

471
00:26:52,359 --> 00:26:56,199
Speaker 6: Therapy is also not harmless, like stuff can come up

472
00:26:56,200 --> 00:26:58,600
Speaker 6: in therapy. It can be hard, but you understand a

473
00:26:58,640 --> 00:27:00,800
Speaker 6: little bit of at least getting in you're talking about

474
00:27:00,800 --> 00:27:03,560
Speaker 6: these emotional things. Here's some of the challenges that might

475
00:27:03,600 --> 00:27:07,120
Speaker 6: come up. Here are some of the potential aspects. If

476
00:27:07,119 --> 00:27:09,760
Speaker 6: you share information about yourself that you know you're gonna

477
00:27:09,760 --> 00:27:12,000
Speaker 6: harm yourself, it's gonna have to be shared with authorities

478
00:27:12,040 --> 00:27:14,160
Speaker 6: like that. There's stuff that can come up in therapy too.

479
00:27:14,560 --> 00:27:16,879
Speaker 6: But I think we need to understand what's the contract,

480
00:27:16,920 --> 00:27:19,200
Speaker 6: what's the agreement, what's the safeguards that are in prace

481
00:27:19,280 --> 00:27:20,480
Speaker 6: when we're talking with these.

482
00:27:20,680 --> 00:27:24,760
Speaker 1: I'd imagine there are probably some people out there who

483
00:27:25,480 --> 00:27:29,800
Speaker 1: hear the phrase AI and therapy mentioned in the same sentence,

484
00:27:29,840 --> 00:27:31,919
Speaker 1: and they say, Nah, hit the kill switch on that.

485
00:27:31,960 --> 00:27:35,120
Speaker 1: We absolutely do not want this. We don't want those

486
00:27:35,200 --> 00:27:40,040
Speaker 1: two words in the same sentence at all. It sounds

487
00:27:40,040 --> 00:27:42,280
Speaker 1: like you're not You wouldn't quite agree with that.

488
00:27:44,119 --> 00:27:47,640
Speaker 6: I'm really excited for the potential, and I really think

489
00:27:47,680 --> 00:27:52,480
Speaker 6: that we have an opportunity to provide more and better

490
00:27:52,600 --> 00:27:55,560
Speaker 6: services to people, and technology has a role to play there.

491
00:27:55,720 --> 00:27:58,119
Speaker 6: I think where I get really nervous is in like

492
00:27:58,320 --> 00:28:01,919
Speaker 6: the near term. If someone tells me, Doctor Schuller, I

493
00:28:02,000 --> 00:28:04,600
Speaker 6: have an AI chot bot. It works awesome, we're going

494
00:28:04,600 --> 00:28:06,760
Speaker 6: to provide it for therapy. Do you want to start

495
00:28:07,119 --> 00:28:09,480
Speaker 6: handing this out to people tomorrow? I'm holding my wallet.

496
00:28:09,480 --> 00:28:12,240
Speaker 6: I'm like, no, I'm not convinced where they are yet.

497
00:28:12,720 --> 00:28:15,800
Speaker 6: So I am excited about the potential, but I really

498
00:28:15,880 --> 00:28:17,679
Speaker 6: think that there's a lot of stuff we need to

499
00:28:17,880 --> 00:28:21,040
Speaker 6: figure out to make this effective and safe.

500
00:28:22,240 --> 00:28:24,439
Speaker 1: All right, And just as an aside here, what I

501
00:28:24,480 --> 00:28:27,000
Speaker 1: was picking up from Stephen in our conversation is that

502
00:28:27,080 --> 00:28:30,920
Speaker 1: there really is potential here, but he's holding his wallet, right.

503
00:28:31,000 --> 00:28:33,200
Speaker 1: He wants to be cautious and do it the right way.

504
00:28:33,760 --> 00:28:37,520
Speaker 1: But if we're talking about the wallet, it's not people

505
00:28:37,640 --> 00:28:40,479
Speaker 1: like doctor Stephen Schuler who are holding that AI wallet.

506
00:28:40,840 --> 00:28:43,640
Speaker 1: The decisions about when these things get released, they are

507
00:28:43,680 --> 00:28:46,280
Speaker 1: not up to people like him. When it comes down

508
00:28:46,280 --> 00:28:51,640
Speaker 1: to it, this is a business. I want the best

509
00:28:51,640 --> 00:28:55,200
Speaker 1: and brightest minds, especially in mental health, to be in

510
00:28:55,240 --> 00:28:58,000
Speaker 1: a place where there it is not a profit motive.

