Halt! Bot or Not?

of AI, and how other people have weighed in. Some people have said that perhaps Musk and Zuckerberg are arguing about something that isn't really relevant right now, and that there are in fact other elements of artificial intelligence that we should be focusing on instead of whether or not it is certain to make our lives better or worse

or rule over us. But it got me to thinking about a related topic, and I touched on it all little bit in that episode, and that was all about how do you tell when a a an entity that is communicating with you is in fact a person or it is a computer program that is mimicking a person. So we're going to look at that, And honestly, I was inspired a lot also by the fact that we've got a new Blade Runner movie coming out. It's Blade Runner two thousand, forty nine, which has no connection to

this show. By the way, they are not sponsoring us. I'm pretty sure no one connected to Blade Runner two thousand forty nine even is aware that I exist. But I'm a fan of the original Blade Runner film and I'm looking forward to seeing what happens in Blade Runner two thousand forty nine. I'm a little hesitant because it depends upon which interpretation of the original film they decided

to ultimately go with. If they went with the director's vision, might not want to see two thousand forty nine, but I wanted to kind of talk about the difference between communicating with a person and a synthetic being. Now in Blade Runner, the synthetic beings are called replicants, and they are not exactly robots. They're often referred to as androids, but I don't really think that's terribly accurate either. They're

more like genetically engineered human simulations. Like they're not fully human. Uh. They have other elements that either augmented abilities and intelligence, but a lower lifespan that sort of stuffy. They tend to be born in the adult stage of their lives and implanted with false memories, but they're meant to do jobs that humans can't or won't do, and they do have a tendency to resent their lot in life, seeing as how in the original film they did just have

that built an expiration data just a few years. They don't live for a few years, and then they would their bodies would break down and blade Runner the story follows an investigator who is seeking out specific replicants that are on the run in order to quote unquote retire them with extreme prejudice. So this is all set up at the beginning of the movie. Now, one thing those investigators or blade Runners do is ask questions of suspects,

suspected replicants. You know, they find someone they think that might be a replicant, and then they interview that person and they look for signs that that is not actually a real human being, because replicants are not exactly human. They're human like, but they do not process emotions the same way that humans do. So blade Runners can look for indications that the suspect is actually a replicant and they use what is called the void comp test in

the movie. This is a test that includes the hypothetical situation you're in a desert walking along in the sand when all of a sudden, you look down and see it towards it. This you reach down and flip the tortoise on its back. The tortoise lays on its back. It's belly baking in the hot sun, beating its legs, trying to turn itself over, but it can't, not without your help. But you're not helping. Why is that? Now? I kind of paraphrase that scene because it actually happens

as dialogue between two characters. But that was the attempt of an interrogator to figure out whether or not the person they were talking to is actually a human being, because the emotional responses would indicate whether or not it was a human response, or if there was a lack of that, that it was perhaps a replicant. Now, that's all science fiction, but in the real world there are times when we encounter bots or AI constructs and we might not know at first, at least they were not

communicating with a real life person. In fact, the Interactive Advertising Bureau reported in two thousand fourteen that thirty six per cent of all web traffic is general to buy bots, not people, and the security firm Imperva reported in early two thousand seventeen that today that figure is now closer to fifty, which means that right now there's more traffic on the web being generated from bots than actual human beings, And that's not exactly great. Much of the web depends

upon advertising for monetization. But how do you figure out what the value of traffic to your website is when you know there's a good chance that more than half of all those page views were generated from algorithms, not from human beings. Now, much of the bot traffic isn't meant to be outright malicious. There might be bots that are essentially trying to to scour the Internet for data for nefarious purposes, but a lot of them are just

they're gathering information for, you know, completely innocent purposes. Really, gathering information on its own is not necessarily a bad thing. It's how we use the information that makes it good or bad. It's kind of paraphrasing Shakespeare there. But there are the various bots on social platforms and websites that also interact with people, and some of them are again benign. They're meant to be helpful, such as bots that can

answer basic customer service questions for companies. You've probably encountered one of these where you were looking for some information about a particular product or service and then a little chat window pops up, and you get the feeling that the entity you're talking to is not exactly another human being on the other end. It may just be a bot.

