This is my first video in a new format, podcast or interview style. I interviewed a German therapist named Ursula, who is also starting out on KZfaq. We discussed three main questions:
1. What it's like to be a therapist
2. How will AI change the job of a therapist?
3. How will AI change society?
If you want to focus on AI content, you can probably skip the first section which is only about 10 minutes anyways. In the last section, I do the majority of the talking and we discuss how AGI can impact the profession of therapy. I also discuss my reasons for creating a KZfaq channel.
#ai #jobs #therapist
0:00 Intro
0:23 Contents
0:29 Part 1: What is it like to be a therapist?
0:33 Ursula introduces herself
1:45 Are people scared of therapists?
3:20 More widespread acceptance by public
4:31 German health care system
5:41 Day in the life of a therapist
6:48 Effectiveness of virtual therapy
8:14 Different cues online, can't see feet
9:39 How much info is beneath the words?
10:58 Mental model of patient's reality
12:42 Sherlock Holmes
13:59 Part 2: How will AI change the job?
14:47 Job requires lived human experience
15:29 Making an AI system to replace the core job
16:49 AI system could be best of many world
17:22 AI filters incoming patients
18:28 Homework and exercise automation
19:28 AI system suggesting ideas to therapist
20:05 Health care apps
21:17 Determining when to start intervention
22:41 Start by talking about goals
23:27 Limitations of large language models
24:27 Models are good at picking up nuance
24:53 Chatbot therapists
26:25 Three ways AI replaces jobs
27:59 Outsider thinking about the field
28:57 Common culture between therapist and patient
30:04 Patient might not want AI
30:55 Difficult to generalize AI therapy across cultures
31:42 Vast data required to train an AI system
32:32 Humans receive lots of data
33:45 System 2 vs system 1 and post hoc explanation
35:05 AI is basically only system 1
36:03 Can an AGI system become a therapist?
37:19 AGI means all bets are off
39:41 Will AGI surprise us like ChatGPT did?
40:04 Part 3: How AI will change society
40:10 That's just search
40:43 Increase in relative ability is what's surprising
41:48 RLHF and the history of ChatGPT
43:01 Taking generative AI in any direction
44:40 Also surprised at ChatGPT
45:33 Why I started a KZfaq channel
46:29 Exponential growth in machine learning
47:37 AI regulation in the European Union
48:46 Parallels with GDPR privacy regulation
50:24 EU will have greater visibility into AI use
51:13 Regulation in other countries
51:59 Are people too afraid or not afraid enough?
53:03 AI is here to stay for ordinary people
54:29 Existential risk is too scary to think about
55:33 Alignment problem and bad human operators
56:37 Alignment and training accidents
57:37 Humanity is the defender, which is a hard problem
58:25 The purpose of this KZfaq channel
59:34 People are most afraid of AI when they know nothing
1:00:32 Closing thoughts
1:02:24 Outro