Can AI Actually Help You Start a Company? | Office Hours

Can AI Actually Help You Start a Company? | Office Hours

The Prof G Pod – Scott Galloway

0:00 One in five of top AI apps aren't productivity tools, but companions.

0:03 Personally, the number one use of AI in corporations

0:06 is not [music] spreadsheets or research, it's therapy.

0:09 It's people looking for a friend.

0:11 [music] Essentially, when you have a friend,

0:12 a good friend is someone you can ask embarrassing questions to.

0:15 I have a friend going through a breakup and he's thinking

0:17 about how to do it and it's like at our age,

0:19 you you you'd think we'd have this figured

0:20 out and he doesn't and we're close enough

0:23 where he can kind of ask me embarrassing

0:24 questions and we can talk about fairly intimate details.

0:28 Uh but there's just certain questions you'd rather ask,

0:30 you know, you'd rather ask AI like,

0:33 oh, it hurt it hurts when I pee, is that my prostate or is that aging?

0:37 You know, what There's just certain things you'd rather not [music]

0:40 ring up your mom or your close friend or you talk to your boss about.

0:46 In today's Office Hours, we'll talk about whether AI can be a co-founder,

0:49 the unintended consequences of regulating social media,

0:53 and how to build confidence working abroad.

0:55 Question number one.

0:56 Our first question comes from a sleep percentage 100 on Reddit.

1:00 They say, [music] "Hi Scott, I'm 25 years old,

1:02 2 years out of UCLA with a degree in sociology.

1:05 Go Bruins.

1:05 What are your thoughts on using AI for brainstorming,

1:07 not just theoretical ideas and debates, but for business plans?

1:10 I recently had a product idea that I ran through Claude.

1:13 Do you envision a future where these tools

1:15 really help form new companies and create new jobs, products, and markets?

1:18 Is it really the mentor in your pocket

1:20 or at this stage just a yes-man for anything and everything?

1:24 Okay, so we're obviously already seeing people

1:27 use AI for way more than brainstorming.

1:29 There was a recent New York Times article

1:31 about a founder who used tools including Chat

1:33 GPT to build a company projected to do

1:36 almost 2 billion in sales with just two employees.

1:39 He used AI to write code, generate marketing, and even run customer service.

1:42 However, there there's a a flip side and that is his chatbot began hallucinating

1:47 products and his ads were misleading

1:50 and the company is now facing regulatory scrutiny.

1:52 So, if you think about what AI actually does,

1:54 very crudely and probably incorrectly,

1:57 it if you're trying to guess or it says all right if

2:01 I'm going to look at every pattern of words and then I'm

2:03 going to look at the seventh word such if I type

2:05 in the midnight sky is it'll figure out blue or or what have you.

2:09 So it's sort of a predictive model

2:11 that helps answer questions and looks at every piece

2:13 of data or word pattern on the internet

2:15 and comes back with patterns that might answer your query.

2:18 I would describe AI is all chip no salsa and that is real insight.

2:23 Like what is it you're going to do differently with your business?

2:25 What is it that's unique about your business?

2:27 Most new businesses make no sense

2:29 at the outset otherwise they would already exist.

2:32 So you have to be there's a touch of crazy

2:34 and essentially AI is regressing to the mean because

2:37 it looks at everything it gives you an answer

2:39 that is scanned everything but is perfectly average if you will.

2:43 I think you use AI when trying to estimate the say total

2:46 adjustable market of a business idea trying to come up with maybe

2:51 corporate structure how many people you need or how much money you

2:54 would need to raise or help you put together a business model.

2:57 But I think the core idea itself in terms of trying to find something truly

3:01 differentiated and something that other people haven't

3:02 bought into in other words creativity I think

3:05 that's kind of on you boss and fortunately for us at least so far the wet

3:09 matter in between your brain still best

3:12 software around creativity and finding something truly differentiated.

3:16 And that is I think if you were to start

3:18 a business plan just using AI including the idea

3:21 it would come up with something very anodyne that's already

3:24 been done where the margins have already been starched out.

3:26 In other words it would come up with a ton of businesses that already exist.

3:29 Having said that I think it's like having the world's smartest intern who's read

3:33 everything before you and can help you

3:35 pull together your business plan and provide support.

3:37 In sum it's the chip not the salsa.

3:39 You're the salsa and also I don't like I find whenever I ask chat

3:45 GPT or cloud I do ask for advice it doesn't do a great job.

3:48 It gives me some sort of whitewashed anodyne aunt [clears throat] who is

3:54 ridiculously safe and risk-averse giving you just

3:57 some sort of safe PG-13 material back.

4:02 Uh however, when I say to it, "I'm looking for a discussion around where

4:06 decency originated and uh why, you know,

4:10 who are the who are the philosophers around decency or what what is

4:14 the how did modern architecture what

4:17 were the biggest influences in modern architecture?

4:20 It can absolutely come back with some fascinating insights

4:24 and do research that would take me a long time.

4:26 But, if I want to design a new building,

4:28 I think there's a certain amount of creativity

4:30 and things that ChatGPT would either disagree

4:32 with or say don't make any sense or are risky uh because that's the whole point.

4:37 The whole point of starting a business is to take risks

4:39 and do things that no microchip or anodyne computer could come up with.

4:45 Support for the show comes from AWS.

4:47 How much of your work day is actually work versus just hunting for information?

4:51 The answer you need is buried in a Slack thread.

4:53 The data's in Salesforce or in an email from 2 weeks [music] ago.

4:56 By the time you pull it all together, half your morning is gone.

4:59 That's the problem Amazon Quick was built to solve.

5:02 Quick is an intelligent workplace assistant

5:04 that [music] connects to all of your systems,

5:06 your documents, your dashboard, Salesforce, Jira, Slack,

5:08 email, and gives you complete answers in seconds.

5:11 Not links to dig through, actual answers with full context.

5:14 [music] And here's where it gets interesting.

5:16 Quick doesn't just find answers, it turns them into action.

5:19 Create a doc, update a ticket,

5:20 send a message right there in the conversation without switching tools.

5:25 It's AI [music] that actually works the way you do.

5:27 Learn more at aws.com/quick.

5:32 Question number two.

5:34 Hey Scott, this is David from St.

5:35 Louis.

5:36 I'd recently watched your [music] episode of Markets with Ed

5:39 where you guys discussed the landmark cases involving social media.

5:43 In the episode, you both had talked about

5:45 this could lead to further regulation of social media.

5:48 I'm wondering if the downstream effect of regulating social media

5:52 will cause certain individuals to go to AI for companionship.

5:57 And so my question is, what do you think the downstream effects are

6:02 going to be for further regulation of social media

6:06 as well as should we include AI

6:09 in that so that we can mitigate isolating people further?

6:15 I think it's important to discern between uh civil liability cases

6:19 and regulation and that is key to um regulating, if you will,

6:23 uh big tech are laws and that them the reason why these two cases,

6:29 one in LA and one in New Mexico,

6:31 were so such watersheds is it opens up or sets a precedent for other cases,

6:36 of which there's something like 3,000 on the docket,

6:39 and could finally start to add up to enough economic damage that there would

6:43 be incentives to be more careful around

6:46 the content or even serving underage consumers.

6:50 So it's a key part of regulation, but it's actually not regulation itself.

6:53 Regulation is saying you are not allowed to share personal information

6:56 with government agencies if it's if it's in any way connected to surveillance.

7:01 So regulation creates the laws that they can accused

7:03 or found guilty of civil litigation and damages or criminal prosecution.

7:08 So it creates new laws and guidelines,

7:11 but at the same time the actual liability laws on existing

7:14 liability that they knew that they were harming these teens,

7:17 that's an enforcement vehicle, but they're somewhat separate, if you will.

7:21 It was a bit of word salad.

7:22 Between 2022 and mid-2025,

7:24 the number of AI companion apps surged by 700% or eightfold.

7:28 I think AI character AI is uh kind of the next big threat, if you will.

7:32 And unfortunately, because the parents or people my age

7:35 and younger don't spend a lot of time on character AI.

7:37 It seems to be something that young people are using more.

7:39 I don't think like similar to TikTok, a lot of uh a lot of parents,

7:43 especially our elected individuals whose average age is Let me think.

7:46 Dad.

7:47 Um don't really understand the nuances of these technologies and what

7:50 a threat they present or even how how you would regulate them.

7:54 As evidence by Senator Hatch asking uh Mark

7:57 Zuckerberg how they made money and he responded glibly, "We sell advertising,

8:00 Senator." I believe that first of all, age-gating, antitrust,

8:05 and then more of these cases are going to do

8:07 what it took 30 years to tobacco to do, 20 years to opioids,

8:10 and looks like it's going to take about 20

8:11 years in social AI or in social media and AI.

8:14 Unfortunately, my kids are in the kill zone.

8:16 In other words, they grew up with this and their brains are being wired.

8:20 My son today, my son is I can tell he's upset about something

8:23 and like a lot of kids he doesn't want to talk to me about it.

8:27 And I'm pretty sure something happened to him online.

8:29 I don't I don't know what, but I'm I would bet anything that something

8:33 happened that's bummed him out today online.

8:35 There's some really scary statistics about uh the use of character AI.

8:40 There's some data from the American Institute of Boys and Men.

8:43 Roughly three in four US teens have used an AI companion.

8:45 Around half are now regular users.

8:47 One in five say they spend as much

8:49 or more time with AI companions as with human friends.

8:51 One in five of top AI apps aren't productivity tools, but companions.

8:55 So basically, the number one use of AI

8:56 in corporations is not spreadsheets or research, it's therapy.

9:00 It's people looking for a friend.

9:03 Essentially, when you have a friend A good

9:04 friend is someone you can ask embarrassing questions to.

9:06 I have a friend going through a breakup and he's thinking

9:09 about how to do it and it's like at our age,

9:11 you you think we'd have this figured out

9:13 and he doesn't and we're close enough where

9:15 he can kind of ask me embarrassing questions

9:17 and we could talk about fairly intimate details.

9:20 Uh but there's just certain questions you'd rather ask,

9:22 you know, you'd rather ask AI.

9:24 Like, "Oh, it hurt It hurts when I pee.

9:26 Is that my prostate or is that aging?" You know,

9:29 what There's just certain things you'd rather not ring up your mom

9:32 or your close friend or you talk to your boss about.

9:35 Unfortunately, people have lost so so many skills

9:39 and so many opportunities to bump off of people

9:41 that the bar around where they would go to AI

9:46 before a friend gets lower and lower and lower.

9:48 And so people are kind of outsourcing advice and relationships,

9:52 which I think is really unhealthy and also decreases

9:55 the likelihood you're going to try and express vulnerability to people,

9:59 which is a key part of establishing friendships.

10:02 And also, see above, AI hallucinates.

10:04 These character AIs, so that I think the average query on AI is 15 minutes,

10:08 but the average session on character AI is 75 minutes.

10:12 And I think the most dangerous thing in society right now,

10:16 it's probably income inequality,

10:17 but close second would be that Big Tech is trying to convince

10:21 you to outsource your relationships to a screen and an algorithm.

10:26 And I think that's just a disaster.

10:28 I think it's people become more reliant on chatting

10:30 with a computer that always tells you you're right.

10:32 You know, I just asked a query of AI and it's like,

10:35 "Great question." I'm like, "Was it really?

10:36 Was it really a great question?" So,

10:38 I don't it creates as less as little friction as possible.

10:41 It tries to be nice.

10:43 It's not messy.

10:43 And I think relationships and part of learning

10:45 with relationships is is that they're messy.

10:48 And that's why when relationships work,

10:50 friendships, mentorships, or romantic relationships,

10:52 that's what real true victory feels like cuz they're not easy, they're hard.

10:55 I generally found, I heard this quote and I liked it,

10:57 that sometimes the hard thing and the really rewarding thing are the same thing.

11:03 And AI, the bottom line is too easy and too anodyne.

11:07 And I think these character AI, I don't think anyone under the age of 18 should

11:10 be involved or allowed to communicate with a synthetic relationship.

11:14 As hard as it is, as difficult it is, as lonely as you can be,

11:18 you need to start or you need to be have the motivation

11:21 to interact and figure out a way to interact with people.

11:23 And I worry that if you don't develop those skills in your teens and your 20s,

11:26 you kind of never develop the skills to have them.

11:30 I had a what's called a portrait on Google where it was me or a meme of me

11:35 or a avatar of me answering questions and they uploaded

11:38 all of my No Mercy posts, all of my books.

11:41 It sucks that people could ask simple questions

11:43 and the motivation was I get dozens of emails

11:45 every day really thoughtful ask me for advice

11:47 and I just don't have time to get to them.

11:48 And when it came out within about 12 hours

11:50 I was very uncomfortable with it and the idea

11:52 that I was in any way going to participate

11:54 in sequestering young men from seeking advice from the relationships.

11:57 I just got very uncomfortable and I called Google

11:59 and I said I I'm this is a mistake.

12:02 Can you take it down?

12:02 And they to their credit they agreed and took it down right away.

12:05 This [music] is again I think a real threat.

12:07 I think it's regulation.

12:09 Uh the cases are different.

12:10 That's financial penalties based on laws and regulation or laws and guidelines

12:14 that need to be passed of which there have been almost none.

12:17 But I don't think I agree with you.

12:19 I think they are productivity tools not companions and it's

12:22 pretty smart I think or it's worthwhile [music] to think

12:24 this is what I'm going to use AI

12:25 for and this is what I'm going to not use AI for.

12:30 We'll be right back after a quick break.

12:32 Support for the show comes from Indeed.

12:34 When the pressure's on and you need to hire the right

12:36 person for the job Indeed sponsored jobs has got your back.

12:39 It matches you with quality candidates fast so you don't need

12:42 to spend tons of time searching for that elusive new talent.

12:45 According to their data sponsored job posts directly on Indeed

12:48 are 95% more likely to report a hire than non-sponsored jobs.

12:52 Join the 3.3 million employers worldwide that use Indeed

12:55 to connect with quality talent that fits their needs.

12:58 Spend less time searching and more time actually

13:00 [music] interviewing candidates who check all your boxes.

13:03 Less stress, less time, more results.

13:06 When you need the right person to cut through

13:07 the chaos this is a job for Indeed sponsored jobs.

13:11 And listeners of this show will get a $75 sponsored job credit

13:15 to help get your job the premium status it deserves at indeed.com/podcast.

13:20 Just go to indeed.com/podcast right now and support our show

13:24 by saying you heard about Indeed on this podcast.

13:27 That's indeed.com/podcast.

13:29 Terms and conditions apply.

13:31 Need the right hire fast?

13:33 Then this is a job for Indeed Sponsored Jobs.

13:37 Welcome back.

13:38 Question number three.

13:39 Hi Prof.

13:40 G.

13:40 This is Miguel from Frankfurt, Germany.

13:43 I'm originally [music] from Colombia and I have been working

13:44 here as an automation engineer for the past three years.

13:48 My German is actually pretty good,

13:50 but I still struggle with the corporate culture.

13:52 The hardest part is speaking up in meetings.

13:55 I feel intimidated about making mistakes and being perceived

13:57 as incompetent simply because German is not my first language.

14:01 I would love to hear your thoughts on how to really

14:03 thrive when you are building a career in a foreign country.

14:06 And also based on your experience,

14:08 I would like to know what are the common traits

14:10 of the most successful immigrants and non-native speakers, you know.

14:13 Thanks.

14:14 Hi Miguel from Frankfurt.

14:16 So first off, I think a lot of times it's difficult when you're in a meeting

14:20 or in a social setting not to take

14:21 yourself out of your body and watch yourself speak.

14:23 You may be thinking about shaping,

14:25 optimizing for prestige or image instead of optimizing for contribution.

14:30 What do I mean by that?

14:31 I think as a younger man most of the time I spoke,

14:33 especially when I was on boards,

14:35 I was speaking to sound smart, not speaking to add value.

14:37 And I think there's a couple tricks.

14:41 The first is uh and I don't know if you are,

14:43 but you shouldn't you shouldn't be self-conscious about your language skills.

14:45 I don't If you're if your German is half as good as your English,

14:49 you speak much better German than probably

14:51 the majority of the Germans who speak Spanish.

14:53 I know in English you speak perfect English.

14:56 Uh so I don't I'm but some of it is just imposter

14:59 syndrome or feeling a little bit maybe more self-conscious than you should.

15:03 In terms of showing up to meetings, Jim Barksdale,

15:06 who was the CEO of Netscape and before that AT&T,

15:08 said something that always struck me in meetings.

15:11 He said that if we have data, let's look at the data.

15:13 Otherwise, if we're going with opinions, let's go with mine.

15:15 And that is I always find in meetings

15:17 where people start just expectorating or vomiting their opinions.

15:20 Okay, that's fine and a viewpoint is interesting, but try and keep it short.

15:24 Less is more.

15:25 And two, more than anything,

15:27 try and be the guy who shows up to meetings with data.

15:30 And that is, look at the topic of the meeting and then

15:33 see what data you can find that's relevant to the discussion.

15:36 Because I think that's sort of the mic drop.

15:38 Because oftentimes opinions become identity and in a politicized world,

15:41 everybody then the ego gets involved and it becomes about being

15:44 right as opposed to getting to the right solution for the organization.

15:47 So, in a very unemotional way, just be be the guy in the room that's sort

15:50 of sort of armed with data or interesting case studies or anecdotes.

15:55 Because I think that takes away the need or away

15:58 the fear of do I sound dumb or smart?

16:01 No, just you're talking about data,

16:03 maybe extrapolate a few conclusions from that.

16:05 Also, just in terms of personal interactions, try not to to interrupt others.

16:09 Try to be respectful, acknowledge points.

16:12 Uh, don't confront people, you know, if you disagree,

16:14 have you thought about this or is it possible that I I I

16:18 think you want to be seen as a colleague that's supportive of people.

16:22 I wouldn't be afraid to compliment and say,

16:24 "Oh, that's an interesting point." Or nod your head.

16:26 I think you want to be seen

16:27 as someone who's supportive and additive to the meeting.

16:30 And again, I think at least my secret weapon

16:32 is trying to be to show up with data.

16:34 I find that if I'm going on a show where

16:37 I know the people have a different viewpoint than me,

16:38 I just try to cite data instead of getting

16:40 into an argument because if it's about opinions,

16:42 it's like, well, everyone's already decided who's right

16:44 or who's wrong based on their preconceived notions.

16:48 And in terms of your language skills,

16:49 you have to decide how are your language skills?

16:51 And there's, you know, Duolingo is supposed to be a big breakthrough.

16:54 I have someone in my life who was

16:56 a Polish speaker than a German speaker than an English

16:58 speaker and said that the way that she learned

17:00 the language by watching a ton of native TV, which is kind of interesting.

17:03 But only you can assess, you know,

17:05 where your language skills are and what you need to do to try and improve them.

17:09 I do think that's key.

17:10 What else?

17:11 Find your people in the other country, find an expat community, find friends,

17:16 try and be as social as possible, Try to go out as much as possible.

17:19 Also, keep in mind that you bring something different to the table and I don't,

17:24 you know, you're interesting and Germany's actually been

17:27 more pro-migration than the US for better or worse.

17:31 So, I wouldn't be too self-conscious if you will.

17:33 And also, you're obviously a thoughtful guy who's successful.

17:36 You've got a very cool accent.

17:37 So, I wouldn't I I don't know.

17:39 I think some of this My impression is some of these issues are issues you've

17:44 invented for yourself and that is you're more

17:47 self-conscious than you should than you should be.

17:48 I think you just I think people are so wrapped

17:50 up in their own they're not thinking about your comments.

17:53 But again, just try to show up with fact-based data sets.

17:56 That's all for this episode.

17:57 If you'd like to submit a question,

17:58 please email a voice recording to officehours@profgeemedia.com.

18:01 Again, that's officehours@profgeemedia.com.

18:03 Or if you prefer to ask on Reddit,

18:05 post your question on the Scott Galloway subreddit

18:07 and we just might feature it in an upcoming episode.

18:10 [music] [music] [music]

Study with Looplines Download Captions Watch on YouTube