Sometimes that's fine. Sometimes it's more frustrating than helpful, because you find that you have to word things in a very particular way for the bot to comprehend what you mean, whereas a human would probably pick it up much faster. But you get the idea of why that was employed, right. That was meant to make things a little more smooth and to remove the necessity of putting a human being in charge of that at all hours of the day.

You can also find these sorts of automated services on phone lines, including bots that call you, which is always fun. There's nothing like having a conversation with a bot for half a minute before you figure out something fishy is going on. But other bots are meant to serve the purposes of some third party, sometimes with malicious intent, such as convincing you to click on a link that leads to malware, and that's where we really run into obvious problems.

Some of the benign ones can run into problems too often. There are unintended consequences if you're scouring the web for data. Data is valuable and sometimes people will want to get hold of it for bad reasons, even if the initial approach wasn't to do anything nefarious. Now, some hackers have used bots to flood a platform with complaints in an effort to silence people that the hackers do not like.

So let's say there's this jerk face hacker who thinks a Facebook page devoted to promoting women in STEM education and careers is dumb. So this jerk face then creates or more likely purchases bots to flood Facebook with complaint reports about that specific page in an attempt to get Facebook to shut the page down. Now that's a pretty lousy thing to do. And to be clear, some of the jerk faces are aiming at pages that the average

person would say is a bad one. It doesn't have to be something that like I feel strongly about and and in favor for. I think STEM education and careers for women is amazing, and I would be very upset to hear about a page that was shut down because of one of these attacks. On the other hand, let's say that there was a page that was promoting something I really do not like. Maybe it was a page that was promoting, uh, you know, racial discrimination. I would

think that was terrible. If someone else were to take bots and direct them to that page in order to shut it down, I would also think that that's not so great. I don't think that a page about racial discrimination should be promoted or exist on Facebook. I don't think that's appropriate. But at the same time, I don't think it's appropriate to use automated systems to bring that down. I would rather see an actual ground swell of human support for that, not to you know, boost it with

a bunch of automated scripts. I don't want to give the indication that the only people who ever use bots are those who want to silence vulnerable or underrepresented populations. There are some who use them to attempt to silence voices of hate. In either case, it's dirty pool. I don't think it's really a legitimate strategy. Uh, it ends up hurting everyone in the long run to use bots in that specific way. Butts in general, I'm not against.

I do think there are times when they are incredibly useful, but to use them specifically to fool people into think ging their actual human beings in order to achieve an ulterior motive that sets me on edge that I can't

really see an upside to that. I can definitely see it from the side of customer service or answering general questions, maybe even just trying to funnel out people who have a very simple issue to resolve versus those who need more attention, whereas you know those people would get directed towards a pathway that would lead to speaking to an actual human being. I get it from that perspective. Now, in a recent episode, I explained in brief what the Turing test was, or at least how we interpret it.

The Turing test is sort of the inspiration for the Void comp test and Blade Runner. Alan Turing, one of the fathers of computer science, proposed the test back in nineteen fifty and in the actual thought experiment that he was proposing, it was a variation on a parlor game called the imitation game. Now, the imitation game is one where you have an interrogator that's player, and the player is presented with two subjects, neither of whom the interrogator

can see or talk to directly. One of the two subjects is a woman, the other is a man. Both of the subjects can communicate with the interrogator in a way that does not require face to face contact or voice or anything like that. Typically it would be through something like typewritten letters, because that would help disguise handwriting as well. And the two subjects have the same task. They have to convince the interrogator that they are female.

