Lenny was highly successful in both wasting the time and annoying the scammers. But the latest AI bots are a totally different ballgame. Lenny just randomly repeated pre recorded snippets. AI bots can engage the scammer much better.
The logical conclusion is AI bots exclusively talking to AI bots to complete the circle. It would be funny if it wasn't for all the wasted computing cycles that could be burned for something actually useful instead.
If they move it to optical compute then we won't have to keep talking so much about the CO2 and compute cycles.
I imagine they're having conversations to do just that pretty often now, and hopefully with some action towards getting it done.
? Optical computing still consumes energy, you can't beat thermodynamics like that.
(there is some theoretical research into "reversible computing", but it's not really practical. A side effect is that destroying information is not reversible: so instead of having a two-input one output gate, you have to have 2 inputs 2 outputs and keep the "spare" around to make it reversible.)
That’s how the whole “life” thing works. I don’t think there’s a viable escape route from that, until we invent at-scale negotiation protocols better than arms and market.
Billboards ARE banned in some states, e.g., Vermont.
It is immediately noticeably nicer to drive there vs neighboring states.
It also generates a few fun stories, such as the grandfathered-still-existing tall sign north of Brattleboro that said "Steak $5.99" for years after the steak house had changed. Eventually, while it was being leased to an Asian restaurant, people protested that it was false advertising (as steak had not been available for years and certainly not at that price). The landlord still wanting to keep the sign, but not allowed to change it, removed the "S", "K", and "$5.99", so the sign read "TEA" for many more years, and AFAIK, still does.
If you don't have the right to hold up a sign saying whatever you want on land that you own, it's a serious limitation on your free speech. So while I think it's reasonable to place safety and aesthetic restrictions on advertising, it's worth balancing those restrictions against free speech ideals.
Man, am I glad that the UK has the Advertising Standards Authority ensuring I don't have to deal with an unlimited stream of fraud. Oh and there's essentially zero political advertising on TV, it's all submarined into "news" programmes.
(I would possibly make an exception for cases where people want to put their speech on their land, but the moment it's a corporation paying someone to put up an advert? That's not individual speech any more, is it?)
You don't need to have free speech right to be paid for someone else's board though to protect that.
We can make it really tedious to advertise by forcing them to buy the land itself and deal with title transfer, without impacting free speech of people.
I may be reading too much into this, but I think the suggestion here is that the "tele" prefix means "at a distance" (tele-phone = distant voice, tele-graph, tele-port, tele-vision, tele-pathy, tele-kenisis, etc.)
You don’t call telemarketers or solicit their services. That is quite a bit different than seeking out a websites content and circumventing the method that content provider chose to generate income.
A web page which integrates ads from external ad networks is just as intrusive though. There's actually nothing wrong with curated self-hosted ads (which would be similar to traditional print magazine ads), but just opening up a part of your webpage to let a 3rd-party run potentially malicious code on my device is borderline criminal. From that perspective, an ad-blocker should be pretty much mandatory for security reasons and should be a standard feature of every browser, just like popup blockers are now expected.
I don’t care if you use an ad blocker, but that is most certainly allowing you to view content you want without providing the intended revenue to the provider. That is just the reality.
I'm aware of that reality, but tbh, I don't care either. It's a business model decision by that content provider, and if that business model no longer works (because for instance too many people install an ad-blocker) it's their problem to find a different business model that works under that new reality (just as they were forced to find a new business model when popup blockers became mainstream). It's not that there are no options (for instance a subscription paywall or a 'pay-per-view' solution, that would also allow to judge whether the content quality is actually worth anything to anybody).
You maintain a connection to the telephone network, which the telemarketers utilize in an entirely uncontroversial way to speak with you. "Using the web" is like "having a telephone", in that I can select what I wish to experience, and screen out that which I don't.
Telemarketers have chosen to architect their business on the predicate that people trustingly pick up their phones and interact with those who call them. That social norm is fading; people screen their calls, or even use eg that Pixel phone thing in which a robot picks up, confronts the telemarketer, and texts the user.