511
00:28:58,080 --> 00:29:00,520
Speaker 1: I don't necessarily want the best and brightest minds in

512
00:29:00,560 --> 00:29:02,720
Speaker 1: the world to be at a company where they it

513
00:29:02,800 --> 00:29:06,400
Speaker 1: being I would imagine being pushed to hey man, we

514
00:29:06,480 --> 00:29:11,760
Speaker 1: got to ship this thing. I want the rigor that

515
00:29:11,880 --> 00:29:17,200
Speaker 1: is happening at a public university where somebody can spend time,

516
00:29:17,320 --> 00:29:21,400
Speaker 1: and people can spend time on really truly making sure

517
00:29:21,440 --> 00:29:24,600
Speaker 1: that something is safe, making sure that something is effective,

518
00:29:25,200 --> 00:29:28,560
Speaker 1: and they're not being pressured to create a product. But

519
00:29:28,560 --> 00:29:31,560
Speaker 1: at the tables have turned. The data is being held

520
00:29:31,560 --> 00:29:34,960
Speaker 1: by companies, and in that sense, the power has shifted

521
00:29:35,000 --> 00:29:38,120
Speaker 1: over to a side of the table that has a

522
00:29:38,120 --> 00:29:40,200
Speaker 1: profit motive where it didn't used to be like that.

523
00:29:40,960 --> 00:29:43,560
Speaker 6: I agree with you, And also, dollars make the world

524
00:29:43,560 --> 00:29:46,360
Speaker 6: go around, and so I think like eventually someone's got

525
00:29:46,400 --> 00:29:49,200
Speaker 6: to pay for things. One thing that's challenging with mental

526
00:29:49,200 --> 00:29:53,320
Speaker 6: health because there's so much need and there's so little

527
00:29:53,360 --> 00:29:56,200
Speaker 6: services that the fact of the matter is if we

528
00:29:56,280 --> 00:29:58,920
Speaker 6: want to solve this problem, we're going to have to

529
00:29:59,080 --> 00:30:01,479
Speaker 6: pay for it. I feel like we need to make

530
00:30:01,520 --> 00:30:06,480
Speaker 6: sure that we go slow and we go responsibly because

531
00:30:06,520 --> 00:30:08,680
Speaker 6: the sort of tech idea of like move fast and

532
00:30:08,720 --> 00:30:11,560
Speaker 6: break things, yeah, it does not work for mental health.

533
00:30:11,600 --> 00:30:13,520
Speaker 6: I do not want my mental health broken anymore.

534
00:30:13,520 --> 00:30:15,360
Speaker 5: I want it fixed. I want it helped. I want

535
00:30:15,360 --> 00:30:16,360
Speaker 5: the system better.

536
00:30:17,680 --> 00:30:20,600
Speaker 1: Maybe the whole idea of AI therapy still sounds really

537
00:30:20,640 --> 00:30:23,640
Speaker 1: bleak to you, but hopefully it makes sense why some

538
00:30:23,680 --> 00:30:27,800
Speaker 1: people might find this promising, either societally or just personally

539
00:30:27,840 --> 00:30:30,760
Speaker 1: for themselves. I asked doctor Schuler if there was anything

540
00:30:30,760 --> 00:30:33,200
Speaker 1: else he wanted to add, and he said there was.

541
00:30:34,200 --> 00:30:37,480
Speaker 6: I'll just say anyone out there who's struggling with mental

542
00:30:37,520 --> 00:30:41,440
Speaker 6: health issues, there's help out there, and you can get better.

543
00:30:41,520 --> 00:30:43,960
Speaker 6: And mental health is also a journey, and it doesn't

544
00:30:44,000 --> 00:30:47,280
Speaker 6: mean you're here today and you're completely better tomorrow.

545
00:30:47,320 --> 00:30:48,320
Speaker 5: There's ebbs and flows.

546
00:30:48,720 --> 00:30:52,360
Speaker 6: Because we've talked about I think AI therapy is a potential,

547
00:30:52,720 --> 00:30:54,440
Speaker 6: has the potential to be a tool in the toolbox,

548
00:30:54,480 --> 00:30:57,120
Speaker 6: but there's other tools out there too, And just also

549
00:30:57,160 --> 00:31:00,360
Speaker 6: appreciate that not everything works for everybody. So if you

550
00:31:00,360 --> 00:31:02,880
Speaker 6: try something and it doesn't work, try something else. But

551
00:31:03,040 --> 00:31:05,920
Speaker 6: I just, yeah, really want to try to have a

552
00:31:05,960 --> 00:31:08,320
Speaker 6: message of hope for people who are struggling with these things,

553
00:31:08,360 --> 00:31:10,240
Speaker 6: because I know it's hard and I know it's challenging,

554
00:31:10,280 --> 00:31:12,760
Speaker 6: and I think definitely when you're at your lowest, you

555
00:31:12,800 --> 00:31:14,480
Speaker 6: want to be protected and you want to be safe,

556
00:31:14,480 --> 00:31:17,160
Speaker 6: and I think that's why this conversation is so important.

557
00:31:26,760 --> 00:31:29,200
Speaker 1: Thank you so much for listening to another episode of

558
00:31:29,280 --> 00:31:31,800
Speaker 1: kill Switch. Let us know what you think. If there's

559
00:31:31,880 --> 00:31:34,000
Speaker 1: something you want to say, or if there's something you

560
00:31:34,000 --> 00:31:36,560
Speaker 1: want us to cover, you can email us at kill

561
00:31:36,600 --> 00:31:40,360
Speaker 1: Switch at Kaleidoscope dot NYC, or you can find us

562
00:31:40,360 --> 00:31:44,320
Speaker 1: on Instagram at kill switch pod or me personally. I'm

563
00:31:44,600 --> 00:31:47,520
Speaker 1: at dex digit that's d e X d I G

564
00:31:47,720 --> 00:31:50,880
Speaker 1: I And wherever you're listening to this podcast, you know,

565
00:31:50,960 --> 00:31:53,400
Speaker 1: think about leaving us a review. It helps other people

566
00:31:53,440 --> 00:31:56,280
Speaker 1: find the show, which in turn helps us keep doing

567
00:31:56,320 --> 00:31:59,920
Speaker 1: our thing. Kill Switch is hosted by Me Dexter Thomas.

