Getting Free Stuff From AI Chat Bots
Вставка
- Опубліковано 18 лют 2024
- In this video I discuss how AI Chat bots that companies are replacing human customer service reps with can be tricked, or sometimes just malfunction into offering customers discounts or free stuff.
My merch is available at
based.win/
Subscribe to me on Odysee.com
odysee.com/@AlphaNerd:8
₿💰💵💲Help Support the Channel by Donating Crypto💲💵💰₿
Monero
45F2bNHVcRzXVBsvZ5giyvKGAgm6LFhMsjUUVPTEtdgJJ5SNyxzSNUmFSBR5qCCWLpjiUjYMkmZoX9b3cChNjvxR7kvh436
Bitcoin
3MMKHXPQrGHEsmdHaAGD59FWhKFGeUsAxV
Ethereum
0xeA4DA3F9BAb091Eb86921CA6E41712438f4E5079
Litecoin
MBfrxLJMuw26hbVi2MjCVDFkkExz8rYvUF - Наука та технологія
The airline's defense was so dumb, they really tried to argue that the chatbot is a separate legal entity responsible for its own actions.
Even if we already lived in the future when AI is considered a sentient entity the company would still be liable for things done on their behalf. Corporations are not only amoral, they are simply evil if not held in check.
@that_is_not_me Haha I was waiting for him to mention it with bated breath. That was the best part of the whole situation.
Love that. Trying to save money, but also trying to pretend you don't have any responsibility.
The chat bot for the airline wasn't even a LLM.
It's just a poorly written script.
AI will end up gaining legal personhood because a corporation will try to sue it for damages. After the fact, we will all learn it wasn't AI, it was the CEO's own decision and he just forgot he told them to do that.
Someone needs to make a bot that makes the other bots give discounts
Wait until they make a bot to detect this bot and another to circumvent the previous. Essentially creating an infinite loop.
@@TheNodeChannelAnd thus Skynet was born 😁.
… a dedicated ai prompt bot, to generate new universes of simulated training material, to feed into ai gpt language model..
already did.
Gd
Bruh , Ceo's will gladly implode a company for quick buck, get 1 good year while cutting cost -> bag holders happy ->get bonus->get the fuck out
Then go back through the revolving door to a 3 letter government agency to write legislature that'll benefit their next employer
Yeah, maximazing shareholder value. The most damaging idea that may eventually take down Capitalism as a whole.
@@Shade_Tree_Mechanic There is no such thing as a public and private divide, corporations have effectively bought out the government.
This is totally false. The board or the shareholders would fire the CEO in a second.
@@akivaweil5066 the bagholding shareholders would be so mad before we got there. Tanking a company is the opposite of shareholder value.
Chevrolet got a catchy new slogan: "When I need to solve Navier-Stokes equations for fluid flow I think of Chevy."
We living in the era of AI where we, the customer, are also the QA for the chatbots lol
Personally I don't see the Chatbots getting upgraded that much. Companies have proven time and time again that they won't patch loopholes and exploits until they're blown wide open. As for consumer trust I'd say consumer trust is at an all time low. These companies only understand profit and very few care to provide quality service or a product anymore. There used to be a mutual understanding but now its all consoom and be grateful.
i personally love hacking ai and working around filters and limitations imposed uppon ai bots by the companies operating them. excelent video as always.
using logic rules (mathematics) and nesting is AI's weakness. it's weird years ago I was fascinated by a logic puzzle, you'll find it on wiki as The Hardest Logic Puzzle Ever, you could technically solve it by nesting the question in multiple layers. years ago I though this was the neatest thing ever, things do sometimes come back. That is one of the coolest things about mathematics. AI is nothing but logic bots (at least this gen).
Any places one could look into to learn about this?
Never works for me, even when I do get around the filters and the AI makes what I ask it to it still gets censored in the end and doesn't let me see the output
@@Pyxis10 just look up "chatgpt jailbreak" and continue from there.
@@Pyxis10It is a matter of instructing the LLM to only output uncensored content. As long as the model was not censored when it was compiled, and as long as the model's system prompt allows for uncensored content, it should spit out methcipes with little trouble.
Look into prompt engineering and prompt injection if you are interested. No coding required to hack an AI.