The DOM doesn't live out there somewhere, it lives inside my browser. Being choosy about which parts actually get displayed, as opposed to dropped on the ground, isn't taking food out of people's mouths, it's my failing to adhere to a social norm of naively rendering a webpage exactly as it's sent to me. It reminds me of psuedo-moral panic about WFH coring out once-lively commercial real estate and all its downstream consequences (eg, the absence of cubicle drones needing to go out to eat at lunch hour, etc) -- who owes who a living, and why?
I've got to say the call in the video is not very convincing, but the idea is pretty cool. Let's have AI and scammers talk to each other and waste each other's time while we do more interesting stuff.
as a telemarketer some 25 years ago, perchance I talked to your friend's dad.
Although, he had said that he had just downed a handful of viagra and I was running the clock out on that so he really, really had to get off... the phone.
Telemarketing was one of my first jobs between high school and college. I hated the job and I was fired because I couldn't sell anything to anyone. Area jobs were very slim and I didn't have much experience, and it paid considerably better than mcdonalds - I apologize to World+Dog for being that obnoxious nuisance. I landed a job in IT shortly therafter and never looked back.
It does? That implication never occurred to me at all. I think it implies nothing about the nature of the relationship at all. It's just a way to avoid saying "fuck".
Though I should point out that some dictionaries will list the "just another word for sex" definition either as a colloquialism or a secondary definition.
Basically bitcoin trajectory burning fuel to do useless hard math just so that you can't compete. I see same thing happening with AI chat which will pitch AI against AI until one party can't keep up all while burning fuel doing useless math.
I think in the short term its cheaper for these scam call centres to use "local" English speakers (in practice many of the employees are fresh-faced out-of-towners looking for their first job) than it is to dev and maintain an entire GPT pipeline. It will require a level of completely different organisation and sophistication than what is currently in place. Right now outside of running the dialler and internal off-the-shelf packages to manage their workforce these orgs are not particularly tech literate as evidenced by the YouTube channels that expose them.
For "legitimate" (the kind that's not a scam, just annoying) they are regulated that it wouldn't be legal in many jurisdictions. (at least in B2C context -- maybe it a bit lax in B2B, considering Google's already using some form of bot to confirm business hours for Google Maps listing.)
Scammers in other hand already employ illegal robocalling tactic, and I can see some may opt to use systems like that if the price is right.
there's a bit where it says: "I just won a talent contest, do you think your insurance product would help cover a party that I'd like to throw for me" which is pretty decent.
Personally, I take bigger issue with the people who hire the telemarketer. While I can imagine a person revelling in their telemarketing position, somehow I suspect most people working are just trying to get enough money to not starve before next months paycheck.
Regardless of the reality, I get the sentiment. Hey we built this system that can trap humans in an infinite social loop requiring either superhuman willpower or outright sociopathy to escape. But don't worry, we only use it on the bad people.
Part of the justification I’ve heard for wasting the scammers time is to limit the number of people they can interact with in a day. If wasting a scammer’s time prevents another innocent person from scammed then there’s a net benefit.
It’s unfortunate that some people need to do bad things to make a living but it’s an explanation of bad behavior, not a justification for it.
Mind you, I’m also sympathetic to the ways in which this practice could be abused.
>infinite social loop requiring either superhuman willpower or outright sociopathy to escape
Or, you can just hang up if you have your suspicions. I don't see this kind of banning particularly useful. If it gets sufficiently popular, telemarketers will just work around it in their scripts.
Heaven Banning is hypothetical, it's not real. At least not yet.
If you look closer at the date in the screen shot of a NYT article in your link, you'll see that the date was deliberately chosen to be in the future (i.e. August 8, 2024).
"Heavenbanning" isn't going to be a thing, because it's no more effective than hellbanning while being drastically more expensive for platforms to implement.
The article describes it working though. It was able to string along some salesman for 5 minutes, much longer than he would wait if someone picked up the phone and said nothing.
