Anthropic's Super Bowl Lies | Sharp Tech with Ben Thompson
Sharp Tech Podcast
0:00 says, "Ben, as online advertising's
0:03 biggest evangelist, I'm curious how you
0:06 received Anthropic's viral Super Bowl ad
0:09 campaign.
0:10 I was a big fan of Anthropic's
0:12 marketing vision early on, but something
0:15 about this roll out leaves me with a
0:17 sour taste.
0:18 After going so hard into
0:19 Enterprise, I'm skeptical consumers will
0:22 view Claude as accessible.
0:24 And it's pretty disingenuous to paint opening
0:26 eyes ads as the thing they explicitly
0:28 said they wouldn't do.
0:30 Context driven inline advertising spots.
0:34 Maybe none of that matters because people hate the
0:36 idea of ads so much.
0:39 So again, a number
0:40 of listeners wanted to get your
0:42 perspective on all this.
0:43 I'm curious what direction you go here.
0:45 What was your reaction to these ads?
0:47 Well, a few things.
0:47 Um, I always thought
0:49 I thought Open Eyes Super Bowl ad last
0:52 year like not that good.
0:55 Uh, Chat GPD's ongoing ads super good.
0:58 I like I I think they're great ads though.
1:00 Like like the one where the guy's like
1:02 making like a pasta dish for his like
1:05 new date or whatever and then it's like,
1:06 "Oh, help me make something." Like it
1:08 just like there there's good music, good
1:10 visuals.
1:11 It captures like it shows what
1:13 the product can do.
1:14 My some of the best
1:15 ads in tech history I think were Apple's
1:17 initial iPhone ads where they just
1:19 literally showed the phone and showed
1:21 you doing stuff on it which was great
1:23 because wow you can do this and number
1:25 two they were literally training people
1:27 how to use the iPhone with their ads and
1:30 it is it's a tough nut to crack figuring
1:32 out how to advertise AI generally like
1:35 my idea would be to just interview
1:37 people and have testimonials with
1:40 different people talking about the ways
1:42 AI just completely transforms their I I
1:44 think the GPT I think the Chad GPT ads
1:47 that run regularly actually do that
1:49 pretty well like they show and they show
1:51 the prompt and they like at the end of
1:52 the commercial is like listing out the
1:54 answer that that it gave or whatever it
1:55 might be.
1:55 These more like abstract brand
1:57 ones that they did last year and we'll
1:59 say they're going to do this year
2:01 like whatever this is.
2:02 Take her or leave them.
2:02 Yeah.
2:03 I mean there's some of the some of it is like
2:06 reputation laundering and my eyes tend
2:08 to glaze over a little bit where it's
2:10 like create a playlist on my long drive
2:12 with my sister.
2:13 or help me join a
2:14 running group, stuff like that.
2:15 Like, I personally would be more persuaded if
2:18 you just put people on camera being
2:21 like, "This[ __] is amazing.
2:23 It's completely changed my workday." Um, and
2:25 that's honestly what has compelled me to
2:28 adopt AI more and more of just talking
2:30 to friends and family.
2:31 So, in any event,
2:32 anthropic
2:34 going a different direction.
2:35 I am going to go the opposite of
2:37 basically everyone on the internet.
2:39 Okay.
2:39 Um, disagreeable.
2:40 It's easy to make a viral ad when you're
2:42 a liar.
2:42 I I thought this ad was
2:44 despicable.
2:45 Like, and it was despicable
2:46 in the same way, and I'm consistent on
2:48 this point where I rad Cook over the
2:52 Kohl's when they were lying about
2:53 Facebook.
2:57 Yeah.
2:56 I I thought this was disgusting.
2:59 Like, they're out there lying.
3:01 That's what this ad was.
3:03 Lying about what OpenAI is doing is is
3:06 the core lie here.
3:06 So yeah, you have the
3:07 like these giving the guy's giving an
3:10 answer like this anthropomorphized AI
3:12 answer and he's inserting an ad in his
3:15 answer and as critical as I am of their
3:18 ad product and what it's going to be,
3:20 that's what it's not going to do.
3:23 And I I just don't have any respect for
3:27 and appreciation for liars.
3:29 And this lowers my
3:32 thinking which frankly was not super
3:34 high about anthropic.
3:36 They frame themselves as the moral ones in this
3:39 space and I think this adds immoral.
3:43 I agree with you.