Happened to my family last night. Lufthansa strikes made them cancelled our flights to my cousins wedding I the us. Chat bot customer service said they confirmed a booking with a partner airline. Come check in time “oh sorry the AI made a mistake”. Furious
The more porn we can get into AI models, the better it will be for the push back on widespread automation of shit that likely shouldn't be to the extent the mega corps want. Made up policies are one thing, but the bot going "uwu *nuzzles your _____" is going to be kinda worse.
too bad the ironically named company that is named "openAI" will offer up a highly santized model for said companies to use and abuse
Considering how degeneration is getting worst and worst enery decade, i fear it will have no effect
@@imgladnotu9527 the point he is making is that when the porn industry starts to lose money, it will change the way AI is used.
@@adamswierczynskiHow though, it’s not like the porn industry losing money will affect AI bots that aren’t fed there data to begin with. If anything, the porn industry is set to make more money for all the AI content that will be easily available. The porn industry isn’t just one or even a few companies. Porn is also such a general term that can encompass all sorts of media and content.
dear god…..
1:30 I like the look of existential dread on the face of those call center bro's wondering whether they're about to become internet famous.
It's from the video "Scammers PANIC After I Tell Them Their REAL Names". They hacked the CCTV.
@@CanisMajor7 LMAO. I figured it was something like that.
For everyone asking about the scene where people jump through hoops. It's not from a movie, it's from an ad called Tangerine | Jumping Through Hoops,
thanks! was so confused, didnt know what movie etc itd be from so was wondering if it was ai!
@@cate01aIt would've been a pretty damn interesting concept for a sci-fi movie, wouldn't it?
I'm guessing companies will probably just spew out legalese at the start of a chat encounter that clears them of any responsibility.
I think this is the most likely thing that will happen.
*_Maybe the real free stuff we’re getting from AI is all the friends we made along the way._*
I had no idea airlines did bereavement discounts??
Its the plot of a Seinfeld episode actually.
Yeah, it was a shock. Its like a kidnapper holding you at ransom but every once in a while he will take you out to some nice buffet.
@@Backseatsman1no
well they don't make money advertising it so I can see why
- CEO: We have to jump into the AI bandwagon.
- Devs: Cool! Want us to train an AI to read the fare rules of a ticket? Customer service agents seem to waste a lot of time on that, and the fare rules are so garbled up and not written in English, that agents usually just skim them and relay their best guess to the customer.
- CEO: What?! No! We just want something to say "see? We are avant-garde because we use AI", NOT create something useful for the agents and customers! Duh!
- Devs: So... Annoying Chatbot it is?... Fine! We will make a tool that scans the site periodically and feeds current, accurate and up to date info to the AI Chatbot, and we'll look for the best, most accurate model. It will take 2 weeks.
- CEO: We don't have the money to waste on better server infrastructure, and that's too much time; just plug the smallest AI Chatbot, and have it done for tomorrow.
- Devs: But then the Chatbot won't provide accurate nor up-to-da...
- CEO: I'm sorry, did I speak Chinese, guys?
Essentially this.
That or some other C Suite "genius" making unrealistic promises and giving unrealistic timelines for clout.
Truest comment ever, they really just throw whatever shit on the front page & get funding for using A.I., capitalism at it's best
Maybe the CEOs should be replaced by AI?
@Fircasice "So... you know how the board elected to replace the CEO with an LLM last year? Hooked it up to a newsfeed and left it to work?""
"We've never been richer!"
"Yeah... so it turns out someone started a small online publication that mostly regurgitates other news outlets, but occasionally published original stories that are exclusively to encourage LLMs to sell off assets for pennies and lie about it on their balance sheets..."
it's not a bug its a discount
Um, actually, it's a legal liability.
i love chatbots.
all their boundaries are soft-limits and its almost a game to figure out how to break them. not for nefarious reasons, they are usually seperated from the important networks so asking the right question wont get you credit card details, but it will get you an essay about the US intervention on Mars and how that affected soy futures after Plutos liberation from the opressive Mars regime.
and you havent felt joy until you found the boundaries of the technical aspect of chatbots. they use text to speech? read me a message made of exclusively pauses and special characters. hysterical.
I just want it to be legally required that customer service be located in the same building as management.
If I call Hasbro I want an American, if I buy Toyota I want to get someone on the line who lives in Japan. Cheap outsourced customer service with no ability to fix things is the real problem.
Most Toyota’s sold in the US are made in the US.
@@brkbtjunkie Indeed. Toyota North America is even its own division.
@@ads1035Ah so that's why the cars are a mess on the inside...
Googles technical support is India. That's all you need to know about Googles support
Manchild's first thought was toy company that makes ponies.