568
00:32:00,240 --> 00:32:04,240
Speaker 1: It's produced by Shena Ozaki, Darluk Potts, and Kate Osborne.

569
00:32:04,560 --> 00:32:07,200
Speaker 1: Our theme song is by me and Kyle Murdoch, and

570
00:32:07,320 --> 00:32:11,280
Speaker 1: Kyle also mixes a show from Kaleidoscope. Our executive producers

571
00:32:11,280 --> 00:32:15,880
Speaker 1: are Oswa lashin On, Gesshakti Kadour and Kate Osborne. From iHeart,

572
00:32:15,880 --> 00:32:19,440
Speaker 1: our executive producers are Katrina Normal and Nikki e Tor.

573
00:32:20,160 --> 00:32:22,800
Speaker 1: One last thing, since you still hear. As Steven and

574
00:32:22,800 --> 00:32:26,120
Speaker 1: I were talking about human versus AI therapists, it occurred

575
00:32:26,160 --> 00:32:28,760
Speaker 1: to me that maybe some people are deciding that they

576
00:32:28,880 --> 00:32:31,720
Speaker 1: actually don't want a human and we talked a little

577
00:32:31,720 --> 00:32:35,040
Speaker 1: about that. Well, I think there's also another element of it,

578
00:32:35,080 --> 00:32:38,760
Speaker 1: which is that maybe it's not even always about believing

579
00:32:38,800 --> 00:32:41,680
Speaker 1: that it's a human or not, which is to say,

580
00:32:41,800 --> 00:32:45,920
Speaker 1: we really tend to trust technology in general. And even

581
00:32:45,920 --> 00:32:50,800
Speaker 1: if I know that CHATCHBT or even a properly made

582
00:32:51,760 --> 00:32:56,200
Speaker 1: AI therapy purposely built therapy chatbot is not a human,

583
00:32:56,640 --> 00:32:59,800
Speaker 1: it still feels very authoritative. Maybe the issue is that

584
00:33:00,280 --> 00:33:02,680
Speaker 1: we look at a computer and we see that glowing

585
00:33:02,720 --> 00:33:05,080
Speaker 1: rectangle and we think this is an authority because in

586
00:33:05,120 --> 00:33:08,560
Speaker 1: almost every other aspect of our lives, the computer is

587
00:33:08,600 --> 00:33:09,240
Speaker 1: an authority.

588
00:33:09,520 --> 00:33:09,720
Speaker 3: Yeah.

589
00:33:09,760 --> 00:33:10,760
Speaker 5: I think that's a great point.

590
00:33:10,800 --> 00:33:15,120
Speaker 6: I mean definitely I see people, you know, understand or

591
00:33:15,160 --> 00:33:18,280
Speaker 6: think that, like the Internet knows everything. My kids say

592
00:33:18,320 --> 00:33:20,680
Speaker 6: whenever they have questions about things, they're not like, you know,

593
00:33:20,960 --> 00:33:23,040
Speaker 6: what's the answer to this, Dad, They're like, ask Google.

594
00:33:23,160 --> 00:33:26,520
Speaker 5: Google knows and Google often does know. It's all knowing,

595
00:33:26,680 --> 00:33:28,120
Speaker 5: It's got all the information.

596
00:33:28,280 --> 00:33:30,480
Speaker 6: At your point's a really good one is that technology

597
00:33:30,520 --> 00:33:33,720
Speaker 6: is also it's a separate thing in our life, snap

598
00:33:33,920 --> 00:33:36,080
Speaker 6: things like Google, things like chat should be t is

599
00:33:36,120 --> 00:33:39,240
Speaker 6: like it knows dus, It's got all the knowledge, Like

600
00:33:39,600 --> 00:33:41,520
Speaker 6: why would it not be better than a person?

601
00:33:41,640 --> 00:33:43,720
Speaker 5: Like it's got the whole Internet knowledge on it.

602
00:33:43,920 --> 00:33:47,560
Speaker 6: My therapist doesn't my therapist doesn't know everything, but Google

603
00:33:47,800 --> 00:33:48,480
Speaker 6: has everything.

604
00:33:48,640 --> 00:33:50,240
Speaker 5: So yeah, it's a good point.

605
00:33:51,040 --> 00:33:53,600
Speaker 1: I'm not really sure what to do with that yet.

606
00:33:54,400 --> 00:33:58,360
Speaker 1: If you got any ideas, let me know anyway, catch

607
00:33:58,360 --> 00:33:58,960
Speaker 1: you next time.

608
00:34:00,080 --> 00:34:00,360
Speaker 3: Bye,