Hacker News new | ask | show | jobs
by fishtoaster 1768 days ago
This seems at least a little unethical. At the very least, I would think worse of a company that sent a video to me that seemed personalized, but later turned out to be mass-produced. He has an ethics section in the article, but doesn't doesn't go very deep and (unsurprisingly) concludes that his own actions were perfectly fine.

The root of the problem is why customers respond well to these videos: they're surprised and impressed that the CEO of the company would take time out of his day to record a thank you video just for them. And the fact is: the CEO did not. It's deception, plain and simple. Not big, evil, Theranos-level deception, but deception still.

To put it another way, how do you think customers would respond if, along with each email, you included some text saying "this video was automatically generated"? I'd assume pretty poorly. The difference between that response and the actual response is the value of the deception here. I think that "if we were honest about this, customers wouldn't like it" is a good litmus test for whether you're doing an ethical thing or not.

Really, I think this is doing a sort of trust arbitrage. The tech exists to fake personal videos now, but customers aren't widely aware of it. You can take advantage of that difference with stunts like this. In a few years, people will know this can easily be done and the value of personal videos will drop. At some point it'll be no different to the average customer than an email with "Hi <name here>!" or a letter with faux-handwriting on it.

11 comments

> And the fact is: the CEO did not. It's deception, plain and simple. Not big, evil, Theranos-level deception, but deception still.

If you think the blog post is bad, just watch the video on windsor's website: https://www.windsor.io/#video

The CEO plainly says he has time to work on other important things. Jesus, even creating the video in the first place, requires you to say Hey First Name

That is so cringe. It's shit like this that will really make the masses hate tech companies. Years and years of education, tons and tons of computational resources, engineer months, years, cloud storage, MASSIVE SCALE, billion-dollar valuations (!?) and it's a crummy Ovaltine commercial with a clumsily edited in "Hey Ralphie!"
Actually, it kind of reminds me of those silly commercials you hear on the radio for some real estate seminar... "The Bay Area is a perfect place for our system," or something like that. The cadence is off just enough to notice and kick it over into uncanny valley territory.
That video is one of the scariest things I've witnessed in less than 60 seconds. The future of communication will be opt-in, I suppose.
I don't find it that scary. It's a gimmick. It will maybe work once or twice, but how many of these "hey firstname" videos do you think people will watch before catching on?

Sure, it's going to improve a lot, but since the first iteration of automated, personalized video messaging is so low-effort, I actually think people will just be inoculated against the whole idea. As in, don't watch the videos at all.

Hey $HN_NICKNAME,

Your opinion is very important to me and I would like to take a moment to tell you personally how much I care about whatever you say.

I'd be available for a call between 1am and 11pm if you want to discuss one amazing opportunity.

Lisa - VP of Deceptive Marketing.

Attn: rebuilder

Hey rebuilder,

Sorry for spelling up your name incorrectly, we probably had a computer problem over here.

I have recorded a video just for you to apologize.

https://share.synthesia.io/e07bf638-1385-4c92-9970-940a0ebe7...

Lisa - VP of mass-emailing.

>Really, I think this is doing a sort of trust arbitrage. The tech exists to fake personal videos now, but customers aren't widely aware of it. You can take advantage of that difference with stunts like this. In a few years, people will know this can easily be done and the value of personal videos will drop. At some point it'll be no different to the average customer than an email with "Hi <name here>!" or a letter with faux-handwriting on it.

This was my exact thought as I read this article. The other part to this is that it comes at a societal cost: it will add another layer of distrust and cynicism in how we communicate, all at the expense of a short-term value to a handful of companies that race to maximize the value of this technology over the coming years.

Agreed.

That said, I would argue that this specific practice will become more ethical over time (as the arbitrage opportunity disappears). Once customers are aware of the practice, the trust difference fades, and the deception goes away. 10 years from now, this practice will be equivalent to claiming "Famous Fishtoaster's has the best fish toast in the world": it's a lie, but one that all parties know is false so no one is being deceived. Just like no one really believes personalized text in an email implies that that email was handwritten anymore, no one will believe these videos are personalized and the problem goes away.

But yeah, it will be a societal cost: one less thing we can trust. Just like "photos don't lie" went out the window with widespread photoshopping, so will trust in personalized videos.

> But yeah, it will be a societal cost: one less thing we can trust. Just like "photos don't lie" went out the window with widespread photoshopping, so will trust in personalized videos.