So the woman will be telling the truth, the man will be lying, and it's the interrogator's job to figure out who is imitating a woman and who actually is a woman. Touring then suggested taking this game a step further by replacing the male subject in this thought experiment with a computer. The computer would also attempt to convince the interrogator that the computer was in fact a woman. Now would the interrogator be able to detect the computer's

ruse if not? Touring suggested that this would indicate some form of intelligence, though not necessarily human intelligence. But you could say the machine is capable of fooling a human being, of of practicing deception, which I think most of us would argue. The ability to practice deception does indicate at

least some form of intelligence. Maybe not the type of intelligence that's gonna go out and teach a class on quantum mechanics, but the type of intelligence that does understand the concept of manipulation at least or at least is able to employ the concept of manipulation, if not understand it from a truly cognitive point of view. Now, the other variations and refinements to the Touring test followed after Touring's death in nineteen fifty four, and Touring's life was

very tragic. We've done an episode on Alan Touring, so if you want to go back and find that in our archives, you can learn all about his his death and why some people rule it a suicide. I think most people do, and some people say it was accidental. But it is an interesting and tragic tale. Today, the general interpretation of the Turing test is that if a certain threshold is met, such as a greater than thirty percent success rate of a computer convincing interrogators that's actually

a human, it has passed the Turing test. So, in other words, if you're an interrogator and you've got a computer terminal in front of you, and you're typing messages and the the response is coming back to you. And if more than thirty percent of the time you cannot tell if that actually is a computer or a person, maybe you misidentified as a person more than thirty percent

of the time, and it's actually the computer. That computer is said to pass the Turing test, and that it is capable of fooling you into thinking it's an actual human being. Now, there was a case in two thousand and four in which a chat bought called Eugene seemed to accomplish this. Eugene's persona was that of a thirteen

year old Ukrainian boy. Critics pointed out that Eugene's limitations as a non native English speaker with a limited knowledge of the world due to his age and the fact that he was from the Ukraine, meant that people were lowering their expectations on his performance when they were chatting with him over a computer. In other words, critics were saying that Eugene was gaming the system by making people think, oh, well, non native English speaker, so if the responses come back

a little weird, that explains that. And being young means that they this kid doesn't have that much knowledge about a lot of things in the world, pop culture, politics, lots of stuff, so your expectations are set low, and then you just think, all right, well, are the messages I'm getting Are those in line with what I would expect a thirteen year old non native English speaker to say to me, or do they stand out as being artificial?

And a lot of this ends up being deflection as well, where if you ask somebody a question and the computer program doesn't have a way of responding, it will try to deflect the question so that it doesn't indicate that

and in fact is a computer program. Well, Eugene managed to to fool a lot of people, But again the critics were saying, well, Eugene was kind of an outlier in the sense that you didn't really think of Eugene as being a native speaker with a lifetime of experience where you could really quiz the the entity and find out, Okay, is this actually a person or is it a computer program. It's sort of beside the point. I'm not here to argue about whether or not machines possess intelligence if they

passed the Turing test, because I did that recently already. Instead, let's focus on the flip side of the scenario, we're human at least I'm assuming you're a human. You might be a bot who subscribed to text stuff. Apparently of you out there are in that case. Thanks. I hope you like the show. But this is all for the humans here, this bit out here, so you bots out there can take a break. How can we humans tell if we're dealing with an actual person or if it

is a bot. Well, one of the ways that we have created a means of separating bots from humans is capture. Capture is an acronym that stands for a completely automated public touring test to tell computers and humans apart. That pretty much sums it up when you break it down. It's completely automated, meaning there's no human oversight necessary for any given implementation of the technology. It's public. It's pretty self explanatory. It's a test that's out there in the public.

I guess I explained it even though it wasn't necessary. That's my bad, y'all. Now, it's said to be a turing test because it's meant to detect human versus automated agents operating on a given web page. We talked about the touring test just now. But h so we're not gonna go over that again. But you know, again, it's just just this indicator. Is there something there that implicates this as being a computer agent not a human being?

And if it is in fact a computer agent, then you have a gate up saying all right, you don't get to participate in this because it's not meant for you. When you have of your web traffic out there generated by bots and you're trying to collect meaningful data about real human being users, you have to have a way to separate the two. Right. So, if I'm a web administrator and let's say that I've got let's just say that I'm running a sweepstakes, have created an online entry form.