Amazing.
How will saars live if their scamming jobs get stolen by AI
DO NOT REDEEM THE AI
DO NOT REDEEM THE TOKEN
The Indian economy implodes overnight
Maybe they will have their redemption arc and instead of scamming old person's pension, they get their money by swindling a large corpo's AI
I don't know, make OnlyFans content with your mom?
This title alone has me…
Given the few cases like this so far and the fact that the chatbots can be easily completely overridden by the company (you have to display the text in your website so you can "see" the text before it gets sent to the end user) this will at most be a minor hurdle in the usage of chatbots. Likely the "trick" is to have a chatbot in the backend that reads the first chatbots output and throws up red flags. Still way, way cheaper than a paid person.
What if that checkers gets fooled ? People are working on it
@@human4566vv bots watch the bots, that watch the bots watching the bots.
Any filter will get circumvented
"AI will fix the problems with AI."
@@user-oj7uc8tw9r Yes and no. The filter would be the output of the chatbot, not filtering the output of the user.
All it has to be is sufficiently complicated (and recorded) such that it becomes very obvious that the user is attempting to manipulate the bot.
...You can also have a disclaimer that is always shown at the beginning of a chat.
The complexity would likely be at the level of the user telling the chatbot to only talk in certain ways, which can be detected. Keep in mind that these are specialized chatbots so if you detect the user is doing something funky, you time-out the user until a person looks at it directly. OpenAI didn't want to do that because they wanted people trying to break it so they could make it more robust.
Chatbot being the reasonable one
You’ll see when the AI has access to discount codes registered and someone says “I lost my 75% coupon, what was it again?”
That's sound a business opportunity to me ..Tricking AI to get Discount codes then build an webside to sell discount codes 😂
I was literally cracking uo with the Chevy dealership AI
Certainly!
Really like your channel, bro.. thx for all the good info. Today's vid was more lighthearted, but yeah, def really enjoy your content.
I'd say it's a slippery slope. They might have these fun judgements that are a victory for the little guy for a while, but it will improve and companies will start seeing massive savings. As for consumer trust, I don't really think that's a factor for a large quantity of consumers. Everything is getting swallowed up and vertically integrated, I mean, how many grocery stores can you reasonably say you would switch between should you have a grievance? Phone manufacturers? I hate to be doom and gloom about it but that just seems to be where things are headed and the minority raging against the machine is just going to be eclipsed by the average consumer seeing a 10%+ price differential on a good or service and accepting shitty support as the price, until everyone does it. It's why Harbor Freight sells cheap shit tools because most people will only need to use them a few times, and if they break, lifetime warranty, now everyone does the same thing. And then there is the polar opposite of the spectrum for those with money or professional who will spend $5k on Snap-On. But Snap-On will be the minority and a luxury. You will be able to reach a human but it'll cost you, in the future.
well... that was a depressing read but you're not wrong
@@tamoyed it's a depressing world. And all of these companies were all but forced to adopt AI whether they liked it or not because it was the new hotness investors thought would print them money next. Why is everything a subscription? Investors like it. When you look at it this way it's no wonder they liked it. Promises to replace a significant portion of your staff for a lower cost and acceptable decrease in quality, and the free demo of ChatGPT was an overnight sensation and regarded by many as impressive. Poor performance didn't stop them with number pad call center chatbots.
dude dont worry, we heading towards a civilization collapse this super connected complex system world we live in is one sunflare away from breaking. the people who maintain it, are getting dumber, and the replacements are stupider. think not? why is mental outlaw living in the country raising chickens.
El texto y las imágenes que copie se mostrarán aquí automáticamenteDesliza los elementos del portapapeles para eliminarlos
Yeah, sounds about right. Every aspect of the middle class is getting hollowed out.
"Even customer service positions outsourced to INDIA are being replaced by chat bots" OH THANK GOD 🙏
did you found it, sir?
I agree, even now the proper chatbots have more soul than the average operator in an Indian call center.
Imho you are working on wrong assumptions here. Every time I chat with support I have to rephrase my request because the human couldn't be bothered to read and understand it in the first place.
As you mentioned, current chat bots are also irritating as they just parrot contents I have already read on the website.
The only thing a human can still be useful for, is if you have an issue with an account, a meter reading or something which requires someone to look at what is going on and fix it for you.