But video is the final frontier of digital trust. Once photos, video, and audio can no longer be trusted, what will people trust?

Digital cynicism will become total.

I was just re-watching "Westworld" recently, and there's a moment when Aaron Paul's character Caleb is having a telephone conversation with a sympathetic HR employee who tells him that he didn't get the job.

"Listen, Caleb. Your application was very strong. Unfortunately, our strategy group just hasn't found an opening for you."

Caleb sighs, and says nothing.

The HR employee gently asks, "Caleb, are you still with me?"

"Ok, thank you. Is there anything I should be working on to make myself a better candidate?"

"Like I said, your application was very strong. We just don't have anything that would be a great fit for you right now."

"If I'm not a good fit, is there a different shape I could squeeze myself into?"

There is a pause.

Caleb asks, "Hey, no offense but... are you human?"

"I'm Shaun," the voice says. "I can help you with all kinds of resources... anything else I can do for you today, Caleb?"

This feels a lot like that.

I recently received a telemarketing robocall that had a convincingly human voice.

It did polite social chitchat quite well.

But there were a few tell-tale signs: the bot would never interrupt or overlap my talking. It would always wait until I finished speaking to parse my speech, and its pauses were very consistent and slightly longer than what was natural.

When I asked it if it were a robot, it replied, "(giggle) why, do I sound like a robot?"

When I threw it a few unconventional questions, it quickly became clear that there was a script.

Well said, especially this line:

> I think that "if we were honest about this, customers wouldn't like it" is a good litmus test for whether you're doing an ethical thing or not.

I used Deepfakes to trick 10k people into thinking I cared
Hey $HN_NICKNAME,

Your opinion is very important to me and I would like to take a moment to tell you personally how much I care about whatever you say.

I'd be available for a call between 1am and 11pm if you want to discuss one amazing opportunity.

Lisa - VP of Deceptive Marketing.

Re: URGENT

Hello $HN_NICKNAME,

Are you still reading HackerNews ? I have this amazing opportunity, lifetime opportunity.

It is the end of Q2, and my boss marshallhaas.com has agreed to make an exceptional discount just for you.

Are you available for a call tonight between 4 am and 6 am ?

Lisa.

Mostly this is just creepy and weird. I bet all those text blocks from people responding positively to it are from people who don't have any practice or previous context experience in distinguishing AI/generated content from authentic content.
As with all new ad technology this works incredibly well. However the ethics and morals of this is quite questionable.

Recipients take time to respond to what is in essence a spam email. If anyone sends this stuff to me, they'll go on the block list - but if it gets widespread enough then its the 'future' - an inbox full of 'personal' video /audio messages..

Quick someone make a spamfilter for AI videos :)

It’s Boomer bait.
Devils advocate on

Since all advertising is the practice of psychological manipulation how is this particularly different? Is all advertising unethical?

Even if the process is automated, if the end result is to evoke a nice sentiment in someone, is that really a bad thing? Wouldn't the ethics really come into play in using such tech to rip people off in a variety of ways rather than to merely say thank you for your purchase?

Devils advocate off

I think there's a strong argument to be made that the efficacy of this approach is more tied to the visual senses and novelty of it. It's not just the personalization since as you've pointed out it's no different than a personalized email.

Ultimately it's just a tool, and how people choose to use those tools is all that really matters.

A friend of mine ran a ecommerce company where for years they would physically record at least a couple hundred personalized thank you videos for people each day. While they were quite successful and people really dug them, I can tell you for a fact that such messages may as well have been automated because there's very little you actually know about your customers beyond their purchase/purchase history that you can truly personalize your message to them with.

> Even if the process is automated, if the end result is to evoke a nice sentiment in someone, is that really a bad thing?

Yes, because it only evokes a nice sentiment if you lie about how it was made. Magic tricks are still fun even if you know they're tricks. This is not at all fun once you know it's a trick.

The fun is in the novelty, not in the <insert first name>.

Even if it's inherently considered dishonest by some, the same should be vigorously argued for all email/snail mail you have received for decades and frankly I can't remember the last time I've seen someone complain about a personalized email being dishonest.

While I also recognize the valid argument that this is deceptive, I have to agree with you:

1. In no world would I ever believe that a cold email was significantly personalized/targeted, except if the overall content were highly specific to me. This might be wrong a good portion of the time, but it is what it is.