I don't want someone flooding my sweepstakes with bots in an effort to try and game the system and win by submitting more entries than anybody else. I want to be able to control that. So I want to have some sort of element on there that can weed out the automated agents out there versus the actual human beings. Now, that last bit and capture to tell computers and humans

apart is the key to all of this. Capture is a Guardian right, Like I was just saying, it's meant to keep people from just writing a script to fill out a form or make a comment on forums, really complete any interaction on the web in an automated way. As someone who creates content online and I get lots of comments on various platforms, I don't want a whole bunch of automated gobbledygook showing up under my various podcasts and videos because then I can't tell where the actual

signal is. All I'm seeing is noise. So you want to have some way of controlling that, and you might use it to limit spam in the message board, or to stop people from abusing the format of an online mistakes, or or again to stop people from harassing others on social platforms. Now, the necessity for cap chub is due to a fundamental flaw of the Internet, and that flaw is this, it doesn't take very many people to make

using the Internet a total drag. You don't want some jerk face to use a script to create thousands of email addresses from a web based email provider and then use those email addresses for spam purposes or for someone to gain the system. In other ways a single person has the potential to impact lots of other people. So everything's out of balance, and the force demands a Jedi

to right the wrongs or something. Now, the ideal application of capture is some sort of test that is very easy for humans to complete, but very difficult for computers to complete. And that requires some creative thinking. So what are some things that people are really good at but computers are aren't so great at. Over time this changes. Computer programmers get better at designing software that allows computers

to simulate more of what humans can do. And that's not a bad thing necessarily because it pushes our development of artificial intelligence. But for the purposes of gate keeping, it does make it more tricky. You've gotta figure out a new way to be able to prevent people from abusing the system. Now, the idea for capture came from

a couple of different teams. One team was at Alta Vista, which started to work on ways to cut down on online abuse way back in the Ulta Vista team was trying to find a way to prevent bots or scripts from adding active u r l s to the search

engine platform. Meanwhile, the other team was at Carnegie Mellon University, And actually this happened a couple of years after Alta Vista's work, and they included some researchers who were really eager to try and find a solution to this problem, and they included Louis on On, Manuel Bloom, Nicholas Hopper, and John Langford. It was the Carnegie Melon team that coined the term capture back in two thousand three, and

it worked pretty well. Humans could get a capture right more often than not, and computers weren't nearly as good at it, at least not at first. Now we'll talk a lot about captures in just a minute and get into some more elements about telling the difference between butts and humans, but right now let's take a quick break to thank our sponsor. So with early capture implementations, things were pretty simple. The capture would take on a pretty

universal form. You'd have a little box and inside that box you would see a couple of different words or collections of letters or other characters, often distorted in some way, and a little field beneath it telling you, hey, tie been what you see here? And it was your job, as a human being type person to type in the correct characters, and that would allow you to gain access to whatever it was that the capture was guarding. And the thought was that computers just weren't as good at

recognizing those characters as humans are. That if you distort them, then the character recognition software couldn't put piece that altogether. The weird shapes would be too far outside the norm for the computer model. So if you had a one, but that number one, the numeral one, it was all wavy and staticky or something like you were, uh, you were breaking up the shape a bit by changing it.

Computers can't really see that and conceptualize that's a one, or at least not in the early days, so it would just look like a weird squiggle to them, and they wouldn't be able to complete the capture. Whereas we human being type people, we'd look and think, that's the worst number one I've ever seen. Some kid must have drawn that, but we understand it is, we recognize it, so we would type that in. That was the basis for capture. Create a test that's relatively easy for humans,

very difficult for computers. Now, not everyone was capable of seeing these captures. Clearly, some people have visual impair impairment, and so they need to have some other element to captures in order to be able to access that same content. So there are also audible captures, which is pretty important option to get around those visual impairments that some people have. And uh, you might get a distorted voice being repeating out the same sort of letters and numbers that you

would encounter with a capture. There might also be some background noise that would include some other elements that would make it hard for a computer program to analyze the audio and figure out what was being said, but hopefully humans would be able to make it out. So again, it was all about making it more challenging for a computer while not making it too challenging for human beings. And sometimes that works great, and sometimes that doesn't work