The only reason it works in a telemarketing context is because "hellbanning" doesn't exist for phones, it's a synchronous medium where you need to keep them on the line somehow, as you note. That's not a concern for asynchronous mediums like forums.
In addition, banning someone in the context of a forum is the responsibility of the operators of the platform, but stringing along a telemarketer is the responsibility of the individual users of the platform. It's not equivalent to a hellban, it's the equivalent of trolling, but now automated. If it ever started happening in large numbers, the platform operators (the phone companies) would change their terms of service to make it illegal, because having machines trolling each other indefinitely uses up their network capacity for no reason.
Regarding information safety on the Internet, I think the cat is long out of the bag now. Forget AI, information is already weaponized at scale. I mean in the way like in this Wiki collection [0]. AI is, while probably powerful, just another tool in this toolbox.
To preface it's a strong gut feeling that I don't know where exactly it is coming from. So I'm trying to understand it myself.
I think the reason I feel so opposed to it is because human connection is one of the most beautiful things. Forming friendships. And heaven banning counter-acts that and counterfeits that. If you find out your friends were fake, in a very literal sense, enough times you may stop trying to find friends. You may fall into a learned helplessness and isolation, where you must scream but there is no real to hear. Only fake phantoms feigning sympathy, gone tomorrow and replaced with new faces.
Thanks, I think I understand. I think the people in the world could do with more compassion. To stay with the example of telemarketers, my issue is that whoever benefits from the operation is not the one bearing the consequences, the externalities of the deed - I mean, they are not the ones yelled at by the desperate people on the other end of the line. So the operators effectively outsource this misery to the pawns that are the telemarketers. We as a society do this thing often, with everything we don't want to face directly, like the butchering of the animals that we consume.
The proper way to go about it would be regulation, I think. For example, to ban telemarketing altogether.
All subscriptions are billed on an annual basis, but you may cancel a subscription at any time. If you cancel within the first 30 days of signup, you will not yet have been charged, and the pending charge to occur at the end of the trial will be cancelled. In other words, the first 30 days after signup are considered a free trial period and there is no charge for the free trial period.
Once your paid subscription period begins, if you cancel within the first 14 days of this period, you will receive a full refund. Neither full or partial refunds will be provided after 14 days from the date of payment. To cancel your subscription, click on the “Cancel Subscription” link on the “My Numbers” page of the Pirate Porthole. Please allow 30 days for your refund to be processed.
Pretty sure the CFPB says this is not an OK practice. Maybe it's a scam in itself? Stir up hype, acquire your 80% of subs in the first 45 days. Pull the plug. No refund.
I think as this technology gets cheaper and better, another, more insidious application is using it to overwhelm support channels.
Fake or false alarm 911 calls are a problem today. Imagine someone "attacking" the 911 system by spamming it with fake calls.
Or calling up / spamming/ etc. help lines for e.g. returning defective merchandise, clogging up support agents and costing money.
Spamming your local coffee shop because one morning you felt they were rude. Leaving 1 star reviews everywhere, but over time, so it's harder to detect.
Placing tons of orders and cancelling them before being charged, hopefully leading to wasted effort.
We may need a CloudFlare for everything, also powered by the same technology
Every interaction that may be abused and spammed should either require payment or require strict identification of a user that it is this particular human being with permission to perform this action. In this case any spam attack can be countervailed either by rising a price for interaction or by withdrawing permissions.
Sometimes that's not possible or leads to less ideal tradeoffs (at least in the short run)
For example, I made a doctor's appointment recently. To do so, I called, told them my name, and a few other details that are mostly in the realm of public information.
No one in the office had met me. They had no ID. That would come later in some other forms to be filled out before the visit.
If they require ID upfront, that friction will cause some appointments to not be made. You can't find your ID while on the phone. Grandma has an expired driver's license.
On the other hand, requiring it upfront reduces fraud.
Eventually we will probably have AI powered agents that perform these tasks for us, have the relevant documents handy in memory, and ultimately reduce fraud when scheduling an appointment.