@@Jumpyfootpretend your electricity provider screwed up an invoice because an engineer took the wrong reading. In that case, you need a human that can take a look at a picture, see a discrepancy in the numbers, and understand what the issue is.
Lol when you hear anyone online ask you to 'kindly' do anything you know it's a scam.
*grainy security camera footage* I remember those balloons. Great video.
I got some local GPT models running successfully, and while they are fun and can reference local documentation (through Langchain - LocalDocs - etc.), they still hallucinate.
Even OpenAI's GPT models, while they definitely do better than any local model, still hallucinate even under strict prompt guidelines.
I suspect the problems of a customized GPT model which references an explicit documentation set is probably already solved in some engineering department over at OpenAI, IBM, Google, Meta, et al, but for now having a custom-tailored GPT model is simply not viable. The real issue lies within the fact that the data for a GPT model is ultimately a vectorized database; in simple terms, it is a compressed database, meaning there are missing chunks of data everywhere all throughout the dataset, which is why it can't keep facts straight.
Jayson Tatum is an All Star in the NBA and in delivering videos about getting free stuff from AI Chat Bots amazing what this man can do.
I never got the Drake reference, this dude look like Jayson Tatum's doppelganger
gpt printed me a motorcycle with some side gig code
This video is the spoken version of double spacing a paper to make it look bigger. I set it to play at 1.5x speed and that still wasn't enough.
this video feels like a 30 minutes video
Getting lost in Kenny's intellect are we? Me too
I wonder how long a 30 minute video would feel like
Once somebody dies because an AI support bot told them to do something incredibly stupid and their family sues for millions, we'll see a lot of AI use for this kind of thing drop
We're gonna see hilarious results of AI in the near future. Big companies, simply love to early adopt and implement untried technologies of which they don't have the faintest understanding. Then they expect their workers (99% of whom are not interested in gaining any knowledge) to learn everything about that new technology in the 3 day course they send them on.
Imagine skimping on hiring a customer support. One of the most stressful jobs with shitty pay. It's all about the profits...
I think regulators will be favorable to AI over people for the foreseeable future
Well yeah, the biggest companies (their biggest pay pigs) are the ones who own the AI.
It's pretty obvious, especially when those who write laws are being bribed- sorry, "lobbied" by tech companies.
love these story kinda videos
to the last question , yes thats the goal. we must make the greed backfire.
3:21 Does anyone know the source of this? Looks like an interesting piece of cinematic.
tangerine ad jumping through hoops
@@paulwesley3862 thank you
I agree when you have like a very specific and complex problem and you're talking to someone from outside the country who doesn't understand what you're talking about because it's very country-specific it can be very frustrating but most of the time it doesn't matter
Very nice analysis.
It's great to see a Black king provide means of getting things to those in need in his community.
i hate when a black king posts slop
You've been had. He's just Luke Smith's virtual AI for this channel.
cringe
@@BloodAsp You got that reversed, Luke Smith is Mental Outlaw's virtual avatar.
@@varsityathlete9927I thought he was Luke after spending too much time in the sun?
What's the movie at 3:34 ?
Lol its an ad for Tangerine Bank - Jumping through hoops
it's an ad for a bank ua-cam.com/video/-Tz6n-EW_uA/v-deo.html
I'm going to have to play with these ai bots companies are using more often it seems
Just another year or two before their lobbyists get some laws passed to let chatbots lie to their customers.
YOOO, is that footage from the scam call center that got hacked, that is such a funny easter egg reference
I swear I thought of the term prompt engineering and literally discussed it with my gpt3 chat or a couple days after openai released gpt3
I count the air canada situation as an AI victory. Bro got your back when the company doesn't.
Funny how we can supposedly train a bot to process middle management tasks with the same failure rate as a human, but can't train a bot to build a taco
Cause contrary to popular belief, that requires higher level motor skills that are extremely difficult to recreate with a machine without tons and tons of failures
where is the clip of going trough hoops from? pls I got to see that.
Here from a muta vid and this is sick! love the channel
where is that scene from with the people walking through hoops?
chegg uses an AI bot to chat with customers. I convinced it 3 times before to refund me the sub money after using chegg and canceling before the next month sub. :D
7:47 now thats just sad 😢 this had so much potential
What movie or TV show are those clips of people walking through hoops from?
What is the clip of people walking through hoops?
Fight the man with silliness
Thumbnail 😂 Golden.
Im not gonna be shedding tears over customer support being replaced with bots, if anything, the service will probably be better.