so great. There are plenty of examples of human beings who could not get through a capture because the distortion was so great that it made the made it almost impossible to recognize what the actual capture was supposed to be. But the first counter to capture wasn't an advance in

computational analysis of visual or audible data. You know, there are a lot of tricks that people figured out later down the line, to make these visual captures easier to analyze, things like switching all the images gray scales so that you take out the different color gradations that could fool a computer, and other elements along those lines. But at first those weren't even really necessary because the people who really wanted to get access to those systems didn't bother

programming better AI. They just went and started paying people to fill out capture forms. Those who wanted to continue the game giving the systems, they created a new industry. They'd pay the people to fill out all these capture fields. There was no need to develop any sort of AI. People were doing what people were supposed to be doing easily. They were solving captures. Now, the pay was super low and the output was super high, and it posed a

threat to the capture system. Now, as an analogy, amount imagine that you build a big fence strong enough to keep bears out. No bears will get in this fence, you say, and you go on your married a little way.

What you didn't notice is that there were gaps in the fence that while the bears are far too big to fit through the gaps, the gaps are big enough to let rabid I don't know possums through, And so the bears who go to employ rabid possums, paying them handsomely, are able to access the stuff behind your fence anyway, because the rapid possums pass right through the security. They weren't intended to be kept out. Of course, in the case the captures, we are talking about people accessing the system.

They were just doing so in massive numbers and for less than ethical reasons. So the Carnegie Melon team began to consider a new approach. That's when they developed recapture. This tech used images of real words and numbers taken from old documents. The original run was of New York Times archival texts, but eventually the teams sold this technology to Google, which began to use it on lots and lots of scanned books. They were trying to transcribe those

old books. The company used recapture to display scanned words or numbers from the texts, and as more people filled out the recaptures, Google began to use that data to transcribe these old works, which meant that they had a digital copy of these books that they had come into possession of, which means anyone filling those fields out was actually technically doing real work for Google, including all those

folks who were being employed to write out captures. Meanwhile, bot developers were making better bots, and character recognition and analysis software was getting better at increasing success rates with visual captures. Now that would prompt capture designers to make more challenging captures, and soon we reach a real problem. The whole point of capture was that it was supposed to be easy for a human to complete, but difficult for a bot to complete. If it becomes tricky for humans,

you've defeated its original purpose. Now Google updated capture to the familiar I'm not a robot check box that you can still find on some online forms. They call it the no capture recapture catchy. It wasn't just a check box that needed checking. Behind the scenes, back if you were able to stare at the back side of the website that you're on, software was analyzing your clicking style all so it would look for stuff like was the box clicked right away, perhaps before or at the same

time as fields were being filled in. If so, that indicates a bot rather than a human being. But this approach also doesn't get around the fact that you could employ real human beings to do this same work. So well, it's an effective way to tell the difference between a bot and a person. It's not necessarily effective in keeping spam traffic away from a site if people are willing to employ actual human beings to do it. In sen Google killed off this version of capture on its own services.

You can still find it everywhere else, but these days Google uses invisible recaptures. Now, this version analyzes your browsing behavior, and there aren't a lot of details released about it yet, but presumably Google is looking at how any given agent on a website uses a web page to determine if, in fact, that is an honest to goodness human being, or if the terminator has the said to pop over

to Zeppos to look for some new kicks. So this is still in a way of being able to differentiate humans from machines based solely upon behavior, just analyzing the behavior and thinking all right, well, this indicates a human being. This is this person, This entity is navigating a web page the way a human would versus this is really efficient and formulaic and repetitive, and that tells me that's possibly a machine. So let's switch over to Twitter. Twitter

has got a lot of bots on it. Twitter and the follower numbers are kind of a type of status online. If you have more followers than the general implication is that you must be more important than someone who has fewer followers, and so there's a healthy market for purchased followers. On Twitter. You can go to several different companies and stores and buy followers by the hundreds or thousands. So if you're desperate to boost that number, you can pay

a service that will link accounts to your account. Now, most of those probably do not have real, live human beings behind those accounts, and so a visit to any of those accounts will show you that they never seem to say anything themselves. They'll retweet what lots of other people are saying, but they don't actually, you know, tweet anything of their own, or if they do, it makes

little to no sense. It might just be kind of like garbled general you know, new a g kind of stuff, the things that that sound like they might have some sort of deep meaning, but if you think about it, you realize, no, that really doesn't mean anything at all. Now, on a one on one basis, Twitter bots are pretty

easy to spot. So let's say your you tweet about something important going on, such as you know something's going on in politics or whether John Snow is going to win the Game of Thrones, and almost immediately after you tweet, you notice a new followed notification and if it popped up sue pretty quickly, like instantly after you posted a tweet.

That might very well be a bot running on a script that is searching for instances of specific keywords, and when it finds those keywords, it then prompts the bot account to follow the account that generated the keywords, assuming that hasn't already followed that account. And some butts do this in order to convince people to follow them back because lots of folks on Twitter have a follow back policy which helps them boost up their own follower numbers.

You know, it's the whole hey, if you follow me, I'll follow you quit bro quo kind of approach. But in this case, one of the two parties is a bot, at least one of them. Anyway, maybe they both are, which is kind of funny and pointless. Now, once you follow the bot, you may start seeing spam messages from that bot pop up in your feed. Whenever it occasionally posts two followers, it's likely trying to get you to

engage in a particular behavior. Now that behavior might be more or less benign, such as convincing you to shop a certain brand which is obnoxious but not you know, malicious, Or it might be more sinisters, such as trying to get you to do something foolish that will compromise your computer and allow it to join like a hacker's bot net army or something. And there's a lot of reasons, most of them annoying, that a bot programmer would want

you to follow their butt. According to a study conducted by researchers at Indiana University and the University of Southern California, somewhere between nine and fifteen percent of all active Twitter accounts are actually bots. It usually doesn't require a lot of work to determine if a single account is the work of an actual human being, but if you have a lot of them, that can be a challenge. I mean, if you've got thousands of followers, sorting through all of

those would take a real long time. So that's what prompted developers to create apps like butt or Not, which scour Twitter followers and look for signs of butts, returning a report to the user to let him or her know how many legitimate followers they have. Those apps, which you can argue are are kind of bots themselves, look for indicators such as each followers Twitter description uh the u r L field, the number of tweets the account has generated of its own, the number of followers the

account has, and so on. So if you come across an account that follows thousands of other accounts but only as a few followers of its own, that's a red flag. That's saying, well, this account is following lots of people, not a lot of people follow it. That tells me

something hinky might be going on. If the description or you are l are empty, that's another indicator because it shows maybe someone didn't want to take the time to try and fool people by creating a bogus description and a bogus u r L. There are several other criteria that the apps look for, and depending upon how many red flag boxes get checked, the app determines if the account is the work of a script or if it's

an actual person. Now, on the one hand, we can look at all these stories about bots, and think of how irritating they are because they generate spam content, they clog up actual communication. They create deception, whether it's an attempt to trick you into following a malicious link or to think someone is particularly notable due to the enormous number of Twitter followers they have. But on the other hand, we can think about how these examples show how we're

getting better at creating more human like agents. Now that's not to say these agents possess intelligence, only that they can imitate human interactions enough to raise the question could this be a bot I'm talking to? If you have to ask that question, then that indicates programmers are getting better at designing bots, or that you're getting pretty bad at recognizing humans. Some days, I certainly have that problem.

We'll talk a little bit more about machine intelligence and communication in just a minute and kind of layout why it's so difficult to really create a truly compelling butt that can fool people into thinking it's a human. But first let's take another quick break to think our sponsor. Now, they're just elements to human communication that bots are not great at handling, or they need a huge amount of help in order to pull it off. So let's take

IBM S Watson for example. Now, Watson is the interface that made the news when it competed against two former Jeopardy champions on a special edition of Jeopardy. Watson beat the opponents, which is pretty impressive when you consider that Jeopardys format includes elements of wordplay includes and machines are typically not very good at interpreting word play and subtext and that sort of thing and getting at what the

actual meaning to a sentence is. Watson even attempted a couple of jokes throughout the course of the game, but they weren't really spontaneous bumb malls designed to get a chuckle of, you know, Alex Trebek. Humor is just one of those aspects of human communication that is difficult to quantify and implement with machines. Typically, it requires programmers to think ahead and imagine specific scenarios and queries to build

out appropriate or, depending upon the context, inappropriate responses. So, for example, when Apple's personal assistant Sirie debuted, people immediately began to test Sirie. They began to ask the digital personal assistant all sorts of odd things and sharing the results. If you create any sort of system. One of the first things you're going to find when you allow people to access that system is they're going to try and break it, or they're at least going to try and

explore what the limitations are within that system. And they're not necessarily doing this malicious with malicious intent, but rather that you know, we're humans, were curious. We want to know how how far do things go? Are they really limitless or are you going to run up against an invisible all if you keep going in one direction long enough.

The same thing is true about personal digital assistance. So in some cases where people were asking weird things of SIRIE, serious responses were particularly hilarious, indicating that someone over at Apple had anticipated some of those shenanigans because SIRIE wasn't coming up with these wacky responses on its own account. It was referring to a database of responses that people

had been compiling ever since they started working on the project. So, if you are working on a personal Digital assistant project and you think, oh, someone's gonna say I love you eventually to this, I want to have a response to come back that isn't just I'm sorry. I don't understand that every time the digital assistant says, I'm sorry, I don't understand that is an overall, like outright admission of limitations. So you will try avoid that as much as you can.

Make it kind of a joke instead. But it means thinking ahead, and it means the humans are thinking ahead. It's not a machine. Uh So, here's an example. One of the early queries that got widespread traction was I need to hide a body, and Siri would respond originally with various sites where you could, you know, possibly dump a body and get away with it, like nearby reservoirs or quarries. It's pretty grim, but darkly humorous, and it showed that someone had been thinking those things through by

the way, that joke became very serious. In two thousand twelve, a Florida man stood accused of murdering a friend of his, a roommate, and on his phone the suspects phone was a screenshot of a query to Sirie, the one about where to hide his roommate, and prosecutors used it as

evidence in the trial. But it turned out the screenshot that used wasn't really a query that the man had made himself, because his iPhone was as an iPhone that was running on Verizons service, and the screenshot was from an iPhone that was running on a T and T S service. Uh. Also, it turned out that the phone he was using, the suspect was an older model of iPhone that didn't even support SIRIE. However, he was later found guilty of his crime, though the Serie connection was

again dismissed for those multiple reasons. Later on, Apple would replace that joking response with a referential but less morbid joke, which was quote I used to know the answer to this question end quote, So, in other words, acknowledging that, in fact, there used to be another response without actually giving it because of you know, these very grim, macabre reasons in real life. But let's say you wanted to create an artificial entity that could respond with humor dynamically.

It wouldn't require you to pre program in responses to different questions you'd have to anticipate. This would let you have a bot that could convincingly stand in as a human without the danger of the bot encountering something you didn't expect and having no response to it, or to misinterpreting the interaction with an actual human being, or if it did misinterpret it that it could follow up in

a very human way. So if I make a joke to my coworkers and I do it well, my coworkers understand what the meaning of the joke was, what the intended meaning of the joke was, and there you get that response. If the joke doesn't go well, I can follow it up by explaining the joke or explaining what I had tried to do with the joke, which doesn't make the joke funny, but at least informs the audience as to what it was I was thinking. Machines would have to be able to do that too, and this

is hard to do. Machines would need to be able to interpret not only the literal meaning of any statement, but the potential intended meanings as well. So I would have to incorporate the concept of novelties, introducing something new and unexpected into the interaction. It's a subversion of our

expectations that tends to lead to humor. So, for example, Douglas Adams, who is one of my favorite authors, once wrote a sentence describing a fleet of spaceships, and the sentence goes like this, The ships hung in the sky in much the same way that bricks didn't. Now that's a great sentence. It gets across the humor and intent to the reader. You know that if you were to see these spaceships in the sky, they would look completely out of place. They might even be remotely brick shaped.