But then we need defense against fake appointments being a way to commit identify fraud.
Before such agents are widely available, there will be growing pains
Luckily we don't need to construct solution out of the most recent hype, we can pick previous instead.
Imagine convenient crypto wallet app that you always has with you on your phone. And you have two options to make an appointment:
* Pay upfront total cost of an appointment. You receive NFT that indicates you appointment. To cancel appointment just send it back to receive your payment depreciated according to the time you hold the appointment NFT.
* Or you may validate your identity by logging in to the clinic website with your crypto wallet. If you hold enough identity proving NFTs issued by various government agencies, you are eligible to make an appointment.
I feel like half of the required pieces already exist. We just need to connect the dots.
Running with the example of 911 from the grandparent post, the service needs to be as low-friction as possible when people call. For them it is a matter of life or death.
Imagine if you or I had to make a payment, or solve a captcha, or confirm our identity in order to make an emergency call. That is simply unacceptable in this case.
Most of the phones are used with SIM cards and have non zero balance. For instance we might make a law that the last $5 from cell phone subscription plan can be spent only on emergency calls, nothing else. This will make sure that any phone in active use will have enough balance to make couple of emergency calls.
Proving identity can be as easy as scanning your fingerprint twice to unlock the phone and then confirm identity proving operation in crypto wallet.
And we can still process calls to 911 without identity or payment but only if we have enough capacity to do so.
> Imagine someone "attacking" the 911 system by spamming it with fake calls.
Imagine?
Hospitals and universities in my area are getting spammed with bomb threats by idiots over "anti-white discrimination" and right-wing conspiracies about young children getting gender-change operations.
> We may need a CloudFlare for everything, also powered by the same technology
If a wordpress install I run is any indication, Cloudflare is pretty useless for preventing bot activity. Not only do I have sensitivity turned up, I've got everywhere outside the US geo-blocked. Tons of bots sailing right through.
It is not GPT-anything. The Register has been laughably duped here, just like the telemarketers.
The "robots" do not listen to the scammer at all. Basically the algorithm just waits for pauses from the other side and plays random audio. Often just "uh huh", and "keep talking", and "can you repeat that, I wasn't paying attention" and things like that, plus random chatter. The core of it is probably some 50 line script.
The responses don't use anything from the telemarketer's utterances; it's dumber than Eliza.
93 comments
[ 0.23 ms ] story [ 2698 ms ] threadhttps://news.ycombinator.com/item?id=36549299
And a couple days before that... https://news.ycombinator.com/item?id=36523774
The state of the art 10 years ago is hilarious: https://www.google.com/search?client=safari&hl=en-us&q=lenny...
(there is some theoretical research into "reversible computing", but it's not really practical. A side effect is that destroying information is not reversible: so instead of having a two-input one output gate, you have to have 2 inputs 2 outputs and keep the "spare" around to make it reversible.)
"...in a single year, São Paulo removed 15,000 billboards, many of which were replaced by street art."
https://www.theguardian.com/cities/2015/aug/11/can-cities-ki...
It is immediately noticeably nicer to drive there vs neighboring states.
It also generates a few fun stories, such as the grandfathered-still-existing tall sign north of Brattleboro that said "Steak $5.99" for years after the steak house had changed. Eventually, while it was being leased to an Asian restaurant, people protested that it was false advertising (as steak had not been available for years and certainly not at that price). The landlord still wanting to keep the sign, but not allowed to change it, removed the "S", "K", and "$5.99", so the sign read "TEA" for many more years, and AFAIK, still does.
(I would possibly make an exception for cases where people want to put their speech on their land, but the moment it's a corporation paying someone to put up an advert? That's not individual speech any more, is it?)
We can make it really tedious to advertise by forcing them to buy the land itself and deal with title transfer, without impacting free speech of people.
There’s not much telephone involved.
Telemarketing refers, both entymologically and colloquially, to marketing via telephone. No one calls billboards or online ads telemarketing.