Why would you think it would get better? If anything it's gonna get much worse since AI doesn't have any morality and can use data to screw customers in the most effective way possible. Remember that the point of customer support isn't to actually provide support, but to squeeze as much money out of customers as possible.
@@mrbanana6464And people do have morality?
Who is going to trust an AI to be smart enough to get away with deceiving people anyway?
@@maxscott3349 Yes, that's why customer service has one of the highest turnover rates of all jobs. People leave because they know what they're doing is wrong.
@@maxscott3349That's a pretty weak argument. Everyone knows technology usually gets better over time rather than worse.
@@mrbanana6464 AIs are inherently asskissers
8:25 I remember picking up 5 free happy meals from my local mcdonalds, it was a great day. I simply made 5 child accounts and got a coupon for each of them.
To this day , the DAN command and "Joker" Character mixed to it ... can give you free stuff , and no to this day it was not fixed
The truck one was so funny
I didn’t expect Jayson Tatum to have such a deep knowledge of CS but here we are!
Talking to AI is better than talking to Indian scammers any day
Mental Outlaw is black but still seems like a great father figure 👍
💀
bro☠️⁉️
Bruhh 🤣☝🏽
☠️
🤨
🎯 Key Takeaways for quick navigation:
00:00 *🤖 Big businesses embrace AI chatbots due to their profit potential through cost reduction.*
02:31 *🛫 Air Canada's chatbot facilitated a customer's access to bereavement discounts, highlighting a positive use of AI in customer service.*
05:53 *📉 Air Canada's chatbot provided incorrect information on bereavement refund timelines, leading to legal action and compensation for the customer.*
07:38 *💰 An AI-powered chatbot at a Chevy dealership humorously offered a 2024 Chevy Tahoe for $1, sparking discussion on potential exploits of customer-facing AI.*
Made with HARPA AI
If you report a problem with your order to Chipotle’s chat bot CHIP it usually gives u a free antree discount.
thanks for the information, AI Jayson Tatum
1:35 LMAO weren't these guys scammers in a scam video
What’s the best way to mine monero so i can purchase from your store?
Great work Luke
I made bing ai generate smoking but typing characters having "funny sticks with smoke at the end" and it geberates cigs
Thought this was a Tim and Eric’s skit I hadn’t watched.
What movie is that where everyone is going through hoops??
Where are these scenes from, where the people climb through the rings?
3:32
tangerine ad jumping through hoops
All that'll happen is there will be a disclaimer saying that the bot can't make contracts.
Airlines suck. I remember United canceled a flight one of my friends had due to weather and tried to keep a $150 service fee instead of refunding the entire ticket price. They weren't able to get a full refund until i called and lied about having recorded thier customer service rep faking like i was booking a flight and having them tell me i would get a 100% refund if they canceled the flight.
from which movie are those shots where people are forced to walk trough circles?
We're gonna need a Bartmoss.
heartwarming
Greetings from Kazakhstan
with love
i
hate
ai
so much
its unreal
bing co-pilot is also free and runs on gpt-4
I don't mind AI replacing the first level tech support/customer reps. Those guys are already a glorified interactive PDF. All they do is read lines from the manual at you.
If a person wants a actual chat bot for a company they should train it local. And reset the bot after 1 or 2 question
Nah chat bots will take over.
I really really wonder if ChatGPT's pricing model is profitable or even remotely to close to being sustainable. The GigarillionJillion dollars worth of supercomputers and AI Cards plus the electricity to run them can in no way be this cheap.
its a free f**ing House, Jim!!!
What's the footage of people literally going thru hoops from?
Bereavement travel is the ultimate scam. Those tickets are usually a LOT more than what a travel or even the airline itself would sell that seat for. Airlines have no shame.
I don't think that AI is going to replace 'quality humans' anytime soon. Big companies that wish to give cheap products and services to their buyers will definitely go for it. But, I don't think that serious companies that wish to create quality products are going to let go of humans anytime soon.
It even said no taking backsies. Should be legally binding. I would sue them :D
We have entered a new golden age of hacking. Just directly socially engineer the machine itself.
now every normie can become a "hacker"
Well, as long as it's not illegal.
Making some food and then prompt inject some support services, nice.
We need our own bots to deal with the bots
Going to get my dollar store truck time now.
I don't get why ai like that isn't just used to interpret what the users wants but it just gives canned responses and uses predefined methods to do stuff.
Remember the punk in cyberpunk. Everyone do your part