2. Even if I were 100% unaware of the existence of AI/ML, a video wouldn't change my mind about #1. I would be more inclined to think that they'd recorded themselves rattling off a list of customer names and then stitched that into the main video, or just rerecorded the same video hundreds or thousands of times. The latter would take some dedication, and be a pretty big waste of time, but it wouldn't convince me that the sender knows or cares more about me personally.

3. Including a video message in an email in itself is uncommon in and of itself, personalized or not. If the technique increases conversions, this alone could be a sufficient explanation. To the extent that the personalization does further increase conversions, it's equally valid to hypothesize that a novel tech demo catches their attention as it is to hypothesize that people are being deceived. Personally, I would lean toward the former.

As an aside, even in the promotional video on Windsor's home page, the videos are kind of uncanny valley. May not be a big deal if the recipient doesn't think too much of one word being slightly garbled because the remaining speech sounds natural, but I'd be concerned about it completely butchering some names (particularly longer ones).

Look at all the responses this guy got from his videos. Reading them, do you really think those people understood how the video was made and just enjoyed the novelty?

"I am amazed by the customer service...It makes me very happy that the team is so invested"

"I can't imagine how much effort must go into doing this"

"This is the first time in my life that I receive a personalized video from the founder of a company"

"This means a lot to me"

"I really appreciated the personal attention"

"I don't know how you have the time!"

What you see is a sub selection of positive responses intentionally picked out to paint a narrative and ultimately market the product.

I actually find the act of doing that far more manipulative than automated thank you videos.

Fair enough, I can't disagree with you there.
> Since all advertising is the practice of psychological manipulation how is this particularly different? Is all advertising unethical?

Do you mean "is psychologically manipulating people to buy shit they neither need nor want" ethical? Because that's largely how modern advertising operates.

Yes, that indeed was more or less my point.
You're throwing him under the bus because this is a pretty recent phenomenon. Before email became so "personalized" I'd never get an email mentioning me by name and if I did it was from people who I knew personally. This all changed once marketers starting using "Hey $firstname," now I have no idea if this is from someone who knows me personally or from someone has my name in a massive db.

The fact that there is no disclaimer on this video shouldn't be worse than the fact that there is no disclaimer on emails saying "this email was automatically generated."

This gives me the same warm fuzzy feeling that I'd get when I read an email with my name in it even though 99% of times I'm sure it's not really sent just to me and is auto-generated.

Over time auto-generated videos will become as common as auto generated emails.

This makes me feel the same way that those "on purpose casual" marketing emails make me feel.

"Tired of your awful insurance. Like ugh, us too! Insurance can be like, so heckin lame sometimes. Let me let you know about [INSURANCE BRAND] ... "

That stuff is absolutely psychotic and is an instant red flag. Just say "Buy this insurance, it's cheap and has features x y z w" and be done with it. I'm not your bff.

This is kind of an aside, but I feel like fast food companies have really taken over this area. Wendy's posting on twitter like "feelin heckin smol right now. normalize self care." NO! Give me McDonalds -- they show a huge picture of the burger with big text that says "Eat This Burger Now" or whatever. That's all I want from you people!

/rant

Good rant. I agree.
> In a few years, people will know this can easily be done and the value of personal videos will drop.

Or put another way, people will have to do different things (which aren't easily automated) to earn that value.

As a personal thank you, I solved these 10 recaptchas for you!
If the work was provably personalized to you (perhaps with some sort of hashing involving your name), then yep, that’s the idea!
Simply add increasingly unlikely and absurd elements to each new video: https://www.youtube.com/watch?v=tzEVRappRC0
I absolutely agree - same thing like when I get from some co-workers a reply to an email and after the body's reply I see each time the same thing (e.g. "Thank you. John"), identical to all other replies that I got previously from the same person during the last weeks => that always makes me feel a little bit sad (thinking "damn, s/he couldn't even spend literally 2 seconds to write it for real"), [sigh].
we are cyborgs, some of us are ahead of the curve.

I should dress as the Joker for Halloween

EDIT: Let me elaborate. in the 90s, remembering someones birthday with a card was meaningful because you REMEMBER their birthday and took action on it. today, people use Google Calendar to remind themselves to send that 'Happy Birthday' texts, and its not dishonest or deceptive.

In this case its taking that concept of using tech to enhance your abilities to the Nth degree.

now I have that song stuck in my head. M O, M O R