Lots of HNer talking pass each other because they won't acknowledge they don't care.
Telemarketers have chosen to architect their business on the predicate that people trustingly pick up their phones and interact with those who call them. That social norm is fading; people screen their calls, or even use eg that Pixel phone thing in which a robot picks up, confronts the telemarketer, and texts the user.
The DOM doesn't live out there somewhere, it lives inside my browser. Being choosy about which parts actually get displayed, as opposed to dropped on the ground, isn't taking food out of people's mouths, it's my failing to adhere to a social norm of naively rendering a webpage exactly as it's sent to me. It reminds me of psuedo-moral panic about WFH coring out once-lively commercial real estate and all its downstream consequences (eg, the absence of cubicle drones needing to go out to eat at lunch hour, etc) -- who owes who a living, and why?
I remember Captain Picard etc. but I want to do it at scale
Could essentially DDoS the scammers by sending realistic seeming replies from all their spam email.
I've got to say the call in the video is not very convincing, but the idea is pretty cool. Let's have AI and scammers talk to each other and waste each other's time while we do more interesting stuff.
Although, he had said that he had just downed a handful of viagra and I was running the clock out on that so he really, really had to get off... the phone.
Telemarketing was one of my first jobs between high school and college. I hated the job and I was fired because I couldn't sell anything to anyone. Area jobs were very slim and I didn't have much experience, and it paid considerably better than mcdonalds - I apologize to World+Dog for being that obnoxious nuisance. I landed a job in IT shortly therafter and never looked back.
Though I should point out that some dictionaries will list the "just another word for sex" definition either as a colloquialism or a secondary definition.
Note that e.g. M-W does not list this secondary definition: https://www.merriam-webster.com/dictionary/fornication
Scammers in other hand already employ illegal robocalling tactic, and I can see some may opt to use systems like that if the price is right.
In the old days I would just get some garbage with max one sentence decorated with multiple exclamation marks. Here I got 500-word essays.
[0] https://archive.is/hPoAo
Regardless of the reality, I get the sentiment. Hey we built this system that can trap humans in an infinite social loop requiring either superhuman willpower or outright sociopathy to escape. But don't worry, we only use it on the bad people.
It’s unfortunate that some people need to do bad things to make a living but it’s an explanation of bad behavior, not a justification for it.
Mind you, I’m also sympathetic to the ways in which this practice could be abused.
Or, you can just hang up if you have your suspicions. I don't see this kind of banning particularly useful. If it gets sufficiently popular, telemarketers will just work around it in their scripts.
If you look closer at the date in the screen shot of a NYT article in your link, you'll see that the date was deliberately chosen to be in the future (i.e. August 8, 2024).
In addition, banning someone in the context of a forum is the responsibility of the operators of the platform, but stringing along a telemarketer is the responsibility of the individual users of the platform. It's not equivalent to a hellban, it's the equivalent of trolling, but now automated. If it ever started happening in large numbers, the platform operators (the phone companies) would change their terms of service to make it illegal, because having machines trolling each other indefinitely uses up their network capacity for no reason.
Regarding information safety on the Internet, I think the cat is long out of the bag now. Forget AI, information is already weaponized at scale. I mean in the way like in this Wiki collection [0]. AI is, while probably powerful, just another tool in this toolbox.
[0] https://en.wikipedia.org/wiki/Russian_web_brigades#Opinion-i...
I think the reason I feel so opposed to it is because human connection is one of the most beautiful things. Forming friendships. And heaven banning counter-acts that and counterfeits that. If you find out your friends were fake, in a very literal sense, enough times you may stop trying to find friends. You may fall into a learned helplessness and isolation, where you must scream but there is no real to hear. Only fake phantoms feigning sympathy, gone tomorrow and replaced with new faces.
The proper way to go about it would be regulation, I think. For example, to ban telemarketing altogether.
Then, click farms may disappear.
Once your paid subscription period begins, if you cancel within the first 14 days of this period, you will receive a full refund. Neither full or partial refunds will be provided after 14 days from the date of payment. To cancel your subscription, click on the “Cancel Subscription” link on the “My Numbers” page of the Pirate Porthole. Please allow 30 days for your refund to be processed.
Pretty sure the CFPB says this is not an OK practice. Maybe it's a scam in itself? Stir up hype, acquire your 80% of subs in the first 45 days. Pull the plug. No refund.
https://www.nytimes.com/2016/02/25/fashion/a-robot-that-has-...
Fake or false alarm 911 calls are a problem today. Imagine someone "attacking" the 911 system by spamming it with fake calls.
Or calling up / spamming/ etc. help lines for e.g. returning defective merchandise, clogging up support agents and costing money.
Spamming your local coffee shop because one morning you felt they were rude. Leaving 1 star reviews everywhere, but over time, so it's harder to detect.
Placing tons of orders and cancelling them before being charged, hopefully leading to wasted effort.
We may need a CloudFlare for everything, also powered by the same technology
For example, I made a doctor's appointment recently. To do so, I called, told them my name, and a few other details that are mostly in the realm of public information.
No one in the office had met me. They had no ID. That would come later in some other forms to be filled out before the visit.
If they require ID upfront, that friction will cause some appointments to not be made. You can't find your ID while on the phone. Grandma has an expired driver's license.
On the other hand, requiring it upfront reduces fraud.
Eventually we will probably have AI powered agents that perform these tasks for us, have the relevant documents handy in memory, and ultimately reduce fraud when scheduling an appointment.
But then we need defense against fake appointments being a way to commit identify fraud.
Before such agents are widely available, there will be growing pains
Luckily we don't need to construct solution out of the most recent hype, we can pick previous instead.
Imagine convenient crypto wallet app that you always has with you on your phone. And you have two options to make an appointment: * Pay upfront total cost of an appointment. You receive NFT that indicates you appointment. To cancel appointment just send it back to receive your payment depreciated according to the time you hold the appointment NFT. * Or you may validate your identity by logging in to the clinic website with your crypto wallet. If you hold enough identity proving NFTs issued by various government agencies, you are eligible to make an appointment.
I feel like half of the required pieces already exist. We just need to connect the dots.
Running with the example of 911 from the grandparent post, the service needs to be as low-friction as possible when people call. For them it is a matter of life or death.
Imagine if you or I had to make a payment, or solve a captcha, or confirm our identity in order to make an emergency call. That is simply unacceptable in this case.
Most of the phones are used with SIM cards and have non zero balance. For instance we might make a law that the last $5 from cell phone subscription plan can be spent only on emergency calls, nothing else. This will make sure that any phone in active use will have enough balance to make couple of emergency calls.
Proving identity can be as easy as scanning your fingerprint twice to unlock the phone and then confirm identity proving operation in crypto wallet.
And we can still process calls to 911 without identity or payment but only if we have enough capacity to do so.
Imagine?
Hospitals and universities in my area are getting spammed with bomb threats by idiots over "anti-white discrimination" and right-wing conspiracies about young children getting gender-change operations.
> We may need a CloudFlare for everything, also powered by the same technology
If a wordpress install I run is any indication, Cloudflare is pretty useless for preventing bot activity. Not only do I have sensitivity turned up, I've got everywhere outside the US geo-blocked. Tons of bots sailing right through.
https://gridpane.com/blog/cloudflare-firewall-rules-for-secu...
Additionally, cloudflare automatically protects with the owasp ruleset:
https://developers.cloudflare.com/waf/managed-rules/referenc...
The "robots" do not listen to the scammer at all. Basically the algorithm just waits for pauses from the other side and plays random audio. Often just "uh huh", and "keep talking", and "can you repeat that, I wasn't paying attention" and things like that, plus random chatter. The core of it is probably some 50 line script.
The responses don't use anything from the telemarketer's utterances; it's dumber than Eliza.
Yet it captivates them, keeping them talking!