But mostly it's the idea that if bricks could hang in the air, those ships would look like that, except obviously bricks can't hang in the air, And in one sentence, Adams is able to convey with humor the mind bendingly weird experience of seeing these spaceships in the Earth's sky. Computers would have a real hard time replicating that, at

least on purpose. A computer program that put rough sentences together using a basic syntax and vocabulary could potentially make weird and funny sentences, but these would be mostly random and frequently meaningless, and you wouldn't be able to hold a context from sentence to sentence. To make something that

has meaning requires aspects of intelligence that computers don't yet possess. Watson, with its jokes, was running on a massively powerful computer system with two thousand, eight hundred eighty processing cores, and that doesn't even approach the power necessary to create real humor spontaneously. To detect and generate sarcasm, and entity must understand context and other cues and machines aren't very good at this, though we've seen some advances in contextual tracking.

For example, Google's Personal Assistant can follow a line of questions about the same subject without you having to restate the subject with each question. If I asked my Google Home what when the next Braves game is, it would give me an answer. Let's say it's day after tomorrow. Well, I could follow that up with what will the weather be like then? And the system would understand that by then I mean the day of the game, So the

day after tomorrow. I might also ask what's the fastest way there, and it will know that by there I mean the stadium, and that I am probably am asking how to get from my current location to that stadium and the most efficient way possible. The subject is stored in temporary memory, I don't have to keep asking specific questions about the game or the stadium. But that's still a long way off from actually understanding context. So one test for bots might be for us to have it

tell us a joke. If it's clear that the bot can create a brand new joke, one that has not been pre programmed, one that is spontaneous and novel and created by the bot itself, and it makes sense and it is funny. We've reached a point where telling bots and humans apart is going to be tremendously complicated, but

right now we're nowhere near that. The jokes that we hear bots tell, for the most part, are ones that have been created by human beings and just stored in a database, and the body just pulls them out and then recites them. It's not creating them. It's just pulling a massive data from a cell in a giant spreadsheet and saying, all right, this is the joke I'm gonna tell us. The joke that's in sell see four and seventeen. That's the joke for today. That's not creating a joke,

it's just reciting one. If we can get to a point where they can create jokes, that's a big jump in computer intelligence and maybe a brand new audience from my type of humor I'm always looking. Well, that pretty much wraps up this episode. Really, the key to determining whether or not it's a bot or a hu Man is testing whether or not it's capable of handling novelty.

Most bots are fairly limited in the scope of things they can handle, and if you step outside of that, you see those limitations pretty quickly, and that then it becomes apparent. But every year we're getting a little bit better at handling wider spectrums of experiences with bots, so that it becomes more and more complicated to tell them apart from human beings. Uh. In most cases it's probably

a moot point. It's not really necessary depending upon what it is you're trying to do, But in some cases you really do want to know whether or not that's a human being or a machine. On the other end, if you guys have any stories about funny times where you were chatting with something that you thought was a human and turned out to be a bot, Like I've got friends who have received robo calls and didn't know until about half a minute in or maybe a minute in,

that it was a robot. Those are great stories. I have specifically love the ones where if you ask the entity are you a robot? It tries to deflect but does not actually answer the question. Those are the best. But you can get in touch with me, let me know your experiences. The email address is tech stuff at how stuff works dot com, or you can drop me

a line on Twitter or Facebook. The handle for the show at both of those is tech stuff hs W. Remember, normally you can watch me record shows live at twitch dot tv slash tech stuff. I record on Wednesdays and Friday's. Today's episode is a little bit outside the norm. There is no one currently watching me live, so when I'm doing my dance like I am right now, no one can see. But most days you can see, and I do the dance then too. So join me at twitch dot tv slash tech stuff to watch the show live.

You get to see all the elements of the show come together, and I will talk to you again. Really for more on this and thousands of other topics, is that how stuff works dot com.