3:44 Honestly, I I was
3:45 worried that you might try to claim
3:46 being right points here because this is
3:49 ultimately the cost of OpenAI waiting
3:52 two years to deploy ads and now you have
3:54 these wellunded competitors who are
3:56 happy to smear you uh in front of 100
3:59 million people or however many people
4:00 are going to be watching the Super Bowl
4:01 on Sunday.
4:02 But ultimately, it's just
4:04 kind of grading like Anthropic is kind
4:07 of echoing the Apple campaign to
4:10 demonize personalized ads and smearing
4:13 the good name of advertising overall.
4:16 And I don't know, it just sort of rings
4:18 hollow.
4:20 If you don't like advertising, why would
4:22 you like someone lying?
4:23 Like I think that that's that like it really bothers
4:25 me.
4:26 Number two, everyone is criticizing
4:28 Sam Alman's response on Twitter, which
4:30 is super long and ongoing.
4:32 and is [laughter]
4:34 let me add my voice to the
4:36 when OpenAI as we've discussed is way
4:38 too online uh for sure.
4:40 Um
4:43 the I have a big problem with the first
4:44 line.
4:45 I don't think the ads were funny
4:46 and I didn't laugh because I think
4:47 they're liars.
4:48 I actually thought the
4:49 rest of his message was pretty spot-on
4:52 and pretty well articulated.
4:54 So number one he's like it's dishonest.
4:57 Yeah.
4:57 Yeah.
4:57 And then he's like,
4:59 "Ananthropic's kind of dishonest all the
5:00 time." Which I actually agree with.
5:02 I think
5:03 that was the one banger.
5:04 What he said specifically, "It's on brand for
5:06 anthropic double speak to use a
5:08 deceptive ad to critique to critique
5:11 theoretical deceptive ads that aren't
5:14 real."
5:15 Actually, Super Bowl ad not where I
5:16 expect to be.
5:17 Yeah.
5:17 Uh spot on like like
5:19 you and also something I think is going
5:22 to resonate uh there's
5:25 you go back to a lot of the chip control
5:27 stuff a lot of the things that's
5:28 happened in Washington the Biden AI act
5:31 a lot of people in the industry place
5:34 that at the feet of anthropic spinning
5:36 these tall tales and
5:39 seeking to secure sort of a certain
5:40 position.
5:41 Um, and so yeah, so I so I
5:45 agree with that.
5:46 And then he makes the
5:46 point which is anthropics for rich
5:48 people.
5:48 You have to pay a lot of money
5:49 to use it.
5:50 We're trying to serve the
5:51 whole world, which is the most
5:53 compelling argument for ads.
5:54 It's it's a it's a good argument.
5:56 And then yeah, the
5:57 authoritarian company trying to get
5:58 their own road like I the
6:00 not wrong.
6:01 I don't think that's going to
6:02 resonate with many people who are, you
6:04 know, normal consumer.
6:05 hard to have a
6:06 thousand word break through compared to
6:08 a uh quote unquote funny ad.
6:11 That's lying.
6:12 But I think he's right here.
6:14 And I think it was actually important
6:17 for him to say it because who is
6:18 Anthropic actually advertising to?
6:21 Who's that?
6:23 Potential employees, research scientists, people who are like, I
6:27 didn't sign up to work for an ad.
6:29 This is always this is the been the biggest
6:30 challenge for open AI is becoming a no
6:33 one signed up to be a consumer tech
6:34 company and I think so much of the
6:36 challenges and issues over the last few
6:38 years was realizing actually
6:41 we're being pulled into this whether we
6:43 like it or not it's a huge opportunity I
6:45 think this part drove their reluctance
6:46 to get into advertising because that
6:48 wasn't it's not what they wanted to do
6:49 that's not what they signed up for
6:51 that is super interesting because that
6:53 was my question watching these ads I was
6:55 wondering like does Claude really think
6:58 that Does Enanthropic really think that
7:00 they're going to be able to go after the
7:02 massive consumer base that OpenAI has
7:04 here?
7:05 Yeah.
7:05 No, they to the extent they're
7:07 they're causing people to rebell it ads
7:09 is going to go to Gemini's, you know,
7:12 Exactly.
7:12 Yeah.
7:12 totally benefit, I would
7:13 think.
7:14 Well, and ironically, uh, the campaign
7:16 was orchestrated by somebody who came
7:19 over from Meta, the most successful
7:21 advertising company in the history of
7:23 the world here.
7:24 But, you know, go with
7:26 whatever you got as if you're anthropic.
7:28 I suppose, but you're ultimately
7:30 advertising to open.
7:32 And so, so when Sam Alman writes this
7:33 long response, that's who he's writing
7:35 to.
7:35 He's saying like, "Look, we're
7:36 actually serving the whole world.
7:38 We want AI for everyone." Um, that's, you
7:42 know, I I was amused the moment the
7:44 commercial dropped, I got approximately
7:46 4,000 tweets.
7:47 Um, that's a slight
7:48 exaggeration, referencing my 2013
7:50 strategy credit article.
7:52 But actually, probably the greatest moniker I ever
7:53 came up with, better than aggregation
7:55 Theory.
7:55 The strategy credit is when you
7:57 get to brag about something that is
7:59 indemic to your business model and you
8:01 get a look like like so Apple is the
8:03 canonical example being like anti-ad and
8:05 pro privacy.
8:06 It's like you're selling
8:07 devices.
8:07 You're not an advertising
8:08 company.
8:08 It's very easy for you to say
8:10 those things.
8:10 This is a good example.
8:11 They lost in the consumer space.
8:13 They're never going to build an ad business.
8:14 It's never going to happen for them.
8:16 So running a commercial anti- ads is pure
8:19 strategy credit.
8:20 You get to do it for
8:20 free.
8:21 And uh and I again I think the the
8:23 battle here is for the hearts and minds
8:25 of people working at these companies.
8:27 Uh and I'm I I'm basically 100% on an open
8:32 eyes side uh battle.
8:34 Yeah.
8:35 Well, I I agree with you.
8:36 And not because I support advertising.
8:37 I maybe because I support advertising.
8:40 That that's obviously a debate in
8:42 general.
8:42 It's funny.
8:42 Aaron Aaron called you an
8:44 advertising adv evangelist.
8:46 The internet's foremost advertising
8:47 evangelist.
8:48 You might be in fact the
8:50 only advertising evangelist.
8:52 I think Eric
8:53 Seford.
8:53 Yeah, Eric Seuford is the is the
8:55 [laughter] internet's uh foremost
8:56 advertising evangelist.
8:57 Uh so
8:59 you know uh I agree with where you
9:01 landed on all that.
9:02 The ads I do
9:03 disagree.
9:04 The ads are all very funny.
9:06 And they also include Dr.
9:07 Dre's What's the Difference?
9:09 Uh really going hard
9:10 after the upwardly mobile millennial who
9:12 wishes it was still the late 90s and
9:15 it's working.
9:16 Um I enjoyed Dr.
9:17 Dre's inclusion and all that.
9:19 But I think the
9:20 OpenAI messaging it was too long.
9:23 It does resonate with me though.
9:25 They are delivering a product to way more people
9:28 than Claude and to do that effectively
9:30 and get the best version of their
9:32 product to the most people.
9:34 They need a different business model and I I hope
9:37 that they succeed.
9:37 OpenI is in many
9:38 respects their biggest problem is they
9:40 have too many customers.
9:43 Mhm.
9:42 Like like that they have more way more
9:45 compute than anthropic but they they're
9:48 serving all the customers.
9:49 That's the issue, right?
9:50 And so you when you hear
9:51 about internal tension about
9:54 re you know resources and different
9:56 teams competing for all that sort of
9:57 thing.
9:58 That's downstream of having so
9:59 many customers like like so that that's
10:03 the problem of them not launching ads is
10:05 that this is a problem of their own.
10:08 You need to be monetizing the fact you
10:10 have a lot of customers.
10:11 You can't just serve lots of customers endlessly.
10:15 Yeah.
10:14 I do worry it's too late.
10:16 I do worry that Gemini is actually the big
10:18 beneficiary of this like for their sake.
10:20 But
10:20 they may be the beneficiary of these
10:22 ads,
10:23 right?
10:23 No.
10:23 Exactly.
10:24 Exactly.
10:24 Um
10:25 and meanwhile Google Google is right
10:27 there.
10:27 Maybe maybe Google is the most
10:29 successful advertising company in the
10:30 history of the world.
10:31 Uh but they're right there with Meta.
10:33 Yeah.
10:33 Yeah.
10:34 They are funding Gemini with advertising
10:37 money.
10:37 Google is I think the most successful.
10:39 Uh I I just like meta ads better.
10:41 Uh but yes, Google, of course.
10:43 Of course.
10:43 No question.
10:44 Um, well, to keep it moving,
10:47 uh